xref: /freebsd/crypto/openssl/CHANGES.md (revision 87b759f0fa1f7554d50ce640c40138512bbded44)
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 3.0.14 and 3.0.15 [3 Sep 2024]
32
33 * Fixed possible denial of service in X.509 name checks.
34
35   Applications performing certificate name checks (e.g., TLS clients checking
36   server certificates) may attempt to read an invalid memory address when
37   comparing the expected name with an `otherName` subject alternative name of
38   an X.509 certificate. This may result in an exception that terminates the
39   application program.
40
41   ([CVE-2024-6119])
42
43   *Viktor Dukhovni*
44
45 * Fixed possible buffer overread in SSL_select_next_proto().
46
47   Calling the OpenSSL API function SSL_select_next_proto with an empty
48   supported client protocols buffer may cause a crash or memory contents
49   to be sent to the peer.
50
51   ([CVE-2024-5535])
52
53   *Matt Caswell*
54
55### Changes between 3.0.13 and 3.0.14 [4 Jun 2024]
56
57 * Fixed potential use after free after SSL_free_buffers() is called.
58
59   The SSL_free_buffers function is used to free the internal OpenSSL
60   buffer used when processing an incoming record from the network.
61   The call is only expected to succeed if the buffer is not currently
62   in use. However, two scenarios have been identified where the buffer
63   is freed even when still in use.
64
65   The first scenario occurs where a record header has been received
66   from the network and processed by OpenSSL, but the full record body
67   has not yet arrived. In this case calling SSL_free_buffers will succeed
68   even though a record has only been partially processed and the buffer
69   is still in use.
70
71   The second scenario occurs where a full record containing application
72   data has been received and processed by OpenSSL but the application has
73   only read part of this data. Again a call to SSL_free_buffers will
74   succeed even though the buffer is still in use.
75
76   ([CVE-2024-4741])
77
78   *Matt Caswell*
79
80 * Fixed an issue where checking excessively long DSA keys or parameters may
81   be very slow.
82
83   Applications that use the functions EVP_PKEY_param_check() or
84   EVP_PKEY_public_check() to check a DSA public key or DSA parameters may
85   experience long delays. Where the key or parameters that are being checked
86   have been obtained from an untrusted source this may lead to a Denial of
87   Service.
88
89   To resolve this issue DSA keys larger than OPENSSL_DSA_MAX_MODULUS_BITS
90   will now fail the check immediately with a DSA_R_MODULUS_TOO_LARGE error
91   reason.
92
93   ([CVE-2024-4603])
94
95   *Tomáš Mráz*
96
97 * Improved EC/DSA nonce generation routines to avoid bias and timing
98   side channel leaks.
99
100   Thanks to Florian Sieck from Universität zu Lübeck and George Pantelakis
101   and Hubert Kario from Red Hat for reporting the issues.
102
103   *Tomáš Mráz and Paul Dale*
104
105 * Fixed an issue where some non-default TLS server configurations can cause
106   unbounded memory growth when processing TLSv1.3 sessions. An attacker may
107   exploit certain server configurations to trigger unbounded memory growth that
108   would lead to a Denial of Service
109
110   This problem can occur in TLSv1.3 if the non-default SSL_OP_NO_TICKET option
111   is being used (but not if early_data is also configured and the default
112   anti-replay protection is in use). In this case, under certain conditions,
113   the session cache can get into an incorrect state and it will fail to flush
114   properly as it fills. The session cache will continue to grow in an unbounded
115   manner. A malicious client could deliberately create the scenario for this
116   failure to force a Denial of Service. It may also happen by accident in
117   normal operation.
118
119   ([CVE-2024-2511])
120
121   *Matt Caswell*
122
123 * New atexit configuration switch, which controls whether the OPENSSL_cleanup
124   is registered when libcrypto is unloaded. This can be used on platforms
125   where using atexit() from shared libraries causes crashes on exit.
126
127   *Randall S. Becker*
128
129### Changes between 3.0.12 and 3.0.13 [30 Jan 2024]
130
131 * A file in PKCS12 format can contain certificates and keys and may come from
132   an untrusted source. The PKCS12 specification allows certain fields to be
133   NULL, but OpenSSL did not correctly check for this case. A fix has been
134   applied to prevent a NULL pointer dereference that results in OpenSSL
135   crashing. If an application processes PKCS12 files from an untrusted source
136   using the OpenSSL APIs then that application will be vulnerable to this
137   issue prior to this fix.
138
139   OpenSSL APIs that were vulnerable to this are: PKCS12_parse(),
140   PKCS12_unpack_p7data(), PKCS12_unpack_p7encdata(), PKCS12_unpack_authsafes()
141   and PKCS12_newpass().
142
143   We have also fixed a similar issue in SMIME_write_PKCS7(). However since this
144   function is related to writing data we do not consider it security
145   significant.
146
147   ([CVE-2024-0727])
148
149   *Matt Caswell*
150
151 * When function EVP_PKEY_public_check() is called on RSA public keys,
152   a computation is done to confirm that the RSA modulus, n, is composite.
153   For valid RSA keys, n is a product of two or more large primes and this
154   computation completes quickly. However, if n is an overly large prime,
155   then this computation would take a long time.
156
157   An application that calls EVP_PKEY_public_check() and supplies an RSA key
158   obtained from an untrusted source could be vulnerable to a Denial of Service
159   attack.
160
161   The function EVP_PKEY_public_check() is not called from other OpenSSL
162   functions however it is called from the OpenSSL pkey command line
163   application. For that reason that application is also vulnerable if used
164   with the "-pubin" and "-check" options on untrusted data.
165
166   To resolve this issue RSA keys larger than OPENSSL_RSA_MAX_MODULUS_BITS will
167   now fail the check immediately with an RSA_R_MODULUS_TOO_LARGE error reason.
168
169   ([CVE-2023-6237])
170
171   *Tomáš Mráz*
172
173 * Restore the encoding of SM2 PrivateKeyInfo and SubjectPublicKeyInfo to
174   have the contained AlgorithmIdentifier.algorithm set to id-ecPublicKey
175   rather than SM2.
176
177   *Richard Levitte*
178
179 * The POLY1305 MAC (message authentication code) implementation in OpenSSL
180   for PowerPC CPUs saves the contents of vector registers in different
181   order than they are restored. Thus the contents of some of these vector
182   registers is corrupted when returning to the caller. The vulnerable code is
183   used only on newer PowerPC processors supporting the PowerISA 2.07
184   instructions.
185
186   The consequences of this kind of internal application state corruption can
187   be various - from no consequences, if the calling application does not
188   depend on the contents of non-volatile XMM registers at all, to the worst
189   consequences, where the attacker could get complete control of the
190   application process. However unless the compiler uses the vector registers
191   for storing pointers, the most likely consequence, if any, would be an
192   incorrect result of some application dependent calculations or a crash
193   leading to a denial of service.
194
195   ([CVE-2023-6129])
196
197   *Rohan McLure*
198
199 * Fix excessive time spent in DH check / generation with large Q parameter
200   value.
201
202   Applications that use the functions DH_generate_key() to generate an
203   X9.42 DH key may experience long delays. Likewise, applications that use
204   DH_check_pub_key(), DH_check_pub_key_ex() or EVP_PKEY_public_check()
205   to check an X9.42 DH key or X9.42 DH parameters may experience long delays.
206   Where the key or parameters that are being checked have been obtained from
207   an untrusted source this may lead to a Denial of Service.
208
209   ([CVE-2023-5678])
210
211   *Richard Levitte*
212
213### Changes between 3.0.11 and 3.0.12 [24 Oct 2023]
214
215 * Fix incorrect key and IV resizing issues when calling EVP_EncryptInit_ex2(),
216   EVP_DecryptInit_ex2() or EVP_CipherInit_ex2() with OSSL_PARAM parameters
217   that alter the key or IV length ([CVE-2023-5363]).
218
219   *Paul Dale*
220
221### Changes between 3.0.10 and 3.0.11 [19 Sep 2023]
222
223 * Fix POLY1305 MAC implementation corrupting XMM registers on Windows.
224
225   The POLY1305 MAC (message authentication code) implementation in OpenSSL
226   does not save the contents of non-volatile XMM registers on Windows 64
227   platform when calculating the MAC of data larger than 64 bytes. Before
228   returning to the caller all the XMM registers are set to zero rather than
229   restoring their previous content. The vulnerable code is used only on newer
230   x86_64 processors supporting the AVX512-IFMA instructions.
231
232   The consequences of this kind of internal application state corruption can
233   be various - from no consequences, if the calling application does not
234   depend on the contents of non-volatile XMM registers at all, to the worst
235   consequences, where the attacker could get complete control of the
236   application process. However given the contents of the registers are just
237   zeroized so the attacker cannot put arbitrary values inside, the most likely
238   consequence, if any, would be an incorrect result of some application
239   dependent calculations or a crash leading to a denial of service.
240
241   ([CVE-2023-4807])
242
243   *Bernd Edlinger*
244
245### Changes between 3.0.9 and 3.0.10 [1 Aug 2023]
246
247 * Fix excessive time spent checking DH q parameter value.
248
249   The function DH_check() performs various checks on DH parameters. After
250   fixing CVE-2023-3446 it was discovered that a large q parameter value can
251   also trigger an overly long computation during some of these checks.
252   A correct q value, if present, cannot be larger than the modulus p
253   parameter, thus it is unnecessary to perform these checks if q is larger
254   than p.
255
256   If DH_check() is called with such q parameter value,
257   DH_CHECK_INVALID_Q_VALUE return flag is set and the computationally
258   intensive checks are skipped.
259
260   ([CVE-2023-3817])
261
262   *Tomáš Mráz*
263
264 * Fix DH_check() excessive time with over sized modulus.
265
266   The function DH_check() performs various checks on DH parameters. One of
267   those checks confirms that the modulus ("p" parameter) is not too large.
268   Trying to use a very large modulus is slow and OpenSSL will not normally use
269   a modulus which is over 10,000 bits in length.
270
271   However the DH_check() function checks numerous aspects of the key or
272   parameters that have been supplied. Some of those checks use the supplied
273   modulus value even if it has already been found to be too large.
274
275   A new limit has been added to DH_check of 32,768 bits. Supplying a
276   key/parameters with a modulus over this size will simply cause DH_check() to
277   fail.
278
279   ([CVE-2023-3446])
280
281   *Matt Caswell*
282
283 * Do not ignore empty associated data entries with AES-SIV.
284
285   The AES-SIV algorithm allows for authentication of multiple associated
286   data entries along with the encryption. To authenticate empty data the
287   application has to call `EVP_EncryptUpdate()` (or `EVP_CipherUpdate()`)
288   with NULL pointer as the output buffer and 0 as the input buffer length.
289   The AES-SIV implementation in OpenSSL just returns success for such call
290   instead of performing the associated data authentication operation.
291   The empty data thus will not be authenticated. ([CVE-2023-2975])
292
293   Thanks to Juerg Wullschleger (Google) for discovering the issue.
294
295   The fix changes the authentication tag value and the ciphertext for
296   applications that use empty associated data entries with AES-SIV.
297   To decrypt data encrypted with previous versions of OpenSSL the application
298   has to skip calls to `EVP_DecryptUpdate()` for empty associated data
299   entries.
300
301   *Tomáš Mráz*
302
303### Changes between 3.0.8 and 3.0.9 [30 May 2023]
304
305 * Mitigate for the time it takes for `OBJ_obj2txt` to translate gigantic
306   OBJECT IDENTIFIER sub-identifiers to canonical numeric text form.
307
308   OBJ_obj2txt() would translate any size OBJECT IDENTIFIER to canonical
309   numeric text form.  For gigantic sub-identifiers, this would take a very
310   long time, the time complexity being O(n^2) where n is the size of that
311   sub-identifier.  ([CVE-2023-2650])
312
313   To mitigitate this, `OBJ_obj2txt()` will only translate an OBJECT
314   IDENTIFIER to canonical numeric text form if the size of that OBJECT
315   IDENTIFIER is 586 bytes or less, and fail otherwise.
316
317   The basis for this restriction is [RFC 2578 (STD 58), section 3.5]. OBJECT
318   IDENTIFIER values, which stipulates that OBJECT IDENTIFIERS may have at
319   most 128 sub-identifiers, and that the maximum value that each sub-
320   identifier may have is 2^32-1 (4294967295 decimal).
321
322   For each byte of every sub-identifier, only the 7 lower bits are part of
323   the value, so the maximum amount of bytes that an OBJECT IDENTIFIER with
324   these restrictions may occupy is 32 * 128 / 7, which is approximately 586
325   bytes.
326
327   *Richard Levitte*
328
329 * Fixed buffer overread in AES-XTS decryption on ARM 64 bit platforms which
330   happens if the buffer size is 4 mod 5 in 16 byte AES blocks. This can
331   trigger a crash of an application using AES-XTS decryption if the memory
332   just after the buffer being decrypted is not mapped.
333   Thanks to Anton Romanov (Amazon) for discovering the issue.
334   ([CVE-2023-1255])
335
336   *Nevine Ebeid*
337
338 * Reworked the Fix for the Timing Oracle in RSA Decryption ([CVE-2022-4304]).
339   The previous fix for this timing side channel turned out to cause
340   a severe 2-3x performance regression in the typical use case
341   compared to 3.0.7. The new fix uses existing constant time
342   code paths, and restores the previous performance level while
343   fully eliminating all existing timing side channels.
344   The fix was developed by Bernd Edlinger with testing support
345   by Hubert Kario.
346
347   *Bernd Edlinger*
348
349 * Corrected documentation of X509_VERIFY_PARAM_add0_policy() to mention
350   that it does not enable policy checking. Thanks to David Benjamin for
351   discovering this issue.
352   ([CVE-2023-0466])
353
354   *Tomáš Mráz*
355
356 * Fixed an issue where invalid certificate policies in leaf certificates are
357   silently ignored by OpenSSL and other certificate policy checks are skipped
358   for that certificate. A malicious CA could use this to deliberately assert
359   invalid certificate policies in order to circumvent policy checking on the
360   certificate altogether.
361   ([CVE-2023-0465])
362
363   *Matt Caswell*
364
365 * Limited the number of nodes created in a policy tree to mitigate
366   against CVE-2023-0464.  The default limit is set to 1000 nodes, which
367   should be sufficient for most installations.  If required, the limit
368   can be adjusted by setting the OPENSSL_POLICY_TREE_NODES_MAX build
369   time define to a desired maximum number of nodes or zero to allow
370   unlimited growth.
371   ([CVE-2023-0464])
372
373   *Paul Dale*
374
375### Changes between 3.0.7 and 3.0.8 [7 Feb 2023]
376
377 * Fixed NULL dereference during PKCS7 data verification.
378
379   A NULL pointer can be dereferenced when signatures are being
380   verified on PKCS7 signed or signedAndEnveloped data. In case the hash
381   algorithm used for the signature is known to the OpenSSL library but
382   the implementation of the hash algorithm is not available the digest
383   initialization will fail. There is a missing check for the return
384   value from the initialization function which later leads to invalid
385   usage of the digest API most likely leading to a crash.
386   ([CVE-2023-0401])
387
388   PKCS7 data is processed by the SMIME library calls and also by the
389   time stamp (TS) library calls. The TLS implementation in OpenSSL does
390   not call these functions however third party applications would be
391   affected if they call these functions to verify signatures on untrusted
392   data.
393
394   *Tomáš Mráz*
395
396 * Fixed X.400 address type confusion in X.509 GeneralName.
397
398   There is a type confusion vulnerability relating to X.400 address processing
399   inside an X.509 GeneralName. X.400 addresses were parsed as an ASN1_STRING
400   but the public structure definition for GENERAL_NAME incorrectly specified
401   the type of the x400Address field as ASN1_TYPE. This field is subsequently
402   interpreted by the OpenSSL function GENERAL_NAME_cmp as an ASN1_TYPE rather
403   than an ASN1_STRING.
404
405   When CRL checking is enabled (i.e. the application sets the
406   X509_V_FLAG_CRL_CHECK flag), this vulnerability may allow an attacker to
407   pass arbitrary pointers to a memcmp call, enabling them to read memory
408   contents or enact a denial of service.
409   ([CVE-2023-0286])
410
411   *Hugo Landau*
412
413 * Fixed NULL dereference validating DSA public key.
414
415   An invalid pointer dereference on read can be triggered when an
416   application tries to check a malformed DSA public key by the
417   EVP_PKEY_public_check() function. This will most likely lead
418   to an application crash. This function can be called on public
419   keys supplied from untrusted sources which could allow an attacker
420   to cause a denial of service attack.
421
422   The TLS implementation in OpenSSL does not call this function
423   but applications might call the function if there are additional
424   security requirements imposed by standards such as FIPS 140-3.
425   ([CVE-2023-0217])
426
427   *Shane Lontis, Tomáš Mráz*
428
429 * Fixed Invalid pointer dereference in d2i_PKCS7 functions.
430
431   An invalid pointer dereference on read can be triggered when an
432   application tries to load malformed PKCS7 data with the
433   d2i_PKCS7(), d2i_PKCS7_bio() or d2i_PKCS7_fp() functions.
434
435   The result of the dereference is an application crash which could
436   lead to a denial of service attack. The TLS implementation in OpenSSL
437   does not call this function however third party applications might
438   call these functions on untrusted data.
439   ([CVE-2023-0216])
440
441   *Tomáš Mráz*
442
443 * Fixed Use-after-free following BIO_new_NDEF.
444
445   The public API function BIO_new_NDEF is a helper function used for
446   streaming ASN.1 data via a BIO. It is primarily used internally to OpenSSL
447   to support the SMIME, CMS and PKCS7 streaming capabilities, but may also
448   be called directly by end user applications.
449
450   The function receives a BIO from the caller, prepends a new BIO_f_asn1
451   filter BIO onto the front of it to form a BIO chain, and then returns
452   the new head of the BIO chain to the caller. Under certain conditions,
453   for example if a CMS recipient public key is invalid, the new filter BIO
454   is freed and the function returns a NULL result indicating a failure.
455   However, in this case, the BIO chain is not properly cleaned up and the
456   BIO passed by the caller still retains internal pointers to the previously
457   freed filter BIO. If the caller then goes on to call BIO_pop() on the BIO
458   then a use-after-free will occur. This will most likely result in a crash.
459   ([CVE-2023-0215])
460
461   *Viktor Dukhovni, Matt Caswell*
462
463 * Fixed Double free after calling PEM_read_bio_ex.
464
465   The function PEM_read_bio_ex() reads a PEM file from a BIO and parses and
466   decodes the "name" (e.g. "CERTIFICATE"), any header data and the payload
467   data. If the function succeeds then the "name_out", "header" and "data"
468   arguments are populated with pointers to buffers containing the relevant
469   decoded data. The caller is responsible for freeing those buffers. It is
470   possible to construct a PEM file that results in 0 bytes of payload data.
471   In this case PEM_read_bio_ex() will return a failure code but will populate
472   the header argument with a pointer to a buffer that has already been freed.
473   If the caller also frees this buffer then a double free will occur. This
474   will most likely lead to a crash.
475
476   The functions PEM_read_bio() and PEM_read() are simple wrappers around
477   PEM_read_bio_ex() and therefore these functions are also directly affected.
478
479   These functions are also called indirectly by a number of other OpenSSL
480   functions including PEM_X509_INFO_read_bio_ex() and
481   SSL_CTX_use_serverinfo_file() which are also vulnerable. Some OpenSSL
482   internal uses of these functions are not vulnerable because the caller does
483   not free the header argument if PEM_read_bio_ex() returns a failure code.
484   ([CVE-2022-4450])
485
486   *Kurt Roeckx, Matt Caswell*
487
488 * Fixed Timing Oracle in RSA Decryption.
489
490   A timing based side channel exists in the OpenSSL RSA Decryption
491   implementation which could be sufficient to recover a plaintext across
492   a network in a Bleichenbacher style attack. To achieve a successful
493   decryption an attacker would have to be able to send a very large number
494   of trial messages for decryption. The vulnerability affects all RSA padding
495   modes: PKCS#1 v1.5, RSA-OEAP and RSASVE.
496   ([CVE-2022-4304])
497
498   *Dmitry Belyavsky, Hubert Kario*
499
500 * Fixed X.509 Name Constraints Read Buffer Overflow.
501
502   A read buffer overrun can be triggered in X.509 certificate verification,
503   specifically in name constraint checking. The read buffer overrun might
504   result in a crash which could lead to a denial of service attack.
505   In a TLS client, this can be triggered by connecting to a malicious
506   server. In a TLS server, this can be triggered if the server requests
507   client authentication and a malicious client connects.
508   ([CVE-2022-4203])
509
510   *Viktor Dukhovni*
511
512 * Fixed X.509 Policy Constraints Double Locking security issue.
513
514   If an X.509 certificate contains a malformed policy constraint and
515   policy processing is enabled, then a write lock will be taken twice
516   recursively.  On some operating systems (most widely: Windows) this
517   results in a denial of service when the affected process hangs.  Policy
518   processing being enabled on a publicly facing server is not considered
519   to be a common setup.
520   ([CVE-2022-3996])
521
522   *Paul Dale*
523
524 * Our provider implementations of `OSSL_FUNC_KEYMGMT_EXPORT` and
525   `OSSL_FUNC_KEYMGMT_GET_PARAMS` for EC and SM2 keys now honor
526   `OSSL_PKEY_PARAM_EC_POINT_CONVERSION_FORMAT` as set (and
527   default to `POINT_CONVERSION_UNCOMPRESSED`) when exporting
528   `OSSL_PKEY_PARAM_PUB_KEY`, instead of unconditionally using
529   `POINT_CONVERSION_COMPRESSED` as in previous 3.x releases.
530   For symmetry, our implementation of `EVP_PKEY_ASN1_METHOD->export_to`
531   for legacy EC and SM2 keys is also changed similarly to honor the
532   equivalent conversion format flag as specified in the underlying
533   `EC_KEY` object being exported to a provider, when this function is
534   called through `EVP_PKEY_export()`.
535
536   *Nicola Tuveri*
537
538### Changes between 3.0.6 and 3.0.7 [1 Nov 2022]
539
540 * Fixed two buffer overflows in punycode decoding functions.
541
542   A buffer overrun can be triggered in X.509 certificate verification,
543   specifically in name constraint checking. Note that this occurs after
544   certificate chain signature verification and requires either a CA to
545   have signed the malicious certificate or for the application to continue
546   certificate verification despite failure to construct a path to a trusted
547   issuer.
548
549   In a TLS client, this can be triggered by connecting to a malicious
550   server.  In a TLS server, this can be triggered if the server requests
551   client authentication and a malicious client connects.
552
553   An attacker can craft a malicious email address to overflow
554   an arbitrary number of bytes containing the `.`  character (decimal 46)
555   on the stack.  This buffer overflow could result in a crash (causing a
556   denial of service).
557   ([CVE-2022-3786])
558
559   An attacker can craft a malicious email address to overflow four
560   attacker-controlled bytes on the stack.  This buffer overflow could
561   result in a crash (causing a denial of service) or potentially remote code
562   execution depending on stack layout for any given platform/compiler.
563   ([CVE-2022-3602])
564
565   *Paul Dale*
566
567 * Removed all references to invalid OSSL_PKEY_PARAM_RSA names for CRT
568   parameters in OpenSSL code.
569   Applications should not use the names OSSL_PKEY_PARAM_RSA_FACTOR,
570   OSSL_PKEY_PARAM_RSA_EXPONENT and OSSL_PKEY_PARAM_RSA_COEFFICIENT.
571   Use the numbered names such as OSSL_PKEY_PARAM_RSA_FACTOR1 instead.
572   Using these invalid names may cause algorithms to use slower methods
573   that ignore the CRT parameters.
574
575   *Shane Lontis*
576
577 * Fixed a regression introduced in 3.0.6 version raising errors on some stack
578   operations.
579
580   *Tomáš Mráz*
581
582 * Fixed a regression introduced in 3.0.6 version not refreshing the certificate
583   data to be signed before signing the certificate.
584
585   *Gibeom Gwon*
586
587 * Added RIPEMD160 to the default provider.
588
589   *Paul Dale*
590
591 * Ensured that the key share group sent or accepted for the key exchange
592   is allowed for the protocol version.
593
594   *Matt Caswell*
595
596### Changes between 3.0.5 and 3.0.6 [11 Oct 2022]
597
598 * OpenSSL supports creating a custom cipher via the legacy
599   EVP_CIPHER_meth_new() function and associated function calls. This function
600   was deprecated in OpenSSL 3.0 and application authors are instead encouraged
601   to use the new provider mechanism in order to implement custom ciphers.
602
603   OpenSSL versions 3.0.0 to 3.0.5 incorrectly handle legacy custom ciphers
604   passed to the EVP_EncryptInit_ex2(), EVP_DecryptInit_ex2() and
605   EVP_CipherInit_ex2() functions (as well as other similarly named encryption
606   and decryption initialisation functions). Instead of using the custom cipher
607   directly it incorrectly tries to fetch an equivalent cipher from the
608   available providers. An equivalent cipher is found based on the NID passed to
609   EVP_CIPHER_meth_new(). This NID is supposed to represent the unique NID for a
610   given cipher. However it is possible for an application to incorrectly pass
611   NID_undef as this value in the call to EVP_CIPHER_meth_new(). When NID_undef
612   is used in this way the OpenSSL encryption/decryption initialisation function
613   will match the NULL cipher as being equivalent and will fetch this from the
614   available providers. This will succeed if the default provider has been
615   loaded (or if a third party provider has been loaded that offers this
616   cipher). Using the NULL cipher means that the plaintext is emitted as the
617   ciphertext.
618
619   Applications are only affected by this issue if they call
620   EVP_CIPHER_meth_new() using NID_undef and subsequently use it in a call to an
621   encryption/decryption initialisation function. Applications that only use
622   SSL/TLS are not impacted by this issue.
623   ([CVE-2022-3358])
624
625   *Matt Caswell*
626
627 * Fix LLVM vs Apple LLVM version numbering confusion that caused build failures
628   on MacOS 10.11
629
630   *Richard Levitte*
631
632 * Fixed the linux-mips64 Configure target which was missing the
633   SIXTY_FOUR_BIT bn_ops flag. This was causing heap corruption on that
634   platform.
635
636   *Adam Joseph*
637
638 * Fix handling of a ticket key callback that returns 0 in TLSv1.3 to not send a
639   ticket
640
641   *Matt Caswell*
642
643 * Correctly handle a retransmitted ClientHello in DTLS
644
645   *Matt Caswell*
646
647 * Fixed detection of ktls support in cross-compile environment on Linux
648
649   *Tomas Mraz*
650
651 * Fixed some regressions and test failures when running the 3.0.0 FIPS provider
652   against 3.0.x
653
654   *Paul Dale*
655
656 * Fixed SSL_pending() and SSL_has_pending() with DTLS which were failing to
657   report correct results in some cases
658
659   *Matt Caswell*
660
661 * Fix UWP builds by defining VirtualLock
662
663   *Charles Milette*
664
665 * For known safe primes use the minimum key length according to RFC 7919.
666   Longer private key sizes unnecessarily raise the cycles needed to compute the
667   shared secret without any increase of the real security. This fixes a
668   regression from 1.1.1 where these shorter keys were generated for the known
669   safe primes.
670
671   *Tomas Mraz*
672
673 * Added the loongarch64 target
674
675   *Shi Pujin*
676
677 * Fixed EC ASM flag passing. Flags for ASM implementations of EC curves were
678   only passed to the FIPS provider and not to the default or legacy provider.
679
680   *Juergen Christ*
681
682 * Fixed reported performance degradation on aarch64. Restored the
683   implementation prior to commit 2621751 ("aes/asm/aesv8-armx.pl: avoid
684   32-bit lane assignment in CTR mode") for 64bit targets only, since it is
685   reportedly 2-17% slower and the silicon errata only affects 32bit targets.
686   The new algorithm is still used for 32 bit targets.
687
688   *Bernd Edlinger*
689
690 * Added a missing header for memcmp that caused compilation failure on some
691   platforms
692
693   *Gregor Jasny*
694
695### Changes between 3.0.4 and 3.0.5 [5 Jul 2022]
696
697 * The OpenSSL 3.0.4 release introduced a serious bug in the RSA
698   implementation for X86_64 CPUs supporting the AVX512IFMA instructions.
699   This issue makes the RSA implementation with 2048 bit private keys
700   incorrect on such machines and memory corruption will happen during
701   the computation. As a consequence of the memory corruption an attacker
702   may be able to trigger a remote code execution on the machine performing
703   the computation.
704
705   SSL/TLS servers or other servers using 2048 bit RSA private keys running
706   on machines supporting AVX512IFMA instructions of the X86_64 architecture
707   are affected by this issue.
708   ([CVE-2022-2274])
709
710   *Xi Ruoyao*
711
712 * AES OCB mode for 32-bit x86 platforms using the AES-NI assembly optimised
713   implementation would not encrypt the entirety of the data under some
714   circumstances.  This could reveal sixteen bytes of data that was
715   preexisting in the memory that wasn't written.  In the special case of
716   "in place" encryption, sixteen bytes of the plaintext would be revealed.
717
718   Since OpenSSL does not support OCB based cipher suites for TLS and DTLS,
719   they are both unaffected.
720   ([CVE-2022-2097])
721
722   *Alex Chernyakhovsky, David Benjamin, Alejandro Sedeño*
723
724### Changes between 3.0.3 and 3.0.4 [21 Jun 2022]
725
726 * In addition to the c_rehash shell command injection identified in
727   CVE-2022-1292, further bugs where the c_rehash script does not
728   properly sanitise shell metacharacters to prevent command injection have been
729   fixed.
730
731   When the CVE-2022-1292 was fixed it was not discovered that there
732   are other places in the script where the file names of certificates
733   being hashed were possibly passed to a command executed through the shell.
734
735   This script is distributed by some operating systems in a manner where
736   it is automatically executed.  On such operating systems, an attacker
737   could execute arbitrary commands with the privileges of the script.
738
739   Use of the c_rehash script is considered obsolete and should be replaced
740   by the OpenSSL rehash command line tool.
741   (CVE-2022-2068)
742
743   *Daniel Fiala, Tomáš Mráz*
744
745 * Case insensitive string comparison no longer uses locales.  It has instead
746   been directly implemented.
747
748   *Paul Dale*
749
750### Changes between 3.0.2 and 3.0.3 [3 May 2022]
751
752 * Case insensitive string comparison is reimplemented via new locale-agnostic
753   comparison functions OPENSSL_str[n]casecmp always using the POSIX locale for
754   comparison. The previous implementation had problems when the Turkish locale
755   was used.
756
757   *Dmitry Belyavskiy*
758
759 * Fixed a bug in the c_rehash script which was not properly sanitising shell
760   metacharacters to prevent command injection.  This script is distributed by
761   some operating systems in a manner where it is automatically executed.  On
762   such operating systems, an attacker could execute arbitrary commands with the
763   privileges of the script.
764
765   Use of the c_rehash script is considered obsolete and should be replaced
766   by the OpenSSL rehash command line tool.
767   (CVE-2022-1292)
768
769   *Tomáš Mráz*
770
771 * Fixed a bug in the function `OCSP_basic_verify` that verifies the signer
772   certificate on an OCSP response. The bug caused the function in the case
773   where the (non-default) flag OCSP_NOCHECKS is used to return a postivie
774   response (meaning a successful verification) even in the case where the
775   response signing certificate fails to verify.
776
777   It is anticipated that most users of `OCSP_basic_verify` will not use the
778   OCSP_NOCHECKS flag. In this case the `OCSP_basic_verify` function will return
779   a negative value (indicating a fatal error) in the case of a certificate
780   verification failure. The normal expected return value in this case would be
781   0.
782
783   This issue also impacts the command line OpenSSL "ocsp" application. When
784   verifying an ocsp response with the "-no_cert_checks" option the command line
785   application will report that the verification is successful even though it
786   has in fact failed. In this case the incorrect successful response will also
787   be accompanied by error messages showing the failure and contradicting the
788   apparently successful result.
789   ([CVE-2022-1343])
790
791   *Matt Caswell*
792
793 * Fixed a bug where the RC4-MD5 ciphersuite incorrectly used the
794   AAD data as the MAC key. This made the MAC key trivially predictable.
795
796   An attacker could exploit this issue by performing a man-in-the-middle attack
797   to modify data being sent from one endpoint to an OpenSSL 3.0 recipient such
798   that the modified data would still pass the MAC integrity check.
799
800   Note that data sent from an OpenSSL 3.0 endpoint to a non-OpenSSL 3.0
801   endpoint will always be rejected by the recipient and the connection will
802   fail at that point. Many application protocols require data to be sent from
803   the client to the server first. Therefore, in such a case, only an OpenSSL
804   3.0 server would be impacted when talking to a non-OpenSSL 3.0 client.
805
806   If both endpoints are OpenSSL 3.0 then the attacker could modify data being
807   sent in both directions. In this case both clients and servers could be
808   affected, regardless of the application protocol.
809
810   Note that in the absence of an attacker this bug means that an OpenSSL 3.0
811   endpoint communicating with a non-OpenSSL 3.0 endpoint will fail to complete
812   the handshake when using this ciphersuite.
813
814   The confidentiality of data is not impacted by this issue, i.e. an attacker
815   cannot decrypt data that has been encrypted using this ciphersuite - they can
816   only modify it.
817
818   In order for this attack to work both endpoints must legitimately negotiate
819   the RC4-MD5 ciphersuite. This ciphersuite is not compiled by default in
820   OpenSSL 3.0, and is not available within the default provider or the default
821   ciphersuite list. This ciphersuite will never be used if TLSv1.3 has been
822   negotiated. In order for an OpenSSL 3.0 endpoint to use this ciphersuite the
823   following must have occurred:
824
825   1) OpenSSL must have been compiled with the (non-default) compile time option
826      enable-weak-ssl-ciphers
827
828   2) OpenSSL must have had the legacy provider explicitly loaded (either
829      through application code or via configuration)
830
831   3) The ciphersuite must have been explicitly added to the ciphersuite list
832
833   4) The libssl security level must have been set to 0 (default is 1)
834
835   5) A version of SSL/TLS below TLSv1.3 must have been negotiated
836
837   6) Both endpoints must negotiate the RC4-MD5 ciphersuite in preference to any
838      others that both endpoints have in common
839   (CVE-2022-1434)
840
841   *Matt Caswell*
842
843 * Fix a bug in the OPENSSL_LH_flush() function that breaks reuse of the memory
844   occuppied by the removed hash table entries.
845
846   This function is used when decoding certificates or keys. If a long lived
847   process periodically decodes certificates or keys its memory usage will
848   expand without bounds and the process might be terminated by the operating
849   system causing a denial of service. Also traversing the empty hash table
850   entries will take increasingly more time.
851
852   Typically such long lived processes might be TLS clients or TLS servers
853   configured to accept client certificate authentication.
854   (CVE-2022-1473)
855
856   *Hugo Landau, Aliaksei Levin*
857
858 * The functions `OPENSSL_LH_stats` and `OPENSSL_LH_stats_bio` now only report
859   the `num_items`, `num_nodes` and `num_alloc_nodes` statistics. All other
860   statistics are no longer supported. For compatibility, these statistics are
861   still listed in the output but are now always reported as zero.
862
863   *Hugo Landau*
864
865### Changes between 3.0.1 and 3.0.2 [15 Mar 2022]
866
867 * Fixed a bug in the BN_mod_sqrt() function that can cause it to loop forever
868   for non-prime moduli.
869
870   Internally this function is used when parsing certificates that contain
871   elliptic curve public keys in compressed form or explicit elliptic curve
872   parameters with a base point encoded in compressed form.
873
874   It is possible to trigger the infinite loop by crafting a certificate that
875   has invalid explicit curve parameters.
876
877   Since certificate parsing happens prior to verification of the certificate
878   signature, any process that parses an externally supplied certificate may thus
879   be subject to a denial of service attack. The infinite loop can also be
880   reached when parsing crafted private keys as they can contain explicit
881   elliptic curve parameters.
882
883   Thus vulnerable situations include:
884
885    - TLS clients consuming server certificates
886    - TLS servers consuming client certificates
887    - Hosting providers taking certificates or private keys from customers
888    - Certificate authorities parsing certification requests from subscribers
889    - Anything else which parses ASN.1 elliptic curve parameters
890
891   Also any other applications that use the BN_mod_sqrt() where the attacker
892   can control the parameter values are vulnerable to this DoS issue.
893   ([CVE-2022-0778])
894
895   *Tomáš Mráz*
896
897 * Add ciphersuites based on DHE_PSK (RFC 4279) and ECDHE_PSK (RFC 5489)
898   to the list of ciphersuites providing Perfect Forward Secrecy as
899   required by SECLEVEL >= 3.
900
901   *Dmitry Belyavskiy, Nicola Tuveri*
902
903 * Made the AES constant time code for no-asm configurations
904   optional due to the resulting 95% performance degradation.
905   The AES constant time code can be enabled, for no assembly
906   builds, with: ./config no-asm -DOPENSSL_AES_CONST_TIME
907
908   *Paul Dale*
909
910 * Fixed PEM_write_bio_PKCS8PrivateKey() to make it possible to use empty
911   passphrase strings.
912
913   *Darshan Sen*
914
915 * The negative return value handling of the certificate verification callback
916   was reverted. The replacement is to set the verification retry state with
917   the SSL_set_retry_verify() function.
918
919   *Tomáš Mráz*
920
921### Changes between 3.0.0 and 3.0.1 [14 Dec 2021]
922
923 * Fixed invalid handling of X509_verify_cert() internal errors in libssl
924   Internally libssl in OpenSSL calls X509_verify_cert() on the client side to
925   verify a certificate supplied by a server. That function may return a
926   negative return value to indicate an internal error (for example out of
927   memory). Such a negative return value is mishandled by OpenSSL and will cause
928   an IO function (such as SSL_connect() or SSL_do_handshake()) to not indicate
929   success and a subsequent call to SSL_get_error() to return the value
930   SSL_ERROR_WANT_RETRY_VERIFY. This return value is only supposed to be
931   returned by OpenSSL if the application has previously called
932   SSL_CTX_set_cert_verify_callback(). Since most applications do not do this
933   the SSL_ERROR_WANT_RETRY_VERIFY return value from SSL_get_error() will be
934   totally unexpected and applications may not behave correctly as a result. The
935   exact behaviour will depend on the application but it could result in
936   crashes, infinite loops or other similar incorrect responses.
937
938   This issue is made more serious in combination with a separate bug in OpenSSL
939   3.0 that will cause X509_verify_cert() to indicate an internal error when
940   processing a certificate chain. This will occur where a certificate does not
941   include the Subject Alternative Name extension but where a Certificate
942   Authority has enforced name constraints. This issue can occur even with valid
943   chains.
944   ([CVE-2021-4044])
945
946   *Matt Caswell*
947
948 * Corrected a few file name and file reference bugs in the build,
949   installation and setup scripts, which lead to installation verification
950   failures.  Slightly enhanced the installation verification script.
951
952   *Richard Levitte*
953
954 * Fixed EVP_PKEY_eq() to make it possible to use it with strictly private
955   keys.
956
957   *Richard Levitte*
958
959 * Fixed PVK encoder to properly query for the passphrase.
960
961   *Tomáš Mráz*
962
963 * Multiple fixes in the OSSL_HTTP API functions.
964
965   *David von Oheimb*
966
967 * Allow sign extension in OSSL_PARAM_allocate_from_text() for the
968   OSSL_PARAM_INTEGER data type and return error on negative numbers
969   used with the OSSL_PARAM_UNSIGNED_INTEGER data type. Make
970   OSSL_PARAM_BLD_push_BN{,_pad}() return an error on negative numbers.
971
972   *Richard Levitte*
973
974 * Allow copying uninitialized digest contexts with EVP_MD_CTX_copy_ex.
975
976   *Tomáš Mráz*
977
978 * Fixed detection of ARMv7 and ARM64 CPU features on FreeBSD.
979
980   *Allan Jude*
981
982 * Multiple threading fixes.
983
984   *Matt Caswell*
985
986 * Added NULL digest implementation to keep compatibility with 1.1.1 version.
987
988   *Tomáš Mráz*
989
990 * Allow fetching an operation from the provider that owns an unexportable key
991   as a fallback if that is still allowed by the property query.
992
993   *Richard Levitte*
994
995### Changes between 1.1.1 and 3.0.0 [7 sep 2021]
996
997 * TLS_MAX_VERSION, DTLS_MAX_VERSION and DTLS_MIN_VERSION constants are now
998   deprecated.
999
1000   *Matt Caswell*
1001
1002 * The `OPENSSL_s390xcap` environment variable can be used to set bits in the
1003   S390X capability vector to zero. This simplifies testing of different code
1004   paths on S390X architecture.
1005
1006   *Patrick Steuer*
1007
1008 * Encrypting more than 2^64 TLS records with AES-GCM is disallowed
1009   as per FIPS 140-2 IG A.5 "Key/IV Pair Uniqueness Requirements from
1010   SP 800-38D". The communication will fail at this point.
1011
1012   *Paul Dale*
1013
1014 * The EC_GROUP_clear_free() function is deprecated as there is nothing
1015   confidential in EC_GROUP data.
1016
1017   *Nicola Tuveri*
1018
1019 * The byte order mark (BOM) character is ignored if encountered at the
1020   beginning of a PEM-formatted file.
1021
1022   *Dmitry Belyavskiy*
1023
1024 * Added CMS support for the Russian GOST algorithms.
1025
1026   *Dmitry Belyavskiy*
1027
1028 * Due to move of the implementation of cryptographic operations
1029   to the providers, validation of various operation parameters can
1030   be postponed until the actual operation is executed where previously
1031   it happened immediately when an operation parameter was set.
1032
1033   For example when setting an unsupported curve with
1034   EVP_PKEY_CTX_set_ec_paramgen_curve_nid() this function call will not
1035   fail but later keygen operations with the EVP_PKEY_CTX will fail.
1036
1037   *OpenSSL team members and many third party contributors*
1038
1039 * The EVP_get_cipherbyname() function will return NULL for algorithms such as
1040   "AES-128-SIV", "AES-128-CBC-CTS" and "CAMELLIA-128-CBC-CTS" which were
1041   previously only accessible via low level interfaces. Use EVP_CIPHER_fetch()
1042   instead to retrieve these algorithms from a provider.
1043
1044   *Shane Lontis*
1045
1046 * On build targets where the multilib postfix is set in the build
1047   configuration the libdir directory was changing based on whether
1048   the lib directory with the multilib postfix exists on the system
1049   or not. This unpredictable behavior was removed and eventual
1050   multilib postfix is now always added to the default libdir. Use
1051   `--libdir=lib` to override the libdir if adding the postfix is
1052   undesirable.
1053
1054   *Jan Lána*
1055
1056 * The triple DES key wrap functionality now conforms to RFC 3217 but is
1057   no longer interoperable with OpenSSL 1.1.1.
1058
1059   *Paul Dale*
1060
1061 * The ERR_GET_FUNC() function was removed.  With the loss of meaningful
1062   function codes, this function can only cause problems for calling
1063   applications.
1064
1065   *Paul Dale*
1066
1067 * Add a configurable flag to output date formats as ISO 8601. Does not
1068   change the default date format.
1069
1070   *William Edmisten*
1071
1072 * Version of MSVC earlier than 1300 could get link warnings, which could
1073   be suppressed if the undocumented -DI_CAN_LIVE_WITH_LNK4049 was set.
1074   Support for this flag has been removed.
1075
1076   *Rich Salz*
1077
1078 * Rework and make DEBUG macros consistent. Remove unused -DCONF_DEBUG,
1079   -DBN_CTX_DEBUG, and REF_PRINT. Add a new tracing category and use it for
1080   printing reference counts. Rename -DDEBUG_UNUSED to -DUNUSED_RESULT_DEBUG
1081   Fix BN_DEBUG_RAND so it compiles and, when set, force DEBUG_RAND to be set
1082   also. Rename engine_debug_ref to be ENGINE_REF_PRINT also for consistency.
1083
1084   *Rich Salz*
1085
1086 * The signatures of the functions to get and set options on SSL and
1087   SSL_CTX objects changed from "unsigned long" to "uint64_t" type.
1088   Some source code changes may be required.
1089
1090   *Rich Salz*
1091
1092 * The public definitions of conf_method_st and conf_st have been
1093   deprecated. They will be made opaque in a future release.
1094
1095   *Rich Salz and Tomáš Mráz*
1096
1097 * Client-initiated renegotiation is disabled by default. To allow it, use
1098   the -client_renegotiation option, the SSL_OP_ALLOW_CLIENT_RENEGOTIATION
1099   flag, or the "ClientRenegotiation" config parameter as appropriate.
1100
1101   *Rich Salz*
1102
1103 * Add "abspath" and "includedir" pragma's to config files, to prevent,
1104   or modify relative pathname inclusion.
1105
1106   *Rich Salz*
1107
1108 * OpenSSL includes a cryptographic module that is intended to be FIPS 140-2
1109   validated. Please consult the README-FIPS and
1110   README-PROVIDERS files, as well as the migration guide.
1111
1112   *OpenSSL team members and many third party contributors*
1113
1114 * For the key types DH and DHX the allowed settable parameters are now different.
1115
1116   *Shane Lontis*
1117
1118 * The openssl commands that read keys, certificates, and CRLs now
1119   automatically detect the PEM or DER format of the input files.
1120
1121   *David von Oheimb, Richard Levitte, and Tomáš Mráz*
1122
1123 * Added enhanced PKCS#12 APIs which accept a library context.
1124
1125   *Jon Spillett*
1126
1127 * The default manual page suffix ($MANSUFFIX) has been changed to "ossl"
1128
1129   *Matt Caswell*
1130
1131 * Added support for Kernel TLS (KTLS).
1132
1133   *Boris Pismenny, John Baldwin and Andrew Gallatin*
1134
1135 * Support for RFC 5746 secure renegotiation is now required by default for
1136   SSL or TLS connections to succeed.
1137
1138   *Benjamin Kaduk*
1139
1140 * The signature of the `copy` functional parameter of the
1141   EVP_PKEY_meth_set_copy() function has changed so its `src` argument is
1142   now `const EVP_PKEY_CTX *` instead of `EVP_PKEY_CTX *`. Similarly
1143   the signature of the `pub_decode` functional parameter of the
1144   EVP_PKEY_asn1_set_public() function has changed so its `pub` argument is
1145   now `const X509_PUBKEY *` instead of `X509_PUBKEY *`.
1146
1147   *David von Oheimb*
1148
1149 * The error return values from some control calls (ctrl) have changed.
1150
1151   *Paul Dale*
1152
1153 * A public key check is now performed during EVP_PKEY_derive_set_peer().
1154
1155   *Shane Lontis*
1156
1157 * Many functions in the EVP_ namespace that are getters of values from
1158   implementations or contexts were renamed to include get or get0 in their
1159   names. Old names are provided as macro aliases for compatibility and
1160   are not deprecated.
1161
1162   *Tomáš Mráz*
1163
1164 * The EVP_PKEY_CTRL_PKCS7_ENCRYPT, EVP_PKEY_CTRL_PKCS7_DECRYPT,
1165   EVP_PKEY_CTRL_PKCS7_SIGN, EVP_PKEY_CTRL_CMS_ENCRYPT,
1166   EVP_PKEY_CTRL_CMS_DECRYPT, and EVP_PKEY_CTRL_CMS_SIGN control operations
1167   are deprecated.
1168
1169   *Tomáš Mráz*
1170
1171 * The EVP_PKEY_public_check() and EVP_PKEY_param_check() functions now work for
1172   more key types.
1173
1174 * The output from the command line applications may have minor
1175   changes.
1176
1177   *Paul Dale*
1178
1179 * The output from numerous "printing" may have minor changes.
1180
1181   *David von Oheimb*
1182
1183 * Windows thread synchronization uses read/write primitives (SRWLock) when
1184   supported by the OS, otherwise CriticalSection continues to be used.
1185
1186   *Vincent Drake*
1187
1188 * Add filter BIO BIO_f_readbuffer() that allows BIO_tell() and BIO_seek() to
1189   work on read only BIO source/sinks that do not support these functions.
1190   This allows piping or redirection of a file BIO using stdin to be buffered
1191   into memory. This is used internally in OSSL_DECODER_from_bio().
1192
1193   *Shane Lontis*
1194
1195 * OSSL_STORE_INFO_get_type() may now return an additional value. In 1.1.1
1196   this function would return one of the values OSSL_STORE_INFO_NAME,
1197   OSSL_STORE_INFO_PKEY, OSSL_STORE_INFO_PARAMS, OSSL_STORE_INFO_CERT or
1198   OSSL_STORE_INFO_CRL. Decoded public keys would previously have been reported
1199   as type OSSL_STORE_INFO_PKEY in 1.1.1. In 3.0 decoded public keys are now
1200   reported as having the new type OSSL_STORE_INFO_PUBKEY. Applications
1201   using this function should be amended to handle the changed return value.
1202
1203   *Richard Levitte*
1204
1205 * Improved adherence to Enhanced Security Services (ESS, RFC 2634 and RFC 5035)
1206   for the TSP and CMS Advanced Electronic Signatures (CAdES) implementations.
1207   As required by RFC 5035 check both ESSCertID and ESSCertIDv2 if both present.
1208   Correct the semantics of checking the validation chain in case ESSCertID{,v2}
1209   contains more than one certificate identifier: This means that all
1210   certificates referenced there MUST be part of the validation chain.
1211
1212   *David von Oheimb*
1213
1214 * The implementation of older EVP ciphers related to CAST, IDEA, SEED, RC2, RC4,
1215   RC5, DESX and DES have been moved to the legacy provider.
1216
1217   *Matt Caswell*
1218
1219 * The implementation of the EVP digests MD2, MD4, MDC2, WHIRLPOOL and
1220   RIPEMD-160 have been moved to the legacy provider.
1221
1222   *Matt Caswell*
1223
1224 * The deprecated function EVP_PKEY_get0() now returns NULL being called for a
1225   provided key.
1226
1227   *Dmitry Belyavskiy*
1228
1229 * The deprecated functions EVP_PKEY_get0_RSA(),
1230   EVP_PKEY_get0_DSA(), EVP_PKEY_get0_EC_KEY(), EVP_PKEY_get0_DH(),
1231   EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and EVP_PKEY_get0_siphash() as
1232   well as the similarly named "get1" functions behave differently in
1233   OpenSSL 3.0.
1234
1235   *Matt Caswell*
1236
1237 * A number of functions handling low-level keys or engines were deprecated
1238   including EVP_PKEY_set1_engine(), EVP_PKEY_get0_engine(), EVP_PKEY_assign(),
1239   EVP_PKEY_get0(), EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and
1240   EVP_PKEY_get0_siphash().
1241
1242   *Matt Caswell*
1243
1244 * PKCS#5 PBKDF1 key derivation has been moved from PKCS5_PBE_keyivgen() into
1245   the legacy crypto provider as an EVP_KDF. Applications requiring this KDF
1246   will need to load the legacy crypto provider. This includes these PBE
1247   algorithms which use this KDF:
1248   - NID_pbeWithMD2AndDES_CBC
1249   - NID_pbeWithMD5AndDES_CBC
1250   - NID_pbeWithSHA1AndRC2_CBC
1251   - NID_pbeWithMD2AndRC2_CBC
1252   - NID_pbeWithMD5AndRC2_CBC
1253   - NID_pbeWithSHA1AndDES_CBC
1254
1255   *Jon Spillett*
1256
1257 * Deprecated obsolete BIO_set_callback(), BIO_get_callback(), and
1258   BIO_debug_callback() functions.
1259
1260   *Tomáš Mráz*
1261
1262 * Deprecated obsolete EVP_PKEY_CTX_get0_dh_kdf_ukm() and
1263   EVP_PKEY_CTX_get0_ecdh_kdf_ukm() functions.
1264
1265   *Tomáš Mráz*
1266
1267 * The RAND_METHOD APIs have been deprecated.
1268
1269   *Paul Dale*
1270
1271 * The SRP APIs have been deprecated.
1272
1273   *Matt Caswell*
1274
1275 * Add a compile time option to prevent the caching of provider fetched
1276   algorithms.  This is enabled by including the no-cached-fetch option
1277   at configuration time.
1278
1279   *Paul Dale*
1280
1281 * pkcs12 now uses defaults of PBKDF2, AES and SHA-256, with a MAC iteration
1282   count of PKCS12_DEFAULT_ITER.
1283
1284   *Tomáš Mráz and Sahana Prasad*
1285
1286 * The openssl speed command does not use low-level API calls anymore.
1287
1288   *Tomáš Mráz*
1289
1290 * Parallel dual-prime 1024-bit modular exponentiation for AVX512_IFMA
1291   capable processors.
1292
1293   *Ilya Albrekht, Sergey Kirillov, Andrey Matyukov (Intel Corp)*
1294
1295 * Combining the Configure options no-ec and no-dh no longer disables TLSv1.3.
1296
1297   *Matt Caswell*
1298
1299 * Implemented support for fully "pluggable" TLSv1.3 groups. This means that
1300   providers may supply their own group implementations (using either the "key
1301   exchange" or the "key encapsulation" methods) which will automatically be
1302   detected and used by libssl.
1303
1304   *Matt Caswell, Nicola Tuveri*
1305
1306 * The undocumented function X509_certificate_type() has been deprecated;
1307
1308   *Rich Salz*
1309
1310 * Deprecated the obsolete BN_pseudo_rand() and BN_pseudo_rand_range().
1311
1312   *Tomáš Mráz*
1313
1314 * Removed RSA padding mode for SSLv23 (which was only used for
1315   SSLv2). This includes the functions RSA_padding_check_SSLv23() and
1316   RSA_padding_add_SSLv23() and the `-ssl` option in the deprecated
1317   `rsautl` command.
1318
1319   *Rich Salz*
1320
1321 * Deprecated the obsolete X9.31 RSA key generation related functions.
1322
1323 * While a callback function set via `SSL_CTX_set_cert_verify_callback()`
1324   is not allowed to return a value > 1, this is no more taken as failure.
1325
1326   *Viktor Dukhovni and David von Oheimb*
1327
1328 * Deprecated the obsolete X9.31 RSA key generation related functions
1329   BN_X931_generate_Xpq(), BN_X931_derive_prime_ex(), and
1330   BN_X931_generate_prime_ex().
1331
1332   *Tomáš Mráz*
1333
1334 * The default key generation method for the regular 2-prime RSA keys was
1335   changed to the FIPS 186-4 B.3.6 method.
1336
1337   *Shane Lontis*
1338
1339 * Deprecated the BN_is_prime_ex() and BN_is_prime_fasttest_ex() functions.
1340
1341   *Kurt Roeckx*
1342
1343 * Deprecated EVP_MD_CTX_set_update_fn() and EVP_MD_CTX_update_fn().
1344
1345   *Rich Salz*
1346
1347 * Deprecated the type OCSP_REQ_CTX and the functions OCSP_REQ_CTX_*() and
1348   replaced with OSSL_HTTP_REQ_CTX and the functions OSSL_HTTP_REQ_CTX_*().
1349
1350   *Rich Salz, Richard Levitte, and David von Oheimb*
1351
1352 * Deprecated `X509_http_nbio()` and `X509_CRL_http_nbio()`.
1353
1354   *David von Oheimb*
1355
1356 * Deprecated `OCSP_parse_url()`.
1357
1358   *David von Oheimb*
1359
1360 * Validation of SM2 keys has been separated from the validation of regular EC
1361   keys.
1362
1363   *Nicola Tuveri*
1364
1365 * Behavior of the `pkey` app is changed, when using the `-check` or `-pubcheck`
1366   switches: a validation failure triggers an early exit, returning a failure
1367   exit status to the parent process.
1368
1369   *Nicola Tuveri*
1370
1371 * Changed behavior of SSL_CTX_set_ciphersuites() and SSL_set_ciphersuites()
1372   to ignore unknown ciphers.
1373
1374   *Otto Hollmann*
1375
1376 * The `-cipher-commands` and `-digest-commands` options
1377   of the command line utility `list` have been deprecated.
1378   Instead use the `-cipher-algorithms` and `-digest-algorithms` options.
1379
1380   *Dmitry Belyavskiy*
1381
1382 * Added convenience functions for generating asymmetric key pairs:
1383   The 'quick' one-shot (yet somewhat limited) function L<EVP_PKEY_Q_keygen(3)>
1384   and macros for the most common cases: <EVP_RSA_gen(3)> and L<EVP_EC_gen(3)>.
1385
1386   *David von Oheimb*
1387
1388 * All of the low level EC_KEY functions have been deprecated.
1389
1390   *Shane Lontis, Paul Dale, Richard Levitte, and Tomáš Mráz*
1391
1392 * Deprecated all the libcrypto and libssl error string loading
1393   functions.
1394
1395   *Richard Levitte*
1396
1397 * The functions SSL_CTX_set_tmp_dh_callback and SSL_set_tmp_dh_callback, as
1398   well as the macros SSL_CTX_set_tmp_dh() and SSL_set_tmp_dh() have been
1399   deprecated.
1400
1401   *Matt Caswell*
1402
1403 * The `-crypt` option to the `passwd` command line tool has been removed.
1404
1405   *Paul Dale*
1406
1407 * The -C option to the `x509`, `dhparam`, `dsaparam`, and `ecparam` commands
1408   were removed.
1409
1410   *Rich Salz*
1411
1412 * Add support for AES Key Wrap inverse ciphers to the EVP layer.
1413
1414   *Shane Lontis*
1415
1416 * Deprecated EVP_PKEY_set1_tls_encodedpoint() and
1417   EVP_PKEY_get1_tls_encodedpoint().
1418
1419   *Matt Caswell*
1420
1421 * The security callback, which can be customised by application code, supports
1422   the security operation SSL_SECOP_TMP_DH. One location of the "other" parameter
1423   was incorrectly passing a DH object. It now passed an EVP_PKEY in all cases.
1424
1425   *Matt Caswell*
1426
1427 * Add PKCS7_get_octet_string() and PKCS7_type_is_other() to the public
1428   interface. Their functionality remains unchanged.
1429
1430   *Jordan Montgomery*
1431
1432 * Added new option for 'openssl list', '-providers', which will display the
1433   list of loaded providers, their names, version and status.  It optionally
1434   displays their gettable parameters.
1435
1436   *Paul Dale*
1437
1438 * Removed EVP_PKEY_set_alias_type().
1439
1440   *Richard Levitte*
1441
1442 * Deprecated `EVP_PKEY_CTX_set_rsa_keygen_pubexp()` and introduced
1443   `EVP_PKEY_CTX_set1_rsa_keygen_pubexp()`, which is now preferred.
1444
1445   *Jeremy Walch*
1446
1447 * Changed all "STACK" functions to be macros instead of inline functions. Macro
1448   parameters are still checked for type safety at compile time via helper
1449   inline functions.
1450
1451   *Matt Caswell*
1452
1453 * Remove the RAND_DRBG API
1454
1455   *Paul Dale and Matthias St. Pierre*
1456
1457 * Allow `SSL_set1_host()` and `SSL_add1_host()` to take IP literal addresses
1458   as well as actual hostnames.
1459
1460   *David Woodhouse*
1461
1462 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
1463   ignore TLS protocol version bounds when configuring DTLS-based contexts, and
1464   conversely, silently ignore DTLS protocol version bounds when configuring
1465   TLS-based contexts.  The commands can be repeated to set bounds of both
1466   types.  The same applies with the corresponding "min_protocol" and
1467   "max_protocol" command-line switches, in case some application uses both TLS
1468   and DTLS.
1469
1470   SSL_CTX instances that are created for a fixed protocol version (e.g.
1471   `TLSv1_server_method()`) also silently ignore version bounds.  Previously
1472   attempts to apply bounds to these protocol versions would result in an
1473   error.  Now only the "version-flexible" SSL_CTX instances are subject to
1474   limits in configuration files in command-line options.
1475
1476   *Viktor Dukhovni*
1477
1478 * Deprecated the `ENGINE` API.  Engines should be replaced with providers
1479   going forward.
1480
1481   *Paul Dale*
1482
1483 * Reworked the recorded ERR codes to make better space for system errors.
1484   To distinguish them, the macro `ERR_SYSTEM_ERROR()` indicates if the
1485   given code is a system error (true) or an OpenSSL error (false).
1486
1487   *Richard Levitte*
1488
1489 * Reworked the test perl framework to better allow parallel testing.
1490
1491   *Nicola Tuveri and David von Oheimb*
1492
1493 * Added ciphertext stealing algorithms AES-128-CBC-CTS, AES-192-CBC-CTS and
1494   AES-256-CBC-CTS to the providers. CS1, CS2 and CS3 variants are supported.
1495
1496   *Shane Lontis*
1497
1498 * 'Configure' has been changed to figure out the configuration target if
1499   none is given on the command line.  Consequently, the 'config' script is
1500   now only a mere wrapper.  All documentation is changed to only mention
1501   'Configure'.
1502
1503   *Rich Salz and Richard Levitte*
1504
1505 * Added a library context `OSSL_LIB_CTX` that applications as well as
1506   other libraries can use to form a separate context within which
1507   libcrypto operations are performed.
1508
1509   *Richard Levitte*
1510
1511 * Added various `_ex` functions to the OpenSSL API that support using
1512   a non-default `OSSL_LIB_CTX`.
1513
1514   *OpenSSL team*
1515
1516 * Handshake now fails if Extended Master Secret extension is dropped
1517   on renegotiation.
1518
1519   *Tomáš Mráz*
1520
1521 * Dropped interactive mode from the `openssl` program.
1522
1523   *Richard Levitte*
1524
1525 * Deprecated `EVP_PKEY_cmp()` and `EVP_PKEY_cmp_parameters()`.
1526
1527   *David von Oheimb and Shane Lontis*
1528
1529 * Deprecated `EC_METHOD_get_field_type()`.
1530
1531   *Billy Bob Brumley*
1532
1533 * Deprecated EC_GFp_simple_method(), EC_GFp_mont_method(),
1534   EC_GF2m_simple_method(), EC_GFp_nist_method(), EC_GFp_nistp224_method()
1535   EC_GFp_nistp256_method(), and EC_GFp_nistp521_method().
1536
1537   *Billy Bob Brumley*
1538
1539 * Deprecated EC_GROUP_new(), EC_GROUP_method_of(), and EC_POINT_method_of().
1540
1541   *Billy Bob Brumley*
1542
1543 * Add CAdES-BES signature verification support, mostly derived
1544   from ESSCertIDv2 TS (RFC 5816) contribution by Marek Klein.
1545
1546   *Filipe Raimundo da Silva*
1547
1548 * Add CAdES-BES signature scheme and attributes support (RFC 5126) to CMS API.
1549
1550   *Antonio Iacono*
1551
1552 * Added the AuthEnvelopedData content type structure (RFC 5083) with AES-GCM
1553   parameter (RFC 5084) for the Cryptographic Message Syntax (CMS).
1554
1555   *Jakub Zelenka*
1556
1557 * Deprecated EC_POINT_make_affine() and EC_POINTs_make_affine().
1558
1559   *Billy Bob Brumley*
1560
1561 * Deprecated EC_GROUP_precompute_mult(), EC_GROUP_have_precompute_mult(), and
1562   EC_KEY_precompute_mult().
1563
1564   *Billy Bob Brumley*
1565
1566 * Deprecated EC_POINTs_mul().
1567
1568   *Billy Bob Brumley*
1569
1570 * Removed FIPS_mode() and FIPS_mode_set().
1571
1572   *Shane Lontis*
1573
1574 * The SSL option SSL_OP_IGNORE_UNEXPECTED_EOF is introduced.
1575
1576   *Dmitry Belyavskiy*
1577
1578 * Deprecated EC_POINT_set_Jprojective_coordinates_GFp() and
1579   EC_POINT_get_Jprojective_coordinates_GFp().
1580
1581   *Billy Bob Brumley*
1582
1583 * Added OSSL_PARAM_BLD to the public interface.  This allows OSSL_PARAM
1584   arrays to be more easily constructed via a series of utility functions.
1585   Create a parameter builder using OSSL_PARAM_BLD_new(), add parameters using
1586   the various push functions and finally convert to a passable OSSL_PARAM
1587   array using OSSL_PARAM_BLD_to_param().
1588
1589   *Paul Dale*
1590
1591 * The security strength of SHA1 and MD5 based signatures in TLS has been
1592   reduced.
1593
1594   *Kurt Roeckx*
1595
1596 * Added EVP_PKEY_set_type_by_keymgmt(), to initialise an EVP_PKEY to
1597   contain a provider side internal key.
1598
1599   *Richard Levitte*
1600
1601 * ASN1_verify(), ASN1_digest() and ASN1_sign() have been deprecated.
1602
1603   *Richard Levitte*
1604
1605 * Project text documents not yet having a proper file name extension
1606   (`HACKING`, `LICENSE`, `NOTES*`, `README*`, `VERSION`) have been renamed to
1607   `*.md` as far as reasonable, else `*.txt`, for better use with file managers.
1608
1609   *David von Oheimb*
1610
1611 * The main project documents (README, NEWS, CHANGES, INSTALL, SUPPORT)
1612   have been converted to Markdown with the goal to produce documents
1613   which not only look pretty when viewed online in the browser, but
1614   remain well readable inside a plain text editor.
1615
1616   To achieve this goal, a 'minimalistic' Markdown style has been applied
1617   which avoids formatting elements that interfere too much with the
1618   reading flow in the text file. For example, it
1619
1620   * avoids [ATX headings][] and uses [setext headings][] instead
1621     (which works for `<h1>` and `<h2>` headings only).
1622   * avoids [inline links][] and uses [reference links][] instead.
1623   * avoids [fenced code blocks][] and uses [indented code blocks][] instead.
1624
1625     [ATX headings]:         https://github.github.com/gfm/#atx-headings
1626     [setext headings]:      https://github.github.com/gfm/#setext-headings
1627     [inline links]:         https://github.github.com/gfm/#inline-link
1628     [reference links]:      https://github.github.com/gfm/#reference-link
1629     [fenced code blocks]:   https://github.github.com/gfm/#fenced-code-blocks
1630     [indented code blocks]: https://github.github.com/gfm/#indented-code-blocks
1631
1632   *Matthias St. Pierre*
1633
1634 * The test suite is changed to preserve results of each test recipe.
1635   A new directory test-runs/ with subdirectories named like the
1636   test recipes are created in the build tree for this purpose.
1637
1638   *Richard Levitte*
1639
1640 * Added an implementation of CMP and CRMF (RFC 4210, RFC 4211 RFC 6712).
1641   This adds `crypto/cmp/`, `crpyto/crmf/`, `apps/cmp.c`, and `test/cmp_*`.
1642   See L<openssl-cmp(1)> and L<OSSL_CMP_exec_IR_ses(3)> as starting points.
1643
1644   *David von Oheimb, Martin Peylo*
1645
1646 * Generalized the HTTP client code from `crypto/ocsp/` into `crpyto/http/`.
1647   It supports arbitrary request and response content types, GET redirection,
1648   TLS, connections via HTTP(S) proxies, connections and exchange via
1649   user-defined BIOs (allowing implicit connections), persistent connections,
1650   and timeout checks.  See L<OSSL_HTTP_transfer(3)> etc. for details.
1651   The legacy OCSP-focused (and only partly documented) API
1652   is retained for backward compatibility, while most of it is deprecated.
1653
1654   *David von Oheimb*
1655
1656 * Added `util/check-format.pl`, a tool for checking adherence to the
1657   OpenSSL coding style <https://www.openssl.org/policies/codingstyle.html>.
1658   The checks performed are incomplete and yield some false positives.
1659   Still the tool should be useful for detecting most typical glitches.
1660
1661   *David von Oheimb*
1662
1663 * `BIO_do_connect()` and `BIO_do_handshake()` have been extended:
1664   If domain name resolution yields multiple IP addresses all of them are tried
1665   after `connect()` failures.
1666
1667   *David von Oheimb*
1668
1669 * All of the low level RSA functions have been deprecated.
1670
1671   *Paul Dale*
1672
1673 * X509 certificates signed using SHA1 are no longer allowed at security
1674   level 1 and above.
1675
1676   *Kurt Roeckx*
1677
1678 * The command line utilities dhparam, dsa, gendsa and dsaparam have been
1679   modified to use PKEY APIs.  These commands are now in maintenance mode
1680   and no new features will be added to them.
1681
1682   *Paul Dale*
1683
1684 * The command line utility rsautl has been deprecated.
1685
1686   *Paul Dale*
1687
1688 * The command line utilities genrsa and rsa have been modified to use PKEY
1689   APIs. They now write PKCS#8 keys by default. These commands are now in
1690   maintenance mode and no new features will be added to them.
1691
1692   *Paul Dale*
1693
1694 * All of the low level DH functions have been deprecated.
1695
1696   *Paul Dale and Matt Caswell*
1697
1698 * All of the low level DSA functions have been deprecated.
1699
1700   *Paul Dale*
1701
1702 * Reworked the treatment of EC EVP_PKEYs with the SM2 curve to
1703   automatically become EVP_PKEY_SM2 rather than EVP_PKEY_EC.
1704
1705   *Richard Levitte*
1706
1707 * Deprecated low level ECDH and ECDSA functions.
1708
1709   *Paul Dale*
1710
1711 * Deprecated EVP_PKEY_decrypt_old() and EVP_PKEY_encrypt_old().
1712
1713   *Richard Levitte*
1714
1715 * Enhanced the documentation of EVP_PKEY_get_size(), EVP_PKEY_get_bits()
1716   and EVP_PKEY_get_security_bits().  Especially EVP_PKEY_get_size() needed
1717   a new formulation to include all the things it can be used for,
1718   as well as words of caution.
1719
1720   *Richard Levitte*
1721
1722 * The SSL_CTX_set_tlsext_ticket_key_cb(3) function has been deprecated.
1723
1724   *Paul Dale*
1725
1726 * All of the low level HMAC functions have been deprecated.
1727
1728   *Paul Dale and David von Oheimb*
1729
1730 * Over two thousand fixes were made to the documentation, including:
1731   - Common options (such as -rand/-writerand, TLS version control, etc)
1732     were refactored and point to newly-enhanced descriptions in openssl.pod.
1733   - Added style conformance for all options (with help from Richard Levitte),
1734     documented all reported missing options, added a CI build to check
1735     that all options are documented and that no unimplemented options
1736     are documented.
1737   - Documented some internals, such as all use of environment variables.
1738   - Addressed all internal broken L<> references.
1739
1740   *Rich Salz*
1741
1742 * All of the low level CMAC functions have been deprecated.
1743
1744   *Paul Dale*
1745
1746 * The low-level MD2, MD4, MD5, MDC2, RIPEMD160 and Whirlpool digest
1747   functions have been deprecated.
1748
1749   *Paul Dale and David von Oheimb*
1750
1751 * Corrected the documentation of the return values from the `EVP_DigestSign*`
1752   set of functions.  The documentation mentioned negative values for some
1753   errors, but this was never the case, so the mention of negative values
1754   was removed.
1755
1756   Code that followed the documentation and thereby check with something
1757   like `EVP_DigestSignInit(...) <= 0` will continue to work undisturbed.
1758
1759   *Richard Levitte*
1760
1761 * All of the low level cipher functions have been deprecated.
1762
1763   *Matt Caswell and Paul Dale*
1764
1765 * Removed include/openssl/opensslconf.h.in and replaced it with
1766   include/openssl/configuration.h.in, which differs in not including
1767   <openssl/macros.h>.  A short header include/openssl/opensslconf.h
1768   was added to include both.
1769
1770   This allows internal hacks where one might need to modify the set
1771   of configured macros, for example this if deprecated symbols are
1772   still supposed to be available internally:
1773
1774       #include <openssl/configuration.h>
1775
1776       #undef OPENSSL_NO_DEPRECATED
1777       #define OPENSSL_SUPPRESS_DEPRECATED
1778
1779       #include <openssl/macros.h>
1780
1781   This should not be used by applications that use the exported
1782   symbols, as that will lead to linking errors.
1783
1784   *Richard Levitte*
1785
1786 * Fixed an overflow bug in the x64_64 Montgomery squaring procedure
1787   used in exponentiation with 512-bit moduli. No EC algorithms are
1788   affected. Analysis suggests that attacks against 2-prime RSA1024,
1789   3-prime RSA1536, and DSA1024 as a result of this defect would be very
1790   difficult to perform and are not believed likely. Attacks against DH512
1791   are considered just feasible. However, for an attack the target would
1792   have to re-use the DH512 private key, which is not recommended anyway.
1793   Also applications directly using the low-level API BN_mod_exp may be
1794   affected if they use BN_FLG_CONSTTIME.
1795   ([CVE-2019-1551])
1796
1797   *Andy Polyakov*
1798
1799 * Most memory-debug features have been deprecated, and the functionality
1800   replaced with no-ops.
1801
1802   *Rich Salz*
1803
1804 * Added documentation for the STACK API.
1805
1806   *Rich Salz*
1807
1808 * Introduced a new method type and API, OSSL_ENCODER, to represent
1809   generic encoders.  These do the same sort of job that PEM writers
1810   and d2i functions do, but with support for methods supplied by
1811   providers, and the possibility for providers to support other
1812   formats as well.
1813
1814   *Richard Levitte*
1815
1816 * Introduced a new method type and API, OSSL_DECODER, to represent
1817   generic decoders.  These do the same sort of job that PEM readers
1818   and i2d functions do, but with support for methods supplied by
1819   providers, and the possibility for providers to support other
1820   formats as well.
1821
1822   *Richard Levitte*
1823
1824 * Added a .pragma directive to the syntax of configuration files, to
1825   allow varying behavior in a supported and predictable manner.
1826   Currently added pragma:
1827
1828           .pragma dollarid:on
1829
1830   This allows dollar signs to be a keyword character unless it's
1831   followed by a opening brace or parenthesis.  This is useful for
1832   platforms where dollar signs are commonly used in names, such as
1833   volume names and system directory names on VMS.
1834
1835   *Richard Levitte*
1836
1837 * Added functionality to create an EVP_PKEY from user data.
1838
1839   *Richard Levitte*
1840
1841 * Change the interpretation of the '--api' configuration option to
1842   mean that this is a desired API compatibility level with no
1843   further meaning.  The previous interpretation, that this would
1844   also mean to remove all deprecated symbols up to and including
1845   the given version, no requires that 'no-deprecated' is also used
1846   in the configuration.
1847
1848   When building applications, the desired API compatibility level
1849   can be set with the OPENSSL_API_COMPAT macro like before.  For
1850   API compatibility version below 3.0, the old style numerical
1851   value is valid as before, such as -DOPENSSL_API_COMPAT=0x10100000L.
1852   For version 3.0 and on, the value is expected to be the decimal
1853   value calculated from the major and minor version like this:
1854
1855           MAJOR * 10000 + MINOR * 100
1856
1857   Examples:
1858
1859           -DOPENSSL_API_COMPAT=30000             For 3.0
1860           -DOPENSSL_API_COMPAT=30200             For 3.2
1861
1862   To hide declarations that are deprecated up to and including the
1863   given API compatibility level, -DOPENSSL_NO_DEPRECATED must be
1864   given when building the application as well.
1865
1866   *Richard Levitte*
1867
1868 * Added the X509_LOOKUP_METHOD called X509_LOOKUP_store, to allow
1869   access to certificate and CRL stores via URIs and OSSL_STORE
1870   loaders.
1871
1872   This adds the following functions:
1873
1874   - X509_LOOKUP_store()
1875   - X509_STORE_load_file()
1876   - X509_STORE_load_path()
1877   - X509_STORE_load_store()
1878   - SSL_add_store_cert_subjects_to_stack()
1879   - SSL_CTX_set_default_verify_store()
1880   - SSL_CTX_load_verify_file()
1881   - SSL_CTX_load_verify_dir()
1882   - SSL_CTX_load_verify_store()
1883
1884   *Richard Levitte*
1885
1886 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY.
1887   The presence of this system service is determined at run-time.
1888
1889   *Richard Levitte*
1890
1891 * Added functionality to create an EVP_PKEY context based on data
1892   for methods from providers.  This takes an algorithm name and a
1893   property query string and simply stores them, with the intent
1894   that any operation that uses this context will use those strings
1895   to fetch the needed methods implicitly, thereby making the port
1896   of application written for pre-3.0 OpenSSL easier.
1897
1898   *Richard Levitte*
1899
1900 * The undocumented function NCONF_WIN32() has been deprecated; for
1901   conversion details see the HISTORY section of doc/man5/config.pod
1902
1903   *Rich Salz*
1904
1905 * Introduced the new functions EVP_DigestSignInit_ex() and
1906   EVP_DigestVerifyInit_ex(). The macros EVP_DigestSignUpdate() and
1907   EVP_DigestVerifyUpdate() have been converted to functions. See the man
1908   pages for further details.
1909
1910   *Matt Caswell*
1911
1912 * Over two thousand fixes were made to the documentation, including:
1913   adding missing command flags, better style conformance, documentation
1914   of internals, etc.
1915
1916   *Rich Salz, Richard Levitte*
1917
1918 * s390x assembly pack: add hardware-support for P-256, P-384, P-521,
1919   X25519, X448, Ed25519 and Ed448.
1920
1921   *Patrick Steuer*
1922
1923 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just
1924   the first value.
1925
1926   *Jon Spillett*
1927
1928 * Deprecated the public definition of `ERR_STATE` as well as the function
1929   `ERR_get_state()`.  This is done in preparation of making `ERR_STATE` an
1930   opaque type.
1931
1932   *Richard Levitte*
1933
1934 * Added ERR functionality to give callers access to the stored function
1935   names that have replaced the older function code based functions.
1936
1937   New functions are ERR_peek_error_func(), ERR_peek_last_error_func(),
1938   ERR_peek_error_data(), ERR_peek_last_error_data(), ERR_get_error_all(),
1939   ERR_peek_error_all() and ERR_peek_last_error_all().
1940
1941   Deprecate ERR functions ERR_get_error_line(), ERR_get_error_line_data(),
1942   ERR_peek_error_line_data(), ERR_peek_last_error_line_data() and
1943   ERR_func_error_string().
1944
1945   *Richard Levitte*
1946
1947 * Extended testing to be verbose for failing tests only.  The make variables
1948   VERBOSE_FAILURE or VF can be used to enable this:
1949
1950           $ make VF=1 test                           # Unix
1951           $ mms /macro=(VF=1) test                   ! OpenVMS
1952           $ nmake VF=1 test                          # Windows
1953
1954   *Richard Levitte*
1955
1956 * Added the `-copy_extensions` option to the `x509` command for use with
1957   `-req` and `-x509toreq`. When given with the `copy` or `copyall` argument,
1958   all extensions in the request are copied to the certificate or vice versa.
1959
1960   *David von Oheimb*, *Kirill Stefanenkov <kirill_stefanenkov@rambler.ru>*
1961
1962 * Added the `-copy_extensions` option to the `req` command for use with
1963   `-x509`. When given with the `copy` or `copyall` argument,
1964   all extensions in the certification request are copied to the certificate.
1965
1966   *David von Oheimb*
1967
1968 * The `x509`, `req`, and `ca` commands now make sure that X.509v3 certificates
1969   they generate are by default RFC 5280 compliant in the following sense:
1970   There is a subjectKeyIdentifier extension with a hash value of the public key
1971   and for not self-signed certs there is an authorityKeyIdentifier extension
1972   with a keyIdentifier field or issuer information identifying the signing key.
1973   This is done unless some configuration overrides the new default behavior,
1974   such as `subjectKeyIdentifier = none` and `authorityKeyIdentifier = none`.
1975
1976   *David von Oheimb*
1977
1978 * Added several checks to `X509_verify_cert()` according to requirements in
1979   RFC 5280 in case `X509_V_FLAG_X509_STRICT` is set
1980   (which may be done by using the CLI option `-x509_strict`):
1981   * The basicConstraints of CA certificates must be marked critical.
1982   * CA certificates must explicitly include the keyUsage extension.
1983   * If a pathlenConstraint is given the key usage keyCertSign must be allowed.
1984   * The issuer name of any certificate must not be empty.
1985   * The subject name of CA certs, certs with keyUsage crlSign,
1986     and certs without subjectAlternativeName must not be empty.
1987   * If a subjectAlternativeName extension is given it must not be empty.
1988   * The signatureAlgorithm field and the cert signature must be consistent.
1989   * Any given authorityKeyIdentifier and any given subjectKeyIdentifier
1990     must not be marked critical.
1991   * The authorityKeyIdentifier must be given for X.509v3 certs
1992     unless they are self-signed.
1993   * The subjectKeyIdentifier must be given for all X.509v3 CA certs.
1994
1995   *David von Oheimb*
1996
1997 * Certificate verification using `X509_verify_cert()` meanwhile rejects EC keys
1998   with explicit curve parameters (specifiedCurve) as required by RFC 5480.
1999
2000   *Tomáš Mráz*
2001
2002 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
2003   used even when parsing explicit parameters, when loading a encoded key
2004   or calling `EC_GROUP_new_from_ecpkparameters()`/
2005   `EC_GROUP_new_from_ecparameters()`.
2006   This prevents bypass of security hardening and performance gains,
2007   especially for curves with specialized EC_METHODs.
2008   By default, if a key encoded with explicit parameters is loaded and later
2009   encoded, the output is still encoded with explicit parameters, even if
2010   internally a "named" EC_GROUP is used for computation.
2011
2012   *Nicola Tuveri*
2013
2014 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
2015   this change, EC_GROUP_set_generator would accept order and/or cofactor as
2016   NULL. After this change, only the cofactor parameter can be NULL. It also
2017   does some minimal sanity checks on the passed order.
2018   ([CVE-2019-1547])
2019
2020   *Billy Bob Brumley*
2021
2022 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
2023   An attack is simple, if the first CMS_recipientInfo is valid but the
2024   second CMS_recipientInfo is chosen ciphertext. If the second
2025   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
2026   encryption key will be replaced by garbage, and the message cannot be
2027   decoded, but if the RSA decryption fails, the correct encryption key is
2028   used and the recipient will not notice the attack.
2029   As a work around for this potential attack the length of the decrypted
2030   key must be equal to the cipher default key length, in case the
2031   certifiate is not given and all recipientInfo are tried out.
2032   The old behaviour can be re-enabled in the CMS code by setting the
2033   CMS_DEBUG_DECRYPT flag.
2034
2035   *Bernd Edlinger*
2036
2037 * Early start up entropy quality from the DEVRANDOM seed source has been
2038   improved for older Linux systems.  The RAND subsystem will wait for
2039   /dev/random to be producing output before seeding from /dev/urandom.
2040   The seeded state is stored for future library initialisations using
2041   a system global shared memory segment.  The shared memory identifier
2042   can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to
2043   the desired value.  The default identifier is 114.
2044
2045   *Paul Dale*
2046
2047 * Revised BN_generate_prime_ex to not avoid factors 2..17863 in p-1
2048   when primes for RSA keys are computed.
2049   Since we previously always generated primes == 2 (mod 3) for RSA keys,
2050   the 2-prime and 3-prime RSA modules were easy to distinguish, since
2051   `N = p*q = 1 (mod 3)`, but `N = p*q*r = 2 (mod 3)`. Therefore fingerprinting
2052   2-prime vs. 3-prime RSA keys was possible by computing N mod 3.
2053   This avoids possible fingerprinting of newly generated RSA modules.
2054
2055   *Bernd Edlinger*
2056
2057 * Correct the extended master secret constant on EBCDIC systems. Without this
2058   fix TLS connections between an EBCDIC system and a non-EBCDIC system that
2059   negotiate EMS will fail. Unfortunately this also means that TLS connections
2060   between EBCDIC systems with this fix, and EBCDIC systems without this
2061   fix will fail if they negotiate EMS.
2062
2063   *Matt Caswell*
2064
2065 * Changed the library initialisation so that the config file is now loaded
2066   by default. This was already the case for libssl. It now occurs for both
2067   libcrypto and libssl. Use the OPENSSL_INIT_NO_LOAD_CONFIG option to
2068   `OPENSSL_init_crypto()` to suppress automatic loading of a config file.
2069
2070   *Matt Caswell*
2071
2072 * Introduced new error raising macros, `ERR_raise()` and `ERR_raise_data()`,
2073   where the former acts as a replacement for `ERR_put_error()`, and the
2074   latter replaces the combination `ERR_put_error()` + `ERR_add_error_data()`.
2075   `ERR_raise_data()` adds more flexibility by taking a format string and
2076   an arbitrary number of arguments following it, to be processed with
2077   `BIO_snprintf()`.
2078
2079   *Richard Levitte*
2080
2081 * Introduced a new function, `OSSL_PROVIDER_available()`, which can be used
2082   to check if a named provider is loaded and available.  When called, it
2083   will also activate all fallback providers if such are still present.
2084
2085   *Richard Levitte*
2086
2087 * Enforce a minimum DH modulus size of 512 bits.
2088
2089   *Bernd Edlinger*
2090
2091 * Changed DH parameters to generate the order q subgroup instead of 2q.
2092   Previously generated DH parameters are still accepted by DH_check
2093   but DH_generate_key works around that by clearing bit 0 of the
2094   private key for those. This avoids leaking bit 0 of the private key.
2095
2096   *Bernd Edlinger*
2097
2098 * Significantly reduce secure memory usage by the randomness pools.
2099
2100   *Paul Dale*
2101
2102 * `{CRYPTO,OPENSSL}_mem_debug_{push,pop}` are now no-ops and have been
2103   deprecated.
2104
2105   *Rich Salz*
2106
2107 * A new type, EVP_KEYEXCH, has been introduced to represent key exchange
2108   algorithms. An implementation of a key exchange algorithm can be obtained
2109   by using the function EVP_KEYEXCH_fetch(). An EVP_KEYEXCH algorithm can be
2110   used in a call to EVP_PKEY_derive_init_ex() which works in a similar way to
2111   the older EVP_PKEY_derive_init() function. See the man pages for the new
2112   functions for further details.
2113
2114   *Matt Caswell*
2115
2116 * The EVP_PKEY_CTX_set_dh_pad() macro has now been converted to a function.
2117
2118   *Matt Caswell*
2119
2120 * Removed the function names from error messages and deprecated the
2121   xxx_F_xxx define's.
2122
2123   *Richard Levitte*
2124
2125 * Removed NextStep support and the macro OPENSSL_UNISTD
2126
2127   *Rich Salz*
2128
2129 * Removed DES_check_key.  Also removed OPENSSL_IMPLEMENT_GLOBAL,
2130   OPENSSL_GLOBAL_REF, OPENSSL_DECLARE_GLOBAL.
2131   Also removed "export var as function" capability; we do not export
2132   variables, only functions.
2133
2134   *Rich Salz*
2135
2136 * RC5_32_set_key has been changed to return an int type, with 0 indicating
2137   an error and 1 indicating success. In previous versions of OpenSSL this
2138   was a void type. If a key was set longer than the maximum possible this
2139   would crash.
2140
2141   *Matt Caswell*
2142
2143 * Support SM2 signing and verification schemes with X509 certificate.
2144
2145   *Paul Yang*
2146
2147 * Use SHA256 as the default digest for TS query in the `ts` app.
2148
2149   *Tomáš Mráz*
2150
2151 * Change PBKDF2 to conform to SP800-132 instead of the older PKCS5 RFC2898.
2152
2153   *Shane Lontis*
2154
2155 * Default cipher lists/suites are now available via a function, the
2156   #defines are deprecated.
2157
2158   *Todd Short*
2159
2160 * Add target VC-WIN32-UWP, VC-WIN64A-UWP, VC-WIN32-ARM-UWP and
2161   VC-WIN64-ARM-UWP in Windows OneCore target for making building libraries
2162   for Windows Store apps easier. Also, the "no-uplink" option has been added.
2163
2164   *Kenji Mouri*
2165
2166 * Join the directories crypto/x509 and crypto/x509v3
2167
2168   *Richard Levitte*
2169
2170 * Added command 'openssl kdf' that uses the EVP_KDF API.
2171
2172   *Shane Lontis*
2173
2174 * Added command 'openssl mac' that uses the EVP_MAC API.
2175
2176   *Shane Lontis*
2177
2178 * Added OPENSSL_info() to get diverse built-in OpenSSL data, such
2179   as default directories.  Also added the command 'openssl info'
2180   for scripting purposes.
2181
2182   *Richard Levitte*
2183
2184 * The functions AES_ige_encrypt() and AES_bi_ige_encrypt() have been
2185   deprecated.
2186
2187   *Matt Caswell*
2188
2189 * Add prediction resistance to the DRBG reseeding process.
2190
2191   *Paul Dale*
2192
2193 * Limit the number of blocks in a data unit for AES-XTS to 2^20 as
2194   mandated by IEEE Std 1619-2018.
2195
2196   *Paul Dale*
2197
2198 * Added newline escaping functionality to a filename when using openssl dgst.
2199   This output format is to replicate the output format found in the `*sum`
2200   checksum programs. This aims to preserve backward compatibility.
2201
2202   *Matt Eaton, Richard Levitte, and Paul Dale*
2203
2204 * Removed the heartbeat message in DTLS feature, as it has very
2205   little usage and doesn't seem to fulfill a valuable purpose.
2206   The configuration option is now deprecated.
2207
2208   *Richard Levitte*
2209
2210 * Changed the output of 'openssl {digestname} < file' to display the
2211   digest name in its output.
2212
2213   *Richard Levitte*
2214
2215 * Added a new generic trace API which provides support for enabling
2216   instrumentation through trace output.
2217
2218   *Richard Levitte & Matthias St. Pierre*
2219
2220 * Added build tests for C++.  These are generated files that only do one
2221   thing, to include one public OpenSSL head file each.  This tests that
2222   the public header files can be usefully included in a C++ application.
2223
2224   This test isn't enabled by default.  It can be enabled with the option
2225   'enable-buildtest-c++'.
2226
2227   *Richard Levitte*
2228
2229 * Added KB KDF (EVP_KDF_KB) to EVP_KDF.
2230
2231   *Robbie Harwood*
2232
2233 * Added SSH KDF (EVP_KDF_SSHKDF) and KRB5 KDF (EVP_KDF_KRB5KDF) to EVP_KDF.
2234
2235   *Simo Sorce*
2236
2237 * Added Single Step KDF (EVP_KDF_SS), X963 KDF, and X942 KDF to EVP_KDF.
2238
2239   *Shane Lontis*
2240
2241 * Added KMAC to EVP_MAC.
2242
2243   *Shane Lontis*
2244
2245 * Added property based algorithm implementation selection framework to
2246   the core.
2247
2248   *Paul Dale*
2249
2250 * Added SCA hardening for modular field inversion in EC_GROUP through
2251   a new dedicated field_inv() pointer in EC_METHOD.
2252   This also addresses a leakage affecting conversions from projective
2253   to affine coordinates.
2254
2255   *Billy Bob Brumley, Nicola Tuveri*
2256
2257 * Added EVP_KDF, an EVP layer KDF API, to simplify adding KDF and PRF
2258   implementations.  This includes an EVP_PKEY to EVP_KDF bridge for
2259   those algorithms that were already supported through the EVP_PKEY API
2260   (scrypt, TLS1 PRF and HKDF).  The low-level KDF functions for PBKDF2
2261   and scrypt are now wrappers that call EVP_KDF.
2262
2263   *David Makepeace*
2264
2265 * Build devcrypto engine as a dynamic engine.
2266
2267   *Eneas U de Queiroz*
2268
2269 * Add keyed BLAKE2 to EVP_MAC.
2270
2271   *Antoine Salon*
2272
2273 * Fix a bug in the computation of the endpoint-pair shared secret used
2274   by DTLS over SCTP. This breaks interoperability with older versions
2275   of OpenSSL like OpenSSL 1.1.0 and OpenSSL 1.0.2. There is a runtime
2276   switch SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG (off by default) enabling
2277   interoperability with such broken implementations. However, enabling
2278   this switch breaks interoperability with correct implementations.
2279
2280 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
2281   re-used X509_PUBKEY object if the second PUBKEY is malformed.
2282
2283   *Bernd Edlinger*
2284
2285 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
2286
2287   *Richard Levitte*
2288
2289 * Changed the license to the Apache License v2.0.
2290
2291   *Richard Levitte*
2292
2293 * Switch to a new version scheme using three numbers MAJOR.MINOR.PATCH.
2294
2295   - Major releases (indicated by incrementing the MAJOR release number)
2296     may introduce incompatible API/ABI changes.
2297   - Minor releases (indicated by incrementing the MINOR release number)
2298     may introduce new features but retain API/ABI compatibility.
2299   - Patch releases (indicated by incrementing the PATCH number)
2300     are intended for bug fixes and other improvements of existing
2301     features only (like improving performance or adding documentation)
2302     and retain API/ABI compatibility.
2303
2304   *Richard Levitte*
2305
2306 * Add support for RFC5297 SIV mode (siv128), including AES-SIV.
2307
2308   *Todd Short*
2309
2310 * Remove the 'dist' target and add a tarball building script.  The
2311   'dist' target has fallen out of use, and it shouldn't be
2312   necessary to configure just to create a source distribution.
2313
2314   *Richard Levitte*
2315
2316 * Recreate the OS390-Unix config target.  It no longer relies on a
2317   special script like it did for OpenSSL pre-1.1.0.
2318
2319   *Richard Levitte*
2320
2321 * Instead of having the source directories listed in Configure, add
2322   a 'build.info' keyword SUBDIRS to indicate what sub-directories to
2323   look into.
2324
2325   *Richard Levitte*
2326
2327 * Add GMAC to EVP_MAC.
2328
2329   *Paul Dale*
2330
2331 * Ported the HMAC, CMAC and SipHash EVP_PKEY_METHODs to EVP_MAC.
2332
2333   *Richard Levitte*
2334
2335 * Added EVP_MAC, an EVP layer MAC API, to simplify adding MAC
2336   implementations.  This includes a generic EVP_PKEY to EVP_MAC bridge,
2337   to facilitate the continued use of MACs through raw private keys in
2338   functionality such as `EVP_DigestSign*` and `EVP_DigestVerify*`.
2339
2340   *Richard Levitte*
2341
2342 * Deprecate ECDH_KDF_X9_62().
2343
2344   *Antoine Salon*
2345
2346 * Added EVP_PKEY_ECDH_KDF_X9_63 and ecdh_KDF_X9_63() as replacements for
2347   the EVP_PKEY_ECDH_KDF_X9_62 KDF type and ECDH_KDF_X9_62(). The old names
2348   are retained for backwards compatibility.
2349
2350   *Antoine Salon*
2351
2352 * AES-XTS mode now enforces that its two keys are different to mitigate
2353   the attacked described in "Efficient Instantiations of Tweakable
2354   Blockciphers and Refinements to Modes OCB and PMAC" by Phillip Rogaway.
2355   Details of this attack can be obtained from:
2356   <http://web.cs.ucdavis.edu/%7Erogaway/papers/offsets.pdf>
2357
2358   *Paul Dale*
2359
2360 * Rename the object files, i.e. give them other names than in previous
2361   versions.  Their names now include the name of the final product, as
2362   well as its type mnemonic (bin, lib, shlib).
2363
2364   *Richard Levitte*
2365
2366 * Added new option for 'openssl list', '-objects', which will display the
2367   list of built in objects, i.e. OIDs with names.
2368
2369   *Richard Levitte*
2370
2371 * Added the options `-crl_lastupdate` and `-crl_nextupdate` to `openssl ca`,
2372   allowing the `lastUpdate` and `nextUpdate` fields in the generated CRL to
2373   be set explicitly.
2374
2375   *Chris Novakovic*
2376
2377 * Added support for Linux Kernel TLS data-path. The Linux Kernel data-path
2378   improves application performance by removing data copies and providing
2379   applications with zero-copy system calls such as sendfile and splice.
2380
2381   *Boris Pismenny*
2382
2383 * The SSL option SSL_OP_CLEANSE_PLAINTEXT is introduced.
2384
2385   *Martin Elshuber*
2386
2387 * `PKCS12_parse` now maintains the order of the parsed certificates
2388   when outputting them via `*ca` (rather than reversing it).
2389
2390   *David von Oheimb*
2391
2392 * Deprecated pthread fork support methods.
2393
2394   *Randall S. Becker*
2395
2396 * Added support for FFDHE key exchange in TLS 1.3.
2397
2398   *Raja Ashok*
2399
2400 * Added a new concept for OpenSSL plugability: providers.  This
2401   functionality is designed to replace the ENGINE API and ENGINE
2402   implementations, and to be much more dynamic, allowing provider
2403   authors to introduce new algorithms among other things, as long as
2404   there's an API that supports the algorithm type.
2405
2406   With this concept comes a new core API for interaction between
2407   libcrypto and provider implementations.  Public libcrypto functions
2408   that want to use providers do so through this core API.
2409
2410   The main documentation for this core API is found in
2411   doc/man7/provider.pod, doc/man7/provider-base.pod, and they in turn
2412   refer to other manuals describing the API specific for supported
2413   algorithm types (also called operations).
2414
2415   *The OpenSSL team*
2416
2417OpenSSL 1.1.1
2418-------------
2419
2420### Changes between 1.1.1l and 1.1.1m [xx XXX xxxx]
2421
2422 * Avoid loading of a dynamic engine twice.
2423
2424   *Bernd Edlinger*
2425
2426 * Prioritise DANE TLSA issuer certs over peer certs
2427
2428   *Viktor Dukhovni*
2429
2430 * Fixed random API for MacOS prior to 10.12
2431
2432   These MacOS versions don't support the CommonCrypto APIs
2433
2434   *Lenny Primak*
2435
2436### Changes between 1.1.1k and 1.1.1l [24 Aug 2021]
2437
2438 * Fixed an SM2 Decryption Buffer Overflow.
2439
2440   In order to decrypt SM2 encrypted data an application is expected to
2441   call the API function EVP_PKEY_decrypt(). Typically an application will
2442   call this function twice. The first time, on entry, the "out" parameter
2443   can be NULL and, on exit, the "outlen" parameter is populated with the
2444   buffer size required to hold the decrypted plaintext. The application
2445   can then allocate a sufficiently sized buffer and call EVP_PKEY_decrypt()
2446   again, but this time passing a non-NULL value for the "out" parameter.
2447
2448   A bug in the implementation of the SM2 decryption code means that the
2449   calculation of the buffer size required to hold the plaintext returned
2450   by the first call to EVP_PKEY_decrypt() can be smaller than the actual
2451   size required by the second call. This can lead to a buffer overflow
2452   when EVP_PKEY_decrypt() is called by the application a second time with
2453   a buffer that is too small.
2454
2455   A malicious attacker who is able present SM2 content for decryption to
2456   an application could cause attacker chosen data to overflow the buffer
2457   by up to a maximum of 62 bytes altering the contents of other data held
2458   after the buffer, possibly changing application behaviour or causing
2459   the application to crash. The location of the buffer is application
2460   dependent but is typically heap allocated.
2461   ([CVE-2021-3711])
2462
2463   *Matt Caswell*
2464
2465 * Fixed various read buffer overruns processing ASN.1 strings
2466
2467   ASN.1 strings are represented internally within OpenSSL as an ASN1_STRING
2468   structure which contains a buffer holding the string data and a field
2469   holding the buffer length. This contrasts with normal C strings which
2470   are repesented as a buffer for the string data which is terminated
2471   with a NUL (0) byte.
2472
2473   Although not a strict requirement, ASN.1 strings that are parsed using
2474   OpenSSL's own "d2i" functions (and other similar parsing functions) as
2475   well as any string whose value has been set with the ASN1_STRING_set()
2476   function will additionally NUL terminate the byte array in the
2477   ASN1_STRING structure.
2478
2479   However, it is possible for applications to directly construct valid
2480   ASN1_STRING structures which do not NUL terminate the byte array by
2481   directly setting the "data" and "length" fields in the ASN1_STRING
2482   array. This can also happen by using the ASN1_STRING_set0() function.
2483
2484   Numerous OpenSSL functions that print ASN.1 data have been found to
2485   assume that the ASN1_STRING byte array will be NUL terminated, even
2486   though this is not guaranteed for strings that have been directly
2487   constructed. Where an application requests an ASN.1 structure to be
2488   printed, and where that ASN.1 structure contains ASN1_STRINGs that have
2489   been directly constructed by the application without NUL terminating
2490   the "data" field, then a read buffer overrun can occur.
2491
2492   The same thing can also occur during name constraints processing
2493   of certificates (for example if a certificate has been directly
2494   constructed by the application instead of loading it via the OpenSSL
2495   parsing functions, and the certificate contains non NUL terminated
2496   ASN1_STRING structures). It can also occur in the X509_get1_email(),
2497   X509_REQ_get1_email() and X509_get1_ocsp() functions.
2498
2499   If a malicious actor can cause an application to directly construct an
2500   ASN1_STRING and then process it through one of the affected OpenSSL
2501   functions then this issue could be hit. This might result in a crash
2502   (causing a Denial of Service attack). It could also result in the
2503   disclosure of private memory contents (such as private keys, or
2504   sensitive plaintext).
2505   ([CVE-2021-3712])
2506
2507   *Matt Caswell*
2508
2509### Changes between 1.1.1j and 1.1.1k [25 Mar 2021]
2510
2511 * Fixed a problem with verifying a certificate chain when using the
2512   X509_V_FLAG_X509_STRICT flag. This flag enables additional security checks of
2513   the certificates present in a certificate chain. It is not set by default.
2514
2515   Starting from OpenSSL version 1.1.1h a check to disallow certificates in
2516   the chain that have explicitly encoded elliptic curve parameters was added
2517   as an additional strict check.
2518
2519   An error in the implementation of this check meant that the result of a
2520   previous check to confirm that certificates in the chain are valid CA
2521   certificates was overwritten. This effectively bypasses the check
2522   that non-CA certificates must not be able to issue other certificates.
2523
2524   If a "purpose" has been configured then there is a subsequent opportunity
2525   for checks that the certificate is a valid CA.  All of the named "purpose"
2526   values implemented in libcrypto perform this check.  Therefore, where
2527   a purpose is set the certificate chain will still be rejected even when the
2528   strict flag has been used. A purpose is set by default in libssl client and
2529   server certificate verification routines, but it can be overridden or
2530   removed by an application.
2531
2532   In order to be affected, an application must explicitly set the
2533   X509_V_FLAG_X509_STRICT verification flag and either not set a purpose
2534   for the certificate verification or, in the case of TLS client or server
2535   applications, override the default purpose.
2536   ([CVE-2021-3450])
2537
2538   *Tomáš Mráz*
2539
2540 * Fixed an issue where an OpenSSL TLS server may crash if sent a maliciously
2541   crafted renegotiation ClientHello message from a client. If a TLSv1.2
2542   renegotiation ClientHello omits the signature_algorithms extension (where it
2543   was present in the initial ClientHello), but includes a
2544   signature_algorithms_cert extension then a NULL pointer dereference will
2545   result, leading to a crash and a denial of service attack.
2546
2547   A server is only vulnerable if it has TLSv1.2 and renegotiation enabled
2548   (which is the default configuration). OpenSSL TLS clients are not impacted by
2549   this issue.
2550   ([CVE-2021-3449])
2551
2552   *Peter Kästle and Samuel Sapalski*
2553
2554### Changes between 1.1.1i and 1.1.1j [16 Feb 2021]
2555
2556 * Fixed the X509_issuer_and_serial_hash() function. It attempts to
2557   create a unique hash value based on the issuer and serial number data
2558   contained within an X509 certificate. However it was failing to correctly
2559   handle any errors that may occur while parsing the issuer field (which might
2560   occur if the issuer field is maliciously constructed). This may subsequently
2561   result in a NULL pointer deref and a crash leading to a potential denial of
2562   service attack.
2563   ([CVE-2021-23841])
2564
2565   *Matt Caswell*
2566
2567 * Fixed the RSA_padding_check_SSLv23() function and the RSA_SSLV23_PADDING
2568   padding mode to correctly check for rollback attacks. This is considered a
2569   bug in OpenSSL 1.1.1 because it does not support SSLv2. In 1.0.2 this is
2570   CVE-2021-23839.
2571
2572   *Matt Caswell*
2573
2574   Fixed the EVP_CipherUpdate, EVP_EncryptUpdate and EVP_DecryptUpdate
2575   functions. Previously they could overflow the output length argument in some
2576   cases where the input length is close to the maximum permissable length for
2577   an integer on the platform. In such cases the return value from the function
2578   call would be 1 (indicating success), but the output length value would be
2579   negative. This could cause applications to behave incorrectly or crash.
2580   ([CVE-2021-23840])
2581
2582   *Matt Caswell*
2583
2584 * Fixed SRP_Calc_client_key so that it runs in constant time. The previous
2585   implementation called BN_mod_exp without setting BN_FLG_CONSTTIME. This
2586   could be exploited in a side channel attack to recover the password. Since
2587   the attack is local host only this is outside of the current OpenSSL
2588   threat model and therefore no CVE is assigned.
2589
2590   Thanks to Mohammed Sabt and Daniel De Almeida Braga for reporting this
2591   issue.
2592
2593   *Matt Caswell*
2594
2595### Changes between 1.1.1h and 1.1.1i [8 Dec 2020]
2596
2597 * Fixed NULL pointer deref in the GENERAL_NAME_cmp function
2598   This function could crash if both GENERAL_NAMEs contain an EDIPARTYNAME.
2599    If an attacker can control both items being compared  then this could lead
2600    to a possible denial of service attack. OpenSSL itself uses the
2601    GENERAL_NAME_cmp function for two purposes:
2602    1) Comparing CRL distribution point names between an available CRL and a
2603       CRL distribution point embedded in an X509 certificate
2604    2) When verifying that a timestamp response token signer matches the
2605       timestamp authority name (exposed via the API functions
2606       TS_RESP_verify_response and TS_RESP_verify_token)
2607   ([CVE-2020-1971])
2608
2609   *Matt Caswell*
2610
2611### Changes between 1.1.1g and 1.1.1h [22 Sep 2020]
2612
2613 * Certificates with explicit curve parameters are now disallowed in
2614   verification chains if the X509_V_FLAG_X509_STRICT flag is used.
2615
2616   *Tomáš Mráz*
2617
2618 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
2619   ignore TLS protocol version bounds when configuring DTLS-based contexts, and
2620   conversely, silently ignore DTLS protocol version bounds when configuring
2621   TLS-based contexts.  The commands can be repeated to set bounds of both
2622   types.  The same applies with the corresponding "min_protocol" and
2623   "max_protocol" command-line switches, in case some application uses both TLS
2624   and DTLS.
2625
2626   SSL_CTX instances that are created for a fixed protocol version (e.g.
2627   TLSv1_server_method()) also silently ignore version bounds.  Previously
2628   attempts to apply bounds to these protocol versions would result in an
2629   error.  Now only the "version-flexible" SSL_CTX instances are subject to
2630   limits in configuration files in command-line options.
2631
2632   *Viktor Dukhovni*
2633
2634 * Handshake now fails if Extended Master Secret extension is dropped
2635   on renegotiation.
2636
2637   *Tomáš Mráz*
2638
2639 * The Oracle Developer Studio compiler will start reporting deprecated APIs
2640
2641### Changes between 1.1.1f and 1.1.1g [21 Apr 2020]
2642
2643 * Fixed segmentation fault in SSL_check_chain()
2644   Server or client applications that call the SSL_check_chain() function
2645   during or after a TLS 1.3 handshake may crash due to a NULL pointer
2646   dereference as a result of incorrect handling of the
2647   "signature_algorithms_cert" TLS extension. The crash occurs if an invalid
2648   or unrecognised signature algorithm is received from the peer. This could
2649   be exploited by a malicious peer in a Denial of Service attack.
2650   ([CVE-2020-1967])
2651
2652   *Benjamin Kaduk*
2653
2654 * Added AES consttime code for no-asm configurations
2655   an optional constant time support for AES was added
2656   when building openssl for no-asm.
2657   Enable with: ./config no-asm -DOPENSSL_AES_CONST_TIME
2658   Disable with: ./config no-asm -DOPENSSL_NO_AES_CONST_TIME
2659   At this time this feature is by default disabled.
2660   It will be enabled by default in 3.0.
2661
2662   *Bernd Edlinger*
2663
2664### Changes between 1.1.1e and 1.1.1f [31 Mar 2020]
2665
2666 * Revert the change of EOF detection while reading in libssl to avoid
2667   regressions in applications depending on the current way of reporting
2668   the EOF. As the existing method is not fully accurate the change to
2669   reporting the EOF via SSL_ERROR_SSL is kept on the current development
2670   branch and will be present in the 3.0 release.
2671
2672   *Tomáš Mráz*
2673
2674 * Revised BN_generate_prime_ex to not avoid factors 3..17863 in p-1
2675   when primes for RSA keys are computed.
2676   Since we previously always generated primes == 2 (mod 3) for RSA keys,
2677   the 2-prime and 3-prime RSA modules were easy to distinguish, since
2678   N = p*q = 1 (mod 3), but N = p*q*r = 2 (mod 3). Therefore fingerprinting
2679   2-prime vs. 3-prime RSA keys was possible by computing N mod 3.
2680   This avoids possible fingerprinting of newly generated RSA modules.
2681
2682   *Bernd Edlinger*
2683
2684### Changes between 1.1.1d and 1.1.1e [17 Mar 2020]
2685
2686 * Properly detect EOF while reading in libssl. Previously if we hit an EOF
2687   while reading in libssl then we would report an error back to the
2688   application (SSL_ERROR_SYSCALL) but errno would be 0. We now add
2689   an error to the stack (which means we instead return SSL_ERROR_SSL) and
2690   therefore give a hint as to what went wrong.
2691
2692   *Matt Caswell*
2693
2694 * Check that ed25519 and ed448 are allowed by the security level. Previously
2695   signature algorithms not using an MD were not being checked that they were
2696   allowed by the security level.
2697
2698   *Kurt Roeckx*
2699
2700 * Fixed SSL_get_servername() behaviour. The behaviour of SSL_get_servername()
2701   was not quite right. The behaviour was not consistent between resumption
2702   and normal handshakes, and also not quite consistent with historical
2703   behaviour. The behaviour in various scenarios has been clarified and
2704   it has been updated to make it match historical behaviour as closely as
2705   possible.
2706
2707   *Matt Caswell*
2708
2709 * *[VMS only]* The header files that the VMS compilers include automatically,
2710   `__DECC_INCLUDE_PROLOGUE.H` and `__DECC_INCLUDE_EPILOGUE.H`, use pragmas
2711   that the C++ compiler doesn't understand.  This is a shortcoming in the
2712   compiler, but can be worked around with `__cplusplus` guards.
2713
2714   C++ applications that use OpenSSL libraries must be compiled using the
2715   qualifier `/NAMES=(AS_IS,SHORTENED)` to be able to use all the OpenSSL
2716   functions.  Otherwise, only functions with symbols of less than 31
2717   characters can be used, as the linker will not be able to successfully
2718   resolve symbols with longer names.
2719
2720   *Richard Levitte*
2721
2722 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY.
2723   The presence of this system service is determined at run-time.
2724
2725   *Richard Levitte*
2726
2727 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just
2728   the first value.
2729
2730   *Jon Spillett*
2731
2732### Changes between 1.1.1c and 1.1.1d [10 Sep 2019]
2733
2734 * Fixed a fork protection issue. OpenSSL 1.1.1 introduced a rewritten random
2735   number generator (RNG). This was intended to include protection in the
2736   event of a fork() system call in order to ensure that the parent and child
2737   processes did not share the same RNG state. However this protection was not
2738   being used in the default case.
2739
2740   A partial mitigation for this issue is that the output from a high
2741   precision timer is mixed into the RNG state so the likelihood of a parent
2742   and child process sharing state is significantly reduced.
2743
2744   If an application already calls OPENSSL_init_crypto() explicitly using
2745   OPENSSL_INIT_ATFORK then this problem does not occur at all.
2746   ([CVE-2019-1549])
2747
2748   *Matthias St. Pierre*
2749
2750 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
2751   used even when parsing explicit parameters, when loading a encoded key
2752   or calling `EC_GROUP_new_from_ecpkparameters()`/
2753   `EC_GROUP_new_from_ecparameters()`.
2754   This prevents bypass of security hardening and performance gains,
2755   especially for curves with specialized EC_METHODs.
2756   By default, if a key encoded with explicit parameters is loaded and later
2757   encoded, the output is still encoded with explicit parameters, even if
2758   internally a "named" EC_GROUP is used for computation.
2759
2760   *Nicola Tuveri*
2761
2762 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
2763   this change, EC_GROUP_set_generator would accept order and/or cofactor as
2764   NULL. After this change, only the cofactor parameter can be NULL. It also
2765   does some minimal sanity checks on the passed order.
2766   ([CVE-2019-1547])
2767
2768   *Billy Bob Brumley*
2769
2770 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
2771   An attack is simple, if the first CMS_recipientInfo is valid but the
2772   second CMS_recipientInfo is chosen ciphertext. If the second
2773   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
2774   encryption key will be replaced by garbage, and the message cannot be
2775   decoded, but if the RSA decryption fails, the correct encryption key is
2776   used and the recipient will not notice the attack.
2777   As a work around for this potential attack the length of the decrypted
2778   key must be equal to the cipher default key length, in case the
2779   certifiate is not given and all recipientInfo are tried out.
2780   The old behaviour can be re-enabled in the CMS code by setting the
2781   CMS_DEBUG_DECRYPT flag.
2782   ([CVE-2019-1563])
2783
2784   *Bernd Edlinger*
2785
2786 * Early start up entropy quality from the DEVRANDOM seed source has been
2787   improved for older Linux systems.  The RAND subsystem will wait for
2788   /dev/random to be producing output before seeding from /dev/urandom.
2789   The seeded state is stored for future library initialisations using
2790   a system global shared memory segment.  The shared memory identifier
2791   can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to
2792   the desired value.  The default identifier is 114.
2793
2794   *Paul Dale*
2795
2796 * Correct the extended master secret constant on EBCDIC systems. Without this
2797   fix TLS connections between an EBCDIC system and a non-EBCDIC system that
2798   negotiate EMS will fail. Unfortunately this also means that TLS connections
2799   between EBCDIC systems with this fix, and EBCDIC systems without this
2800   fix will fail if they negotiate EMS.
2801
2802   *Matt Caswell*
2803
2804 * Use Windows installation paths in the mingw builds
2805
2806   Mingw isn't a POSIX environment per se, which means that Windows
2807   paths should be used for installation.
2808   ([CVE-2019-1552])
2809
2810   *Richard Levitte*
2811
2812 * Changed DH_check to accept parameters with order q and 2q subgroups.
2813   With order 2q subgroups the bit 0 of the private key is not secret
2814   but DH_generate_key works around that by clearing bit 0 of the
2815   private key for those. This avoids leaking bit 0 of the private key.
2816
2817   *Bernd Edlinger*
2818
2819 * Significantly reduce secure memory usage by the randomness pools.
2820
2821   *Paul Dale*
2822
2823 * Revert the DEVRANDOM_WAIT feature for Linux systems
2824
2825   The DEVRANDOM_WAIT feature added a select() call to wait for the
2826   /dev/random device to become readable before reading from the
2827   /dev/urandom device.
2828
2829   It turned out that this change had negative side effects on
2830   performance which were not acceptable. After some discussion it
2831   was decided to revert this feature and leave it up to the OS
2832   resp. the platform maintainer to ensure a proper initialization
2833   during early boot time.
2834
2835   *Matthias St. Pierre*
2836
2837### Changes between 1.1.1b and 1.1.1c [28 May 2019]
2838
2839 * Add build tests for C++.  These are generated files that only do one
2840   thing, to include one public OpenSSL head file each.  This tests that
2841   the public header files can be usefully included in a C++ application.
2842
2843   This test isn't enabled by default.  It can be enabled with the option
2844   'enable-buildtest-c++'.
2845
2846   *Richard Levitte*
2847
2848 * Enable SHA3 pre-hashing for ECDSA and DSA.
2849
2850   *Patrick Steuer*
2851
2852 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
2853   This changes the size when using the `genpkey` command when no size is given.
2854   It fixes an omission in earlier changes that changed all RSA, DSA and DH
2855   generation commands to use 2048 bits by default.
2856
2857   *Kurt Roeckx*
2858
2859 * Reorganize the manual pages to consistently have RETURN VALUES,
2860   EXAMPLES, SEE ALSO and HISTORY come in that order, and adjust
2861   util/fix-doc-nits accordingly.
2862
2863   *Paul Yang, Joshua Lock*
2864
2865 * Add the missing accessor EVP_PKEY_get0_engine()
2866
2867   *Matt Caswell*
2868
2869 * Have commands like `s_client` and `s_server` output the signature scheme
2870   along with other cipher suite parameters when debugging.
2871
2872   *Lorinczy Zsigmond*
2873
2874 * Make OPENSSL_config() error agnostic again.
2875
2876   *Richard Levitte*
2877
2878 * Do the error handling in RSA decryption constant time.
2879
2880   *Bernd Edlinger*
2881
2882 * Prevent over long nonces in ChaCha20-Poly1305.
2883
2884   ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input
2885   for every encryption operation. RFC 7539 specifies that the nonce value
2886   (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length
2887   and front pads the nonce with 0 bytes if it is less than 12
2888   bytes. However it also incorrectly allows a nonce to be set of up to 16
2889   bytes. In this case only the last 12 bytes are significant and any
2890   additional leading bytes are ignored.
2891
2892   It is a requirement of using this cipher that nonce values are
2893   unique. Messages encrypted using a reused nonce value are susceptible to
2894   serious confidentiality and integrity attacks. If an application changes
2895   the default nonce length to be longer than 12 bytes and then makes a
2896   change to the leading bytes of the nonce expecting the new value to be a
2897   new unique nonce then such an application could inadvertently encrypt
2898   messages with a reused nonce.
2899
2900   Additionally the ignored bytes in a long nonce are not covered by the
2901   integrity guarantee of this cipher. Any application that relies on the
2902   integrity of these ignored leading bytes of a long nonce may be further
2903   affected. Any OpenSSL internal use of this cipher, including in SSL/TLS,
2904   is safe because no such use sets such a long nonce value. However user
2905   applications that use this cipher directly and set a non-default nonce
2906   length to be longer than 12 bytes may be vulnerable.
2907
2908   This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk
2909   Greef of Ronomon.
2910   ([CVE-2019-1543])
2911
2912   *Matt Caswell*
2913
2914 * Add DEVRANDOM_WAIT feature for Linux systems
2915
2916   On older Linux systems where the getrandom() system call is not available,
2917   OpenSSL normally uses the /dev/urandom device for seeding its CSPRNG.
2918   Contrary to getrandom(), the /dev/urandom device will not block during
2919   early boot when the kernel CSPRNG has not been seeded yet.
2920
2921   To mitigate this known weakness, use select() to wait for /dev/random to
2922   become readable before reading from /dev/urandom.
2923
2924 * Ensure that SM2 only uses SM3 as digest algorithm
2925
2926   *Paul Yang*
2927
2928### Changes between 1.1.1a and 1.1.1b [26 Feb 2019]
2929
2930 * Change the info callback signals for the start and end of a post-handshake
2931   message exchange in TLSv1.3. In 1.1.1/1.1.1a we used SSL_CB_HANDSHAKE_START
2932   and SSL_CB_HANDSHAKE_DONE. Experience has shown that many applications get
2933   confused by this and assume that a TLSv1.2 renegotiation has started. This
2934   can break KeyUpdate handling. Instead we no longer signal the start and end
2935   of a post handshake message exchange (although the messages themselves are
2936   still signalled). This could break some applications that were expecting
2937   the old signals. However without this KeyUpdate is not usable for many
2938   applications.
2939
2940   *Matt Caswell*
2941
2942### Changes between 1.1.1 and 1.1.1a [20 Nov 2018]
2943
2944 * Timing vulnerability in DSA signature generation
2945
2946   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
2947   timing side channel attack. An attacker could use variations in the signing
2948   algorithm to recover the private key.
2949
2950   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
2951   ([CVE-2018-0734])
2952
2953   *Paul Dale*
2954
2955 * Timing vulnerability in ECDSA signature generation
2956
2957   The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
2958   timing side channel attack. An attacker could use variations in the signing
2959   algorithm to recover the private key.
2960
2961   This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
2962   ([CVE-2018-0735])
2963
2964   *Paul Dale*
2965
2966 * Fixed the issue that RAND_add()/RAND_seed() silently discards random input
2967   if its length exceeds 4096 bytes. The limit has been raised to a buffer size
2968   of two gigabytes and the error handling improved.
2969
2970   This issue was reported to OpenSSL by Dr. Falko Strenzke. It has been
2971   categorized as a normal bug, not a security issue, because the DRBG reseeds
2972   automatically and is fully functional even without additional randomness
2973   provided by the application.
2974
2975### Changes between 1.1.0i and 1.1.1 [11 Sep 2018]
2976
2977 * Add a new ClientHello callback. Provides a callback interface that gives
2978   the application the ability to adjust the nascent SSL object at the
2979   earliest stage of ClientHello processing, immediately after extensions have
2980   been collected but before they have been processed. In particular, this
2981   callback can adjust the supported TLS versions in response to the contents
2982   of the ClientHello
2983
2984   *Benjamin Kaduk*
2985
2986 * Add SM2 base algorithm support.
2987
2988   *Jack Lloyd*
2989
2990 * s390x assembly pack: add (improved) hardware-support for the following
2991   cryptographic primitives: sha3, shake, aes-gcm, aes-ccm, aes-ctr, aes-ofb,
2992   aes-cfb/cfb8, aes-ecb.
2993
2994   *Patrick Steuer*
2995
2996 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
2997   parameter is no longer accepted, as it leads to a corrupt table.  NULL
2998   pem_str is reserved for alias entries only.
2999
3000   *Richard Levitte*
3001
3002 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
3003   step for prime curves. The new implementation is based on formulae from
3004   differential addition-and-doubling in homogeneous projective coordinates
3005   from Izu-Takagi "A fast parallel elliptic curve multiplication resistant
3006   against side channel attacks" and Brier-Joye "Weierstrass Elliptic Curves
3007   and Side-Channel Attacks" Eq. (8) for y-coordinate recovery, modified
3008   to work in projective coordinates.
3009
3010   *Billy Bob Brumley, Nicola Tuveri*
3011
3012 * Change generating and checking of primes so that the error rate of not
3013   being prime depends on the intended use based on the size of the input.
3014   For larger primes this will result in more rounds of Miller-Rabin.
3015   The maximal error rate for primes with more than 1080 bits is lowered
3016   to 2^-128.
3017
3018   *Kurt Roeckx, Annie Yousar*
3019
3020 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
3021
3022   *Kurt Roeckx*
3023
3024 * The 'tsget' script is renamed to 'tsget.pl', to avoid confusion when
3025   moving between systems, and to avoid confusion when a Windows build is
3026   done with mingw vs with MSVC.  For POSIX installs, there's still a
3027   symlink or copy named 'tsget' to avoid that confusion as well.
3028
3029   *Richard Levitte*
3030
3031 * Revert blinding in ECDSA sign and instead make problematic addition
3032   length-invariant. Switch even to fixed-length Montgomery multiplication.
3033
3034   *Andy Polyakov*
3035
3036 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
3037   step for binary curves. The new implementation is based on formulae from
3038   differential addition-and-doubling in mixed Lopez-Dahab projective
3039   coordinates, modified to independently blind the operands.
3040
3041   *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri*
3042
3043 * Add a scaffold to optionally enhance the Montgomery ladder implementation
3044   for `ec_scalar_mul_ladder` (formerly `ec_mul_consttime`) allowing
3045   EC_METHODs to implement their own specialized "ladder step", to take
3046   advantage of more favorable coordinate systems or more efficient
3047   differential addition-and-doubling algorithms.
3048
3049   *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri*
3050
3051 * Modified the random device based seed sources to keep the relevant
3052   file descriptors open rather than reopening them on each access.
3053   This allows such sources to operate in a chroot() jail without
3054   the associated device nodes being available. This behaviour can be
3055   controlled using RAND_keep_random_devices_open().
3056
3057   *Paul Dale*
3058
3059 * Numerous side-channel attack mitigations have been applied. This may have
3060   performance impacts for some algorithms for the benefit of improved
3061   security. Specific changes are noted in this change log by their respective
3062   authors.
3063
3064   *Matt Caswell*
3065
3066 * AIX shared library support overhaul. Switch to AIX "natural" way of
3067   handling shared libraries, which means collecting shared objects of
3068   different versions and bitnesses in one common archive. This allows to
3069   mitigate conflict between 1.0 and 1.1 side-by-side installations. It
3070   doesn't affect the way 3rd party applications are linked, only how
3071   multi-version installation is managed.
3072
3073   *Andy Polyakov*
3074
3075 * Make ec_group_do_inverse_ord() more robust and available to other
3076   EC cryptosystems, so that irrespective of BN_FLG_CONSTTIME, SCA
3077   mitigations are applied to the fallback BN_mod_inverse().
3078   When using this function rather than BN_mod_inverse() directly, new
3079   EC cryptosystem implementations are then safer-by-default.
3080
3081   *Billy Bob Brumley*
3082
3083 * Add coordinate blinding for EC_POINT and implement projective
3084   coordinate blinding for generic prime curves as a countermeasure to
3085   chosen point SCA attacks.
3086
3087   *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley*
3088
3089 * Add blinding to ECDSA and DSA signatures to protect against side channel
3090   attacks discovered by Keegan Ryan (NCC Group).
3091
3092   *Matt Caswell*
3093
3094 * Enforce checking in the `pkeyutl` command to ensure that the input
3095   length does not exceed the maximum supported digest length when performing
3096   a sign, verify or verifyrecover operation.
3097
3098   *Matt Caswell*
3099
3100 * SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking
3101   I/O in combination with something like select() or poll() will hang. This
3102   can be turned off again using SSL_CTX_clear_mode().
3103   Many applications do not properly handle non-application data records, and
3104   TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works
3105   around the problems in those applications, but can also break some.
3106   It's recommended to read the manpages about SSL_read(), SSL_write(),
3107   SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and
3108   SSL_CTX_set_read_ahead() again.
3109
3110   *Kurt Roeckx*
3111
3112 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
3113   now allow empty (zero character) pass phrases.
3114
3115   *Richard Levitte*
3116
3117 * Apply blinding to binary field modular inversion and remove patent
3118   pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation.
3119
3120   *Billy Bob Brumley*
3121
3122 * Deprecate ec2_mult.c and unify scalar multiplication code paths for
3123   binary and prime elliptic curves.
3124
3125   *Billy Bob Brumley*
3126
3127 * Remove ECDSA nonce padding: EC_POINT_mul is now responsible for
3128   constant time fixed point multiplication.
3129
3130   *Billy Bob Brumley*
3131
3132 * Revise elliptic curve scalar multiplication with timing attack
3133   defenses: ec_wNAF_mul redirects to a constant time implementation
3134   when computing fixed point and variable point multiplication (which
3135   in OpenSSL are mostly used with secret scalars in keygen, sign,
3136   ECDH derive operations).
3137   *Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García,
3138    Sohaib ul Hassan*
3139
3140 * Updated CONTRIBUTING
3141
3142   *Rich Salz*
3143
3144 * Updated DRBG / RAND to request nonce and additional low entropy
3145   randomness from the system.
3146
3147   *Matthias St. Pierre*
3148
3149 * Updated 'openssl rehash' to use OpenSSL consistent default.
3150
3151   *Richard Levitte*
3152
3153 * Moved the load of the ssl_conf module to libcrypto, which helps
3154   loading engines that libssl uses before libssl is initialised.
3155
3156   *Matt Caswell*
3157
3158 * Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA
3159
3160   *Matt Caswell*
3161
3162 * Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases.
3163
3164   *Ingo Schwarze, Rich Salz*
3165
3166 * Added output of accepting IP address and port for 'openssl s_server'
3167
3168   *Richard Levitte*
3169
3170 * Added a new API for TLSv1.3 ciphersuites:
3171      SSL_CTX_set_ciphersuites()
3172      SSL_set_ciphersuites()
3173
3174   *Matt Caswell*
3175
3176 * Memory allocation failures consistently add an error to the error
3177   stack.
3178
3179   *Rich Salz*
3180
3181 * Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values
3182   in libcrypto when run as setuid/setgid.
3183
3184   *Bernd Edlinger*
3185
3186 * Load any config file by default when libssl is used.
3187
3188   *Matt Caswell*
3189
3190 * Added new public header file <openssl/rand_drbg.h> and documentation
3191   for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview.
3192
3193   *Matthias St. Pierre*
3194
3195 * QNX support removed (cannot find contributors to get their approval
3196   for the license change).
3197
3198   *Rich Salz*
3199
3200 * TLSv1.3 replay protection for early data has been implemented. See the
3201   SSL_read_early_data() man page for further details.
3202
3203   *Matt Caswell*
3204
3205 * Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite
3206   configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and
3207   below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3.
3208   In order to avoid issues where legacy TLSv1.2 ciphersuite configuration
3209   would otherwise inadvertently disable all TLSv1.3 ciphersuites the
3210   configuration has been separated out. See the ciphers man page or the
3211   SSL_CTX_set_ciphersuites() man page for more information.
3212
3213   *Matt Caswell*
3214
3215 * On POSIX (BSD, Linux, ...) systems the ocsp(1) command running
3216   in responder mode now supports the new "-multi" option, which
3217   spawns the specified number of child processes to handle OCSP
3218   requests.  The "-timeout" option now also limits the OCSP
3219   responder's patience to wait to receive the full client request
3220   on a newly accepted connection. Child processes are respawned
3221   as needed, and the CA index file is automatically reloaded
3222   when changed.  This makes it possible to run the "ocsp" responder
3223   as a long-running service, making the OpenSSL CA somewhat more
3224   feature-complete.  In this mode, most diagnostic messages logged
3225   after entering the event loop are logged via syslog(3) rather than
3226   written to stderr.
3227
3228   *Viktor Dukhovni*
3229
3230 * Added support for X448 and Ed448. Heavily based on original work by
3231   Mike Hamburg.
3232
3233   *Matt Caswell*
3234
3235 * Extend OSSL_STORE with capabilities to search and to narrow the set of
3236   objects loaded.  This adds the functions OSSL_STORE_expect() and
3237   OSSL_STORE_find() as well as needed tools to construct searches and
3238   get the search data out of them.
3239
3240   *Richard Levitte*
3241
3242 * Support for TLSv1.3 added. Note that users upgrading from an earlier
3243   version of OpenSSL should review their configuration settings to ensure
3244   that they are still appropriate for TLSv1.3. For further information see:
3245   <https://wiki.openssl.org/index.php/TLS1.3>
3246
3247   *Matt Caswell*
3248
3249 * Grand redesign of the OpenSSL random generator
3250
3251   The default RAND method now utilizes an AES-CTR DRBG according to
3252   NIST standard SP 800-90Ar1. The new random generator is essentially
3253   a port of the default random generator from the OpenSSL FIPS 2.0
3254   object module. It is a hybrid deterministic random bit generator
3255   using an AES-CTR bit stream and which seeds and reseeds itself
3256   automatically using trusted system entropy sources.
3257
3258   Some of its new features are:
3259    - Support for multiple DRBG instances with seed chaining.
3260    - The default RAND method makes use of a DRBG.
3261    - There is a public and private DRBG instance.
3262    - The DRBG instances are fork-safe.
3263    - Keep all global DRBG instances on the secure heap if it is enabled.
3264    - The public and private DRBG instance are per thread for lock free
3265      operation
3266
3267   *Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre*
3268
3269 * Changed Configure so it only says what it does and doesn't dump
3270   so much data.  Instead, ./configdata.pm should be used as a script
3271   to display all sorts of configuration data.
3272
3273   *Richard Levitte*
3274
3275 * Added processing of "make variables" to Configure.
3276
3277   *Richard Levitte*
3278
3279 * Added SHA512/224 and SHA512/256 algorithm support.
3280
3281   *Paul Dale*
3282
3283 * The last traces of Netware support, first removed in 1.1.0, have
3284   now been removed.
3285
3286   *Rich Salz*
3287
3288 * Get rid of Makefile.shared, and in the process, make the processing
3289   of certain files (rc.obj, or the .def/.map/.opt files produced from
3290   the ordinal files) more visible and hopefully easier to trace and
3291   debug (or make silent).
3292
3293   *Richard Levitte*
3294
3295 * Make it possible to have environment variable assignments as
3296   arguments to config / Configure.
3297
3298   *Richard Levitte*
3299
3300 * Add multi-prime RSA (RFC 8017) support.
3301
3302   *Paul Yang*
3303
3304 * Add SM3 implemented according to GB/T 32905-2016
3305   *Jack Lloyd <jack.lloyd@ribose.com>,*
3306   *Ronald Tse <ronald.tse@ribose.com>,*
3307   *Erick Borsboom <erick.borsboom@ribose.com>*
3308
3309 * Add 'Maximum Fragment Length' TLS extension negotiation and support
3310   as documented in RFC6066.
3311   Based on a patch from Tomasz Moń
3312
3313   *Filipe Raimundo da Silva*
3314
3315 * Add SM4 implemented according to GB/T 32907-2016.
3316   *Jack Lloyd <jack.lloyd@ribose.com>,*
3317   *Ronald Tse <ronald.tse@ribose.com>,*
3318   *Erick Borsboom <erick.borsboom@ribose.com>*
3319
3320 * Reimplement -newreq-nodes and ERR_error_string_n; the
3321   original author does not agree with the license change.
3322
3323   *Rich Salz*
3324
3325 * Add ARIA AEAD TLS support.
3326
3327   *Jon Spillett*
3328
3329 * Some macro definitions to support VS6 have been removed.  Visual
3330   Studio 6 has not worked since 1.1.0
3331
3332   *Rich Salz*
3333
3334 * Add ERR_clear_last_mark(), to allow callers to clear the last mark
3335   without clearing the errors.
3336
3337   *Richard Levitte*
3338
3339 * Add "atfork" functions.  If building on a system that without
3340   pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application
3341   requirements.  The RAND facility now uses/requires this.
3342
3343   *Rich Salz*
3344
3345 * Add SHA3.
3346
3347   *Andy Polyakov*
3348
3349 * The UI API becomes a permanent and integral part of libcrypto, i.e.
3350   not possible to disable entirely.  However, it's still possible to
3351   disable the console reading UI method, UI_OpenSSL() (use UI_null()
3352   as a fallback).
3353
3354   To disable, configure with 'no-ui-console'.  'no-ui' is still
3355   possible to use as an alias.  Check at compile time with the
3356   macro OPENSSL_NO_UI_CONSOLE.  The macro OPENSSL_NO_UI is still
3357   possible to check and is an alias for OPENSSL_NO_UI_CONSOLE.
3358
3359   *Richard Levitte*
3360
3361 * Add a STORE module, which implements a uniform and URI based reader of
3362   stores that can contain keys, certificates, CRLs and numerous other
3363   objects.  The main API is loosely based on a few stdio functions,
3364   and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof,
3365   OSSL_STORE_error and OSSL_STORE_close.
3366   The implementation uses backends called "loaders" to implement arbitrary
3367   URI schemes.  There is one built in "loader" for the 'file' scheme.
3368
3369   *Richard Levitte*
3370
3371 * Add devcrypto engine.  This has been implemented against cryptodev-linux,
3372   then adjusted to work on FreeBSD 8.4 as well.
3373   Enable by configuring with 'enable-devcryptoeng'.  This is done by default
3374   on BSD implementations, as cryptodev.h is assumed to exist on all of them.
3375
3376   *Richard Levitte*
3377
3378 * Module names can prefixed with OSSL_ or OPENSSL_.  This affects
3379   util/mkerr.pl, which is adapted to allow those prefixes, leading to
3380   error code calls like this:
3381
3382           OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER);
3383
3384   With this change, we claim the namespaces OSSL and OPENSSL in a manner
3385   that can be encoded in C.  For the foreseeable future, this will only
3386   affect new modules.
3387
3388   *Richard Levitte and Tim Hudson*
3389
3390 * Removed BSD cryptodev engine.
3391
3392   *Rich Salz*
3393
3394 * Add a build target 'build_all_generated', to build all generated files
3395   and only that.  This can be used to prepare everything that requires
3396   things like perl for a system that lacks perl and then move everything
3397   to that system and do the rest of the build there.
3398
3399   *Richard Levitte*
3400
3401 * In the UI interface, make it possible to duplicate the user data.  This
3402   can be used by engines that need to retain the data for a longer time
3403   than just the call where this user data is passed.
3404
3405   *Richard Levitte*
3406
3407 * Ignore the '-named_curve auto' value for compatibility of applications
3408   with OpenSSL 1.0.2.
3409
3410   *Tomáš Mráz <tmraz@fedoraproject.org>*
3411
3412 * Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2
3413   bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such
3414   alerts across multiple records (some of which could be empty). In practice
3415   it make no sense to send an empty alert record, or to fragment one. TLSv1.3
3416   prohibits this altogether and other libraries (BoringSSL, NSS) do not
3417   support this at all. Supporting it adds significant complexity to the
3418   record layer, and its removal is unlikely to cause interoperability
3419   issues.
3420
3421   *Matt Caswell*
3422
3423 * Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed
3424   with Z.  These are meant to replace LONG and ZLONG and to be size safe.
3425   The use of LONG and ZLONG is discouraged and scheduled for deprecation
3426   in OpenSSL 1.2.0.
3427
3428   *Richard Levitte*
3429
3430 * Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string,
3431   'z' is to be used for [s]size_t, and 'j' - with [u]int64_t.
3432
3433   *Richard Levitte, Andy Polyakov*
3434
3435 * Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine()
3436   does for RSA, etc.
3437
3438   *Richard Levitte*
3439
3440 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
3441   platform rather than 'mingw'.
3442
3443   *Richard Levitte*
3444
3445 * The functions X509_STORE_add_cert and X509_STORE_add_crl return
3446   success if they are asked to add an object which already exists
3447   in the store. This change cascades to other functions which load
3448   certificates and CRLs.
3449
3450   *Paul Dale*
3451
3452 * x86_64 assembly pack: annotate code with DWARF CFI directives to
3453   facilitate stack unwinding even from assembly subroutines.
3454
3455   *Andy Polyakov*
3456
3457 * Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN.
3458   Also remove OPENSSL_GLOBAL entirely, as it became a no-op.
3459
3460   *Richard Levitte*
3461
3462 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
3463   VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
3464   which is the minimum version we support.
3465
3466   *Richard Levitte*
3467
3468 * Certificate time validation (X509_cmp_time) enforces stricter
3469   compliance with RFC 5280. Fractional seconds and timezone offsets
3470   are no longer allowed.
3471
3472   *Emilia Käsper*
3473
3474 * Add support for ARIA
3475
3476   *Paul Dale*
3477
3478 * s_client will now send the Server Name Indication (SNI) extension by
3479   default unless the new "-noservername" option is used. The server name is
3480   based on the host provided to the "-connect" option unless overridden by
3481   using "-servername".
3482
3483   *Matt Caswell*
3484
3485 * Add support for SipHash
3486
3487   *Todd Short*
3488
3489 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0
3490   or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
3491   prevent issues where no progress is being made and the peer continually
3492   sends unrecognised record types, using up resources processing them.
3493
3494   *Matt Caswell*
3495
3496 * 'openssl passwd' can now produce SHA256 and SHA512 based output,
3497   using the algorithm defined in
3498   <https://www.akkadia.org/drepper/SHA-crypt.txt>
3499
3500   *Richard Levitte*
3501
3502 * Heartbeat support has been removed; the ABI is changed for now.
3503
3504   *Richard Levitte, Rich Salz*
3505
3506 * Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd.
3507
3508   *Emilia Käsper*
3509
3510 * The RSA "null" method, which was partially supported to avoid patent
3511   issues, has been replaced to always returns NULL.
3512
3513   *Rich Salz*
3514
3515OpenSSL 1.1.0
3516-------------
3517
3518### Changes between 1.1.0k and 1.1.0l [10 Sep 2019]
3519
3520 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
3521   used even when parsing explicit parameters, when loading a encoded key
3522   or calling `EC_GROUP_new_from_ecpkparameters()`/
3523   `EC_GROUP_new_from_ecparameters()`.
3524   This prevents bypass of security hardening and performance gains,
3525   especially for curves with specialized EC_METHODs.
3526   By default, if a key encoded with explicit parameters is loaded and later
3527   encoded, the output is still encoded with explicit parameters, even if
3528   internally a "named" EC_GROUP is used for computation.
3529
3530   *Nicola Tuveri*
3531
3532 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
3533   this change, EC_GROUP_set_generator would accept order and/or cofactor as
3534   NULL. After this change, only the cofactor parameter can be NULL. It also
3535   does some minimal sanity checks on the passed order.
3536   ([CVE-2019-1547])
3537
3538   *Billy Bob Brumley*
3539
3540 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
3541   An attack is simple, if the first CMS_recipientInfo is valid but the
3542   second CMS_recipientInfo is chosen ciphertext. If the second
3543   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
3544   encryption key will be replaced by garbage, and the message cannot be
3545   decoded, but if the RSA decryption fails, the correct encryption key is
3546   used and the recipient will not notice the attack.
3547   As a work around for this potential attack the length of the decrypted
3548   key must be equal to the cipher default key length, in case the
3549   certifiate is not given and all recipientInfo are tried out.
3550   The old behaviour can be re-enabled in the CMS code by setting the
3551   CMS_DEBUG_DECRYPT flag.
3552   ([CVE-2019-1563])
3553
3554   *Bernd Edlinger*
3555
3556 * Use Windows installation paths in the mingw builds
3557
3558   Mingw isn't a POSIX environment per se, which means that Windows
3559   paths should be used for installation.
3560   ([CVE-2019-1552])
3561
3562   *Richard Levitte*
3563
3564### Changes between 1.1.0j and 1.1.0k [28 May 2019]
3565
3566 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
3567   This changes the size when using the `genpkey` command when no size is given.
3568   It fixes an omission in earlier changes that changed all RSA, DSA and DH
3569   generation commands to use 2048 bits by default.
3570
3571   *Kurt Roeckx*
3572
3573 * Prevent over long nonces in ChaCha20-Poly1305.
3574
3575   ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input
3576   for every encryption operation. RFC 7539 specifies that the nonce value
3577   (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length
3578   and front pads the nonce with 0 bytes if it is less than 12
3579   bytes. However it also incorrectly allows a nonce to be set of up to 16
3580   bytes. In this case only the last 12 bytes are significant and any
3581   additional leading bytes are ignored.
3582
3583   It is a requirement of using this cipher that nonce values are
3584   unique. Messages encrypted using a reused nonce value are susceptible to
3585   serious confidentiality and integrity attacks. If an application changes
3586   the default nonce length to be longer than 12 bytes and then makes a
3587   change to the leading bytes of the nonce expecting the new value to be a
3588   new unique nonce then such an application could inadvertently encrypt
3589   messages with a reused nonce.
3590
3591   Additionally the ignored bytes in a long nonce are not covered by the
3592   integrity guarantee of this cipher. Any application that relies on the
3593   integrity of these ignored leading bytes of a long nonce may be further
3594   affected. Any OpenSSL internal use of this cipher, including in SSL/TLS,
3595   is safe because no such use sets such a long nonce value. However user
3596   applications that use this cipher directly and set a non-default nonce
3597   length to be longer than 12 bytes may be vulnerable.
3598
3599   This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk
3600   Greef of Ronomon.
3601   ([CVE-2019-1543])
3602
3603   *Matt Caswell*
3604
3605 * Added SCA hardening for modular field inversion in EC_GROUP through
3606   a new dedicated field_inv() pointer in EC_METHOD.
3607   This also addresses a leakage affecting conversions from projective
3608   to affine coordinates.
3609
3610   *Billy Bob Brumley, Nicola Tuveri*
3611
3612 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
3613   re-used X509_PUBKEY object if the second PUBKEY is malformed.
3614
3615   *Bernd Edlinger*
3616
3617 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
3618
3619   *Richard Levitte*
3620
3621 * Remove the 'dist' target and add a tarball building script.  The
3622   'dist' target has fallen out of use, and it shouldn't be
3623   necessary to configure just to create a source distribution.
3624
3625   *Richard Levitte*
3626
3627### Changes between 1.1.0i and 1.1.0j [20 Nov 2018]
3628
3629 * Timing vulnerability in DSA signature generation
3630
3631   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
3632   timing side channel attack. An attacker could use variations in the signing
3633   algorithm to recover the private key.
3634
3635   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
3636   ([CVE-2018-0734])
3637
3638   *Paul Dale*
3639
3640 * Timing vulnerability in ECDSA signature generation
3641
3642   The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
3643   timing side channel attack. An attacker could use variations in the signing
3644   algorithm to recover the private key.
3645
3646   This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
3647   ([CVE-2018-0735])
3648
3649   *Paul Dale*
3650
3651 * Add coordinate blinding for EC_POINT and implement projective
3652   coordinate blinding for generic prime curves as a countermeasure to
3653   chosen point SCA attacks.
3654
3655   *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley*
3656
3657### Changes between 1.1.0h and 1.1.0i [14 Aug 2018]
3658
3659 * Client DoS due to large DH parameter
3660
3661   During key agreement in a TLS handshake using a DH(E) based ciphersuite a
3662   malicious server can send a very large prime value to the client. This will
3663   cause the client to spend an unreasonably long period of time generating a
3664   key for this prime resulting in a hang until the client has finished. This
3665   could be exploited in a Denial Of Service attack.
3666
3667   This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
3668   ([CVE-2018-0732])
3669
3670   *Guido Vranken*
3671
3672 * Cache timing vulnerability in RSA Key Generation
3673
3674   The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
3675   a cache timing side channel attack. An attacker with sufficient access to
3676   mount cache timing attacks during the RSA key generation process could
3677   recover the private key.
3678
3679   This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
3680   Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
3681   ([CVE-2018-0737])
3682
3683   *Billy Brumley*
3684
3685 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
3686   parameter is no longer accepted, as it leads to a corrupt table.  NULL
3687   pem_str is reserved for alias entries only.
3688
3689   *Richard Levitte*
3690
3691 * Revert blinding in ECDSA sign and instead make problematic addition
3692   length-invariant. Switch even to fixed-length Montgomery multiplication.
3693
3694   *Andy Polyakov*
3695
3696 * Change generating and checking of primes so that the error rate of not
3697   being prime depends on the intended use based on the size of the input.
3698   For larger primes this will result in more rounds of Miller-Rabin.
3699   The maximal error rate for primes with more than 1080 bits is lowered
3700   to 2^-128.
3701
3702   *Kurt Roeckx, Annie Yousar*
3703
3704 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
3705
3706   *Kurt Roeckx*
3707
3708 * Add blinding to ECDSA and DSA signatures to protect against side channel
3709   attacks discovered by Keegan Ryan (NCC Group).
3710
3711   *Matt Caswell*
3712
3713 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
3714   now allow empty (zero character) pass phrases.
3715
3716   *Richard Levitte*
3717
3718 * Certificate time validation (X509_cmp_time) enforces stricter
3719   compliance with RFC 5280. Fractional seconds and timezone offsets
3720   are no longer allowed.
3721
3722   *Emilia Käsper*
3723
3724 * Fixed a text canonicalisation bug in CMS
3725
3726   Where a CMS detached signature is used with text content the text goes
3727   through a canonicalisation process first prior to signing or verifying a
3728   signature. This process strips trailing space at the end of lines, converts
3729   line terminators to CRLF and removes additional trailing line terminators
3730   at the end of a file. A bug in the canonicalisation process meant that
3731   some characters, such as form-feed, were incorrectly treated as whitespace
3732   and removed. This is contrary to the specification (RFC5485). This fix
3733   could mean that detached text data signed with an earlier version of
3734   OpenSSL 1.1.0 may fail to verify using the fixed version, or text data
3735   signed with a fixed OpenSSL may fail to verify with an earlier version of
3736   OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data
3737   and use the "-binary" flag (for the "cms" command line application) or set
3738   the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()).
3739
3740   *Matt Caswell*
3741
3742### Changes between 1.1.0g and 1.1.0h [27 Mar 2018]
3743
3744 * Constructed ASN.1 types with a recursive definition could exceed the stack
3745
3746   Constructed ASN.1 types with a recursive definition (such as can be found
3747   in PKCS7) could eventually exceed the stack given malicious input with
3748   excessive recursion. This could result in a Denial Of Service attack. There
3749   are no such structures used within SSL/TLS that come from untrusted sources
3750   so this is considered safe.
3751
3752   This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
3753   project.
3754   ([CVE-2018-0739])
3755
3756   *Matt Caswell*
3757
3758 * Incorrect CRYPTO_memcmp on HP-UX PA-RISC
3759
3760   Because of an implementation bug the PA-RISC CRYPTO_memcmp function is
3761   effectively reduced to only comparing the least significant bit of each
3762   byte. This allows an attacker to forge messages that would be considered as
3763   authenticated in an amount of tries lower than that guaranteed by the
3764   security claims of the scheme. The module can only be compiled by the
3765   HP-UX assembler, so that only HP-UX PA-RISC targets are affected.
3766
3767   This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg
3768   (IBM).
3769   ([CVE-2018-0733])
3770
3771   *Andy Polyakov*
3772
3773 * Add a build target 'build_all_generated', to build all generated files
3774   and only that.  This can be used to prepare everything that requires
3775   things like perl for a system that lacks perl and then move everything
3776   to that system and do the rest of the build there.
3777
3778   *Richard Levitte*
3779
3780 * Backport SSL_OP_NO_RENGOTIATION
3781
3782   OpenSSL 1.0.2 and below had the ability to disable renegotiation using the
3783   (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity
3784   changes this is no longer possible in 1.1.0. Therefore the new
3785   SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to
3786   1.1.0 to provide equivalent functionality.
3787
3788   Note that if an application built against 1.1.0h headers (or above) is run
3789   using an older version of 1.1.0 (prior to 1.1.0h) then the option will be
3790   accepted but nothing will happen, i.e. renegotiation will not be prevented.
3791
3792   *Matt Caswell*
3793
3794 * Removed the OS390-Unix config target.  It relied on a script that doesn't
3795   exist.
3796
3797   *Rich Salz*
3798
3799 * rsaz_1024_mul_avx2 overflow bug on x86_64
3800
3801   There is an overflow bug in the AVX2 Montgomery multiplication procedure
3802   used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
3803   Analysis suggests that attacks against RSA and DSA as a result of this
3804   defect would be very difficult to perform and are not believed likely.
3805   Attacks against DH1024 are considered just feasible, because most of the
3806   work necessary to deduce information about a private key may be performed
3807   offline. The amount of resources required for such an attack would be
3808   significant. However, for an attack on TLS to be meaningful, the server
3809   would have to share the DH1024 private key among multiple clients, which is
3810   no longer an option since CVE-2016-0701.
3811
3812   This only affects processors that support the AVX2 but not ADX extensions
3813   like Intel Haswell (4th generation).
3814
3815   This issue was reported to OpenSSL by David Benjamin (Google). The issue
3816   was originally found via the OSS-Fuzz project.
3817   ([CVE-2017-3738])
3818
3819   *Andy Polyakov*
3820
3821### Changes between 1.1.0f and 1.1.0g [2 Nov 2017]
3822
3823 * bn_sqrx8x_internal carry bug on x86_64
3824
3825   There is a carry propagating bug in the x86_64 Montgomery squaring
3826   procedure. No EC algorithms are affected. Analysis suggests that attacks
3827   against RSA and DSA as a result of this defect would be very difficult to
3828   perform and are not believed likely. Attacks against DH are considered just
3829   feasible (although very difficult) because most of the work necessary to
3830   deduce information about a private key may be performed offline. The amount
3831   of resources required for such an attack would be very significant and
3832   likely only accessible to a limited number of attackers. An attacker would
3833   additionally need online access to an unpatched system using the target
3834   private key in a scenario with persistent DH parameters and a private
3835   key that is shared between multiple clients.
3836
3837   This only affects processors that support the BMI1, BMI2 and ADX extensions
3838   like Intel Broadwell (5th generation) and later or AMD Ryzen.
3839
3840   This issue was reported to OpenSSL by the OSS-Fuzz project.
3841   ([CVE-2017-3736])
3842
3843   *Andy Polyakov*
3844
3845 * Malformed X.509 IPAddressFamily could cause OOB read
3846
3847   If an X.509 certificate has a malformed IPAddressFamily extension,
3848   OpenSSL could do a one-byte buffer overread. The most likely result
3849   would be an erroneous display of the certificate in text format.
3850
3851   This issue was reported to OpenSSL by the OSS-Fuzz project.
3852   ([CVE-2017-3735])
3853
3854   *Rich Salz*
3855
3856### Changes between 1.1.0e and 1.1.0f [25 May 2017]
3857
3858 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
3859   platform rather than 'mingw'.
3860
3861   *Richard Levitte*
3862
3863 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
3864   VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
3865   which is the minimum version we support.
3866
3867   *Richard Levitte*
3868
3869### Changes between 1.1.0d and 1.1.0e [16 Feb 2017]
3870
3871 * Encrypt-Then-Mac renegotiation crash
3872
3873   During a renegotiation handshake if the Encrypt-Then-Mac extension is
3874   negotiated where it was not in the original handshake (or vice-versa) then
3875   this can cause OpenSSL to crash (dependant on ciphersuite). Both clients
3876   and servers are affected.
3877
3878   This issue was reported to OpenSSL by Joe Orton (Red Hat).
3879   ([CVE-2017-3733])
3880
3881   *Matt Caswell*
3882
3883### Changes between 1.1.0c and 1.1.0d [26 Jan 2017]
3884
3885 * Truncated packet could crash via OOB read
3886
3887   If one side of an SSL/TLS path is running on a 32-bit host and a specific
3888   cipher is being used, then a truncated packet can cause that host to
3889   perform an out-of-bounds read, usually resulting in a crash.
3890
3891   This issue was reported to OpenSSL by Robert Święcki of Google.
3892   ([CVE-2017-3731])
3893
3894   *Andy Polyakov*
3895
3896 * Bad (EC)DHE parameters cause a client crash
3897
3898   If a malicious server supplies bad parameters for a DHE or ECDHE key
3899   exchange then this can result in the client attempting to dereference a
3900   NULL pointer leading to a client crash. This could be exploited in a Denial
3901   of Service attack.
3902
3903   This issue was reported to OpenSSL by Guido Vranken.
3904   ([CVE-2017-3730])
3905
3906   *Matt Caswell*
3907
3908 * BN_mod_exp may produce incorrect results on x86_64
3909
3910   There is a carry propagating bug in the x86_64 Montgomery squaring
3911   procedure. No EC algorithms are affected. Analysis suggests that attacks
3912   against RSA and DSA as a result of this defect would be very difficult to
3913   perform and are not believed likely. Attacks against DH are considered just
3914   feasible (although very difficult) because most of the work necessary to
3915   deduce information about a private key may be performed offline. The amount
3916   of resources required for such an attack would be very significant and
3917   likely only accessible to a limited number of attackers. An attacker would
3918   additionally need online access to an unpatched system using the target
3919   private key in a scenario with persistent DH parameters and a private
3920   key that is shared between multiple clients. For example this can occur by
3921   default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
3922   similar to CVE-2015-3193 but must be treated as a separate problem.
3923
3924   This issue was reported to OpenSSL by the OSS-Fuzz project.
3925   ([CVE-2017-3732])
3926
3927   *Andy Polyakov*
3928
3929### Changes between 1.1.0b and 1.1.0c [10 Nov 2016]
3930
3931 * ChaCha20/Poly1305 heap-buffer-overflow
3932
3933   TLS connections using `*-CHACHA20-POLY1305` ciphersuites are susceptible to
3934   a DoS attack by corrupting larger payloads. This can result in an OpenSSL
3935   crash. This issue is not considered to be exploitable beyond a DoS.
3936
3937   This issue was reported to OpenSSL by Robert Święcki (Google Security Team)
3938   ([CVE-2016-7054])
3939
3940   *Richard Levitte*
3941
3942 * CMS Null dereference
3943
3944   Applications parsing invalid CMS structures can crash with a NULL pointer
3945   dereference. This is caused by a bug in the handling of the ASN.1 CHOICE
3946   type in OpenSSL 1.1.0 which can result in a NULL value being passed to the
3947   structure callback if an attempt is made to free certain invalid encodings.
3948   Only CHOICE structures using a callback which do not handle NULL value are
3949   affected.
3950
3951   This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure.
3952   ([CVE-2016-7053])
3953
3954   *Stephen Henson*
3955
3956 * Montgomery multiplication may produce incorrect results
3957
3958   There is a carry propagating bug in the Broadwell-specific Montgomery
3959   multiplication procedure that handles input lengths divisible by, but
3960   longer than 256 bits. Analysis suggests that attacks against RSA, DSA
3961   and DH private keys are impossible. This is because the subroutine in
3962   question is not used in operations with the private key itself and an input
3963   of the attacker's direct choice. Otherwise the bug can manifest itself as
3964   transient authentication and key negotiation failures or reproducible
3965   erroneous outcome of public-key operations with specially crafted input.
3966   Among EC algorithms only Brainpool P-512 curves are affected and one
3967   presumably can attack ECDH key negotiation. Impact was not analyzed in
3968   detail, because pre-requisites for attack are considered unlikely. Namely
3969   multiple clients have to choose the curve in question and the server has to
3970   share the private key among them, neither of which is default behaviour.
3971   Even then only clients that chose the curve will be affected.
3972
3973   This issue was publicly reported as transient failures and was not
3974   initially recognized as a security issue. Thanks to Richard Morgan for
3975   providing reproducible case.
3976   ([CVE-2016-7055])
3977
3978   *Andy Polyakov*
3979
3980 * Removed automatic addition of RPATH in shared libraries and executables,
3981   as this was a remainder from OpenSSL 1.0.x and isn't needed any more.
3982
3983   *Richard Levitte*
3984
3985### Changes between 1.1.0a and 1.1.0b [26 Sep 2016]
3986
3987 * Fix Use After Free for large message sizes
3988
3989   The patch applied to address CVE-2016-6307 resulted in an issue where if a
3990   message larger than approx 16k is received then the underlying buffer to
3991   store the incoming message is reallocated and moved. Unfortunately a
3992   dangling pointer to the old location is left which results in an attempt to
3993   write to the previously freed location. This is likely to result in a
3994   crash, however it could potentially lead to execution of arbitrary code.
3995
3996   This issue only affects OpenSSL 1.1.0a.
3997
3998   This issue was reported to OpenSSL by Robert Święcki.
3999   ([CVE-2016-6309])
4000
4001   *Matt Caswell*
4002
4003### Changes between 1.1.0 and 1.1.0a [22 Sep 2016]
4004
4005 * OCSP Status Request extension unbounded memory growth
4006
4007   A malicious client can send an excessively large OCSP Status Request
4008   extension. If that client continually requests renegotiation, sending a
4009   large OCSP Status Request extension each time, then there will be unbounded
4010   memory growth on the server. This will eventually lead to a Denial Of
4011   Service attack through memory exhaustion. Servers with a default
4012   configuration are vulnerable even if they do not support OCSP. Builds using
4013   the "no-ocsp" build time option are not affected.
4014
4015   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
4016   ([CVE-2016-6304])
4017
4018   *Matt Caswell*
4019
4020 * SSL_peek() hang on empty record
4021
4022   OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer
4023   sends an empty record. This could be exploited by a malicious peer in a
4024   Denial Of Service attack.
4025
4026   This issue was reported to OpenSSL by Alex Gaynor.
4027   ([CVE-2016-6305])
4028
4029   *Matt Caswell*
4030
4031 * Excessive allocation of memory in tls_get_message_header() and
4032   dtls1_preprocess_fragment()
4033
4034   A (D)TLS message includes 3 bytes for its length in the header for the
4035   message. This would allow for messages up to 16Mb in length. Messages of
4036   this length are excessive and OpenSSL includes a check to ensure that a
4037   peer is sending reasonably sized messages in order to avoid too much memory
4038   being consumed to service a connection. A flaw in the logic of version
4039   1.1.0 means that memory for the message is allocated too early, prior to
4040   the excessive message length check. Due to way memory is allocated in
4041   OpenSSL this could mean an attacker could force up to 21Mb to be allocated
4042   to service a connection. This could lead to a Denial of Service through
4043   memory exhaustion. However, the excessive message length check still takes
4044   place, and this would cause the connection to immediately fail. Assuming
4045   that the application calls SSL_free() on the failed connection in a timely
4046   manner then the 21Mb of allocated memory will then be immediately freed
4047   again. Therefore the excessive memory allocation will be transitory in
4048   nature. This then means that there is only a security impact if:
4049
4050   1) The application does not call SSL_free() in a timely manner in the event
4051   that the connection fails
4052   or
4053   2) The application is working in a constrained environment where there is
4054   very little free memory
4055   or
4056   3) The attacker initiates multiple connection attempts such that there are
4057   multiple connections in a state where memory has been allocated for the
4058   connection; SSL_free() has not yet been called; and there is insufficient
4059   memory to service the multiple requests.
4060
4061   Except in the instance of (1) above any Denial Of Service is likely to be
4062   transitory because as soon as the connection fails the memory is
4063   subsequently freed again in the SSL_free() call. However there is an
4064   increased risk during this period of application crashes due to the lack of
4065   memory - which would then mean a more serious Denial of Service.
4066
4067   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
4068   (CVE-2016-6307 and CVE-2016-6308)
4069
4070   *Matt Caswell*
4071
4072 * solaris-x86-cc, i.e. 32-bit configuration with vendor compiler,
4073   had to be removed. Primary reason is that vendor assembler can't
4074   assemble our modules with -KPIC flag. As result it, assembly
4075   support, was not even available as option. But its lack means
4076   lack of side-channel resistant code, which is incompatible with
4077   security by todays standards. Fortunately gcc is readily available
4078   prepackaged option, which we firmly point at...
4079
4080   *Andy Polyakov*
4081
4082### Changes between 1.0.2h and 1.1.0  [25 Aug 2016]
4083
4084 * Windows command-line tool supports UTF-8 opt-in option for arguments
4085   and console input. Setting OPENSSL_WIN32_UTF8 environment variable
4086   (to any value) allows Windows user to access PKCS#12 file generated
4087   with Windows CryptoAPI and protected with non-ASCII password, as well
4088   as files generated under UTF-8 locale on Linux also protected with
4089   non-ASCII password.
4090
4091   *Andy Polyakov*
4092
4093 * To mitigate the SWEET32 attack ([CVE-2016-2183]), 3DES cipher suites
4094   have been disabled by default and removed from DEFAULT, just like RC4.
4095   See the RC4 item below to re-enable both.
4096
4097   *Rich Salz*
4098
4099 * The method for finding the storage location for the Windows RAND seed file
4100   has changed. First we check %RANDFILE%. If that is not set then we check
4101   the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If
4102   all else fails we fall back to C:\.
4103
4104   *Matt Caswell*
4105
4106 * The EVP_EncryptUpdate() function has had its return type changed from void
4107   to int. A return of 0 indicates and error while a return of 1 indicates
4108   success.
4109
4110   *Matt Caswell*
4111
4112 * The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and
4113   DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch
4114   off the constant time implementation for RSA, DSA and DH have been made
4115   no-ops and deprecated.
4116
4117   *Matt Caswell*
4118
4119 * Windows RAND implementation was simplified to only get entropy by
4120   calling CryptGenRandom(). Various other RAND-related tickets
4121   were also closed.
4122
4123   *Joseph Wylie Yandle, Rich Salz*
4124
4125 * The stack and lhash API's were renamed to start with `OPENSSL_SK_`
4126   and `OPENSSL_LH_`, respectively.  The old names are available
4127   with API compatibility.  They new names are now completely documented.
4128
4129   *Rich Salz*
4130
4131 * Unify TYPE_up_ref(obj) methods signature.
4132   SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(),
4133   X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an
4134   int (instead of void) like all others TYPE_up_ref() methods.
4135   So now these methods also check the return value of CRYPTO_atomic_add(),
4136   and the validity of object reference counter.
4137
4138   *fdasilvayy@gmail.com*
4139
4140 * With Windows Visual Studio builds, the .pdb files are installed
4141   alongside the installed libraries and executables.  For a static
4142   library installation, ossl_static.pdb is the associate compiler
4143   generated .pdb file to be used when linking programs.
4144
4145   *Richard Levitte*
4146
4147 * Remove openssl.spec.  Packaging files belong with the packagers.
4148
4149   *Richard Levitte*
4150
4151 * Automatic Darwin/OSX configuration has had a refresh, it will now
4152   recognise x86_64 architectures automatically.  You can still decide
4153   to build for a different bitness with the environment variable
4154   KERNEL_BITS (can be 32 or 64), for example:
4155
4156           KERNEL_BITS=32 ./config
4157
4158   *Richard Levitte*
4159
4160 * Change default algorithms in pkcs8 utility to use PKCS#5 v2.0,
4161   256 bit AES and HMAC with SHA256.
4162
4163   *Steve Henson*
4164
4165 * Remove support for MIPS o32 ABI on IRIX (and IRIX only).
4166
4167   *Andy Polyakov*
4168
4169 * Triple-DES ciphers have been moved from HIGH to MEDIUM.
4170
4171   *Rich Salz*
4172
4173 * To enable users to have their own config files and build file templates,
4174   Configure looks in the directory indicated by the environment variable
4175   OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/
4176   directory.  On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical
4177   name and is used as is.
4178
4179   *Richard Levitte*
4180
4181 * The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX,
4182   X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD.  The unused type
4183   X509_CERT_FILE_CTX was removed.
4184
4185   *Rich Salz*
4186
4187 * "shared" builds are now the default. To create only static libraries use
4188   the "no-shared" Configure option.
4189
4190   *Matt Caswell*
4191
4192 * Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options.
4193   All of these option have not worked for some while and are fundamental
4194   algorithms.
4195
4196   *Matt Caswell*
4197
4198 * Make various cleanup routines no-ops and mark them as deprecated. Most
4199   global cleanup functions are no longer required because they are handled
4200   via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages).
4201   Explicitly de-initing can cause problems (e.g. where a library that uses
4202   OpenSSL de-inits, but an application is still using it). The affected
4203   functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(),
4204   EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(),
4205   RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and
4206   COMP_zlib_cleanup().
4207
4208   *Matt Caswell*
4209
4210 * --strict-warnings no longer enables runtime debugging options
4211   such as REF_DEBUG. Instead, debug options are automatically
4212   enabled with '--debug' builds.
4213
4214   *Andy Polyakov, Emilia Käsper*
4215
4216 * Made DH and DH_METHOD opaque. The structures for managing DH objects
4217   have been moved out of the public header files. New functions for managing
4218   these have been added.
4219
4220   *Matt Caswell*
4221
4222 * Made RSA and RSA_METHOD opaque. The structures for managing RSA
4223   objects have been moved out of the public header files. New
4224   functions for managing these have been added.
4225
4226   *Richard Levitte*
4227
4228 * Made DSA and DSA_METHOD opaque. The structures for managing DSA objects
4229   have been moved out of the public header files. New functions for managing
4230   these have been added.
4231
4232   *Matt Caswell*
4233
4234 * Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been
4235   moved out of the public header files. New functions for managing these
4236   have been added.
4237
4238   *Matt Caswell*
4239
4240 * Removed no-rijndael as a config option. Rijndael is an old name for AES.
4241
4242   *Matt Caswell*
4243
4244 * Removed the mk1mf build scripts.
4245
4246   *Richard Levitte*
4247
4248 * Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so
4249   it is always safe to #include a header now.
4250
4251   *Rich Salz*
4252
4253 * Removed the aged BC-32 config and all its supporting scripts
4254
4255   *Richard Levitte*
4256
4257 * Removed support for Ultrix, Netware, and OS/2.
4258
4259   *Rich Salz*
4260
4261 * Add support for HKDF.
4262
4263   *Alessandro Ghedini*
4264
4265 * Add support for blake2b and blake2s
4266
4267   *Bill Cox*
4268
4269 * Added support for "pipelining". Ciphers that have the
4270   EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple
4271   encryptions/decryptions simultaneously. There are currently no built-in
4272   ciphers with this property but the expectation is that engines will be able
4273   to offer it to significantly improve throughput. Support has been extended
4274   into libssl so that multiple records for a single connection can be
4275   processed in one go (for >=TLS 1.1).
4276
4277   *Matt Caswell*
4278
4279 * Added the AFALG engine. This is an async capable engine which is able to
4280   offload work to the Linux kernel. In this initial version it only supports
4281   AES128-CBC. The kernel must be version 4.1.0 or greater.
4282
4283   *Catriona Lucey*
4284
4285 * OpenSSL now uses a new threading API. It is no longer necessary to
4286   set locking callbacks to use OpenSSL in a multi-threaded environment. There
4287   are two supported threading models: pthreads and windows threads. It is
4288   also possible to configure OpenSSL at compile time for "no-threads". The
4289   old threading API should no longer be used. The functions have been
4290   replaced with "no-op" compatibility macros.
4291
4292   *Alessandro Ghedini, Matt Caswell*
4293
4294 * Modify behavior of ALPN to invoke callback after SNI/servername
4295   callback, such that updates to the SSL_CTX affect ALPN.
4296
4297   *Todd Short*
4298
4299 * Add SSL_CIPHER queries for authentication and key-exchange.
4300
4301   *Todd Short*
4302
4303 * Changes to the DEFAULT cipherlist:
4304   - Prefer (EC)DHE handshakes over plain RSA.
4305   - Prefer AEAD ciphers over legacy ciphers.
4306   - Prefer ECDSA over RSA when both certificates are available.
4307   - Prefer TLSv1.2 ciphers/PRF.
4308   - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the
4309     default cipherlist.
4310
4311   *Emilia Käsper*
4312
4313 * Change the ECC default curve list to be this, in order: x25519,
4314   secp256r1, secp521r1, secp384r1.
4315
4316   *Rich Salz*
4317
4318 * RC4 based libssl ciphersuites are now classed as "weak" ciphers and are
4319   disabled by default. They can be re-enabled using the
4320   enable-weak-ssl-ciphers option to Configure.
4321
4322   *Matt Caswell*
4323
4324 * If the server has ALPN configured, but supports no protocols that the
4325   client advertises, send a fatal "no_application_protocol" alert.
4326   This behaviour is SHALL in RFC 7301, though it isn't universally
4327   implemented by other servers.
4328
4329   *Emilia Käsper*
4330
4331 * Add X25519 support.
4332   Add ASN.1 and EVP_PKEY methods for X25519. This includes support
4333   for public and private key encoding using the format documented in
4334   draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports
4335   key generation and key derivation.
4336
4337   TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses
4338   X25519(29).
4339
4340   *Steve Henson*
4341
4342 * Deprecate SRP_VBASE_get_by_user.
4343   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
4344   In order to fix an unavoidable memory leak ([CVE-2016-0798]),
4345   SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP
4346   seed, even if the seed is configured.
4347
4348   Users should use SRP_VBASE_get1_by_user instead. Note that in
4349   SRP_VBASE_get1_by_user, caller must free the returned value. Note
4350   also that even though configuring the SRP seed attempts to hide
4351   invalid usernames by continuing the handshake with fake
4352   credentials, this behaviour is not constant time and no strong
4353   guarantees are made that the handshake is indistinguishable from
4354   that of a valid user.
4355
4356   *Emilia Käsper*
4357
4358 * Configuration change; it's now possible to build dynamic engines
4359   without having to build shared libraries and vice versa.  This
4360   only applies to the engines in `engines/`, those in `crypto/engine/`
4361   will always be built into libcrypto (i.e. "static").
4362
4363   Building dynamic engines is enabled by default; to disable, use
4364   the configuration option "disable-dynamic-engine".
4365
4366   The only requirements for building dynamic engines are the
4367   presence of the DSO module and building with position independent
4368   code, so they will also automatically be disabled if configuring
4369   with "disable-dso" or "disable-pic".
4370
4371   The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE
4372   are also taken away from openssl/opensslconf.h, as they are
4373   irrelevant.
4374
4375   *Richard Levitte*
4376
4377 * Configuration change; if there is a known flag to compile
4378   position independent code, it will always be applied on the
4379   libcrypto and libssl object files, and never on the application
4380   object files.  This means other libraries that use routines from
4381   libcrypto / libssl can be made into shared libraries regardless
4382   of how OpenSSL was configured.
4383
4384   If this isn't desirable, the configuration options "disable-pic"
4385   or "no-pic" can be used to disable the use of PIC.  This will
4386   also disable building shared libraries and dynamic engines.
4387
4388   *Richard Levitte*
4389
4390 * Removed JPAKE code.  It was experimental and has no wide use.
4391
4392   *Rich Salz*
4393
4394 * The INSTALL_PREFIX Makefile variable has been renamed to
4395   DESTDIR.  That makes for less confusion on what this variable
4396   is for.  Also, the configuration option --install_prefix is
4397   removed.
4398
4399   *Richard Levitte*
4400
4401 * Heartbeat for TLS has been removed and is disabled by default
4402   for DTLS; configure with enable-heartbeats.  Code that uses the
4403   old #define's might need to be updated.
4404
4405   *Emilia Käsper, Rich Salz*
4406
4407 * Rename REF_CHECK to REF_DEBUG.
4408
4409   *Rich Salz*
4410
4411 * New "unified" build system
4412
4413   The "unified" build system is aimed to be a common system for all
4414   platforms we support.  With it comes new support for VMS.
4415
4416   This system builds supports building in a different directory tree
4417   than the source tree.  It produces one Makefile (for unix family
4418   or lookalikes), or one descrip.mms (for VMS).
4419
4420   The source of information to make the Makefile / descrip.mms is
4421   small files called 'build.info', holding the necessary
4422   information for each directory with source to compile, and a
4423   template in Configurations, like unix-Makefile.tmpl or
4424   descrip.mms.tmpl.
4425
4426   With this change, the library names were also renamed on Windows
4427   and on VMS.  They now have names that are closer to the standard
4428   on Unix, and include the major version number, and in certain
4429   cases, the architecture they are built for.  See "Notes on shared
4430   libraries" in INSTALL.
4431
4432   We rely heavily on the perl module Text::Template.
4433
4434   *Richard Levitte*
4435
4436 * Added support for auto-initialisation and de-initialisation of the library.
4437   OpenSSL no longer requires explicit init or deinit routines to be called,
4438   except in certain circumstances. See the OPENSSL_init_crypto() and
4439   OPENSSL_init_ssl() man pages for further information.
4440
4441   *Matt Caswell*
4442
4443 * The arguments to the DTLSv1_listen function have changed. Specifically the
4444   "peer" argument is now expected to be a BIO_ADDR object.
4445
4446 * Rewrite of BIO networking library. The BIO library lacked consistent
4447   support of IPv6, and adding it required some more extensive
4448   modifications.  This introduces the BIO_ADDR and BIO_ADDRINFO types,
4449   which hold all types of addresses and chains of address information.
4450   It also introduces a new API, with functions like BIO_socket,
4451   BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept.
4452   The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram
4453   have been adapted accordingly.
4454
4455   *Richard Levitte*
4456
4457 * RSA_padding_check_PKCS1_type_1 now accepts inputs with and without
4458   the leading 0-byte.
4459
4460   *Emilia Käsper*
4461
4462 * CRIME protection: disable compression by default, even if OpenSSL is
4463   compiled with zlib enabled. Applications can still enable compression
4464   by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by
4465   using the SSL_CONF library to configure compression.
4466
4467   *Emilia Käsper*
4468
4469 * The signature of the session callback configured with
4470   SSL_CTX_sess_set_get_cb was changed. The read-only input buffer
4471   was explicitly marked as `const unsigned char*` instead of
4472   `unsigned char*`.
4473
4474   *Emilia Käsper*
4475
4476 * Always DPURIFY. Remove the use of uninitialized memory in the
4477   RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op.
4478
4479   *Emilia Käsper*
4480
4481 * Removed many obsolete configuration items, including
4482      DES_PTR, DES_RISC1, DES_RISC2, DES_INT
4483      MD2_CHAR, MD2_INT, MD2_LONG
4484      BF_PTR, BF_PTR2
4485      IDEA_SHORT, IDEA_LONG
4486      RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX
4487
4488   *Rich Salz, with advice from Andy Polyakov*
4489
4490 * Many BN internals have been moved to an internal header file.
4491
4492   *Rich Salz with help from Andy Polyakov*
4493
4494 * Configuration and writing out the results from it has changed.
4495   Files such as Makefile include/openssl/opensslconf.h and are now
4496   produced through general templates, such as Makefile.in and
4497   crypto/opensslconf.h.in and some help from the perl module
4498   Text::Template.
4499
4500   Also, the center of configuration information is no longer
4501   Makefile.  Instead, Configure produces a perl module in
4502   configdata.pm which holds most of the config data (in the hash
4503   table %config), the target data that comes from the target
4504   configuration in one of the `Configurations/*.conf` files (in
4505   %target).
4506
4507   *Richard Levitte*
4508
4509 * To clarify their intended purposes, the Configure options
4510   --prefix and --openssldir change their semantics, and become more
4511   straightforward and less interdependent.
4512
4513   --prefix shall be used exclusively to give the location INSTALLTOP
4514   where programs, scripts, libraries, include files and manuals are
4515   going to be installed.  The default is now /usr/local.
4516
4517   --openssldir shall be used exclusively to give the default
4518   location OPENSSLDIR where certificates, private keys, CRLs are
4519   managed.  This is also where the default openssl.cnf gets
4520   installed.
4521   If the directory given with this option is a relative path, the
4522   values of both the --prefix value and the --openssldir value will
4523   be combined to become OPENSSLDIR.
4524   The default for --openssldir is INSTALLTOP/ssl.
4525
4526   Anyone who uses --openssldir to specify where OpenSSL is to be
4527   installed MUST change to use --prefix instead.
4528
4529   *Richard Levitte*
4530
4531 * The GOST engine was out of date and therefore it has been removed. An up
4532   to date GOST engine is now being maintained in an external repository.
4533   See: <https://wiki.openssl.org/index.php/Binaries>. Libssl still retains
4534   support for GOST ciphersuites (these are only activated if a GOST engine
4535   is present).
4536
4537   *Matt Caswell*
4538
4539 * EGD is no longer supported by default; use enable-egd when
4540   configuring.
4541
4542   *Ben Kaduk and Rich Salz*
4543
4544 * The distribution now has Makefile.in files, which are used to
4545   create Makefile's when Configure is run.  *Configure must be run
4546   before trying to build now.*
4547
4548   *Rich Salz*
4549
4550 * The return value for SSL_CIPHER_description() for error conditions
4551   has changed.
4552
4553   *Rich Salz*
4554
4555 * Support for RFC6698/RFC7671 DANE TLSA peer authentication.
4556
4557   Obtaining and performing DNSSEC validation of TLSA records is
4558   the application's responsibility.  The application provides
4559   the TLSA records of its choice to OpenSSL, and these are then
4560   used to authenticate the peer.
4561
4562   The TLSA records need not even come from DNS.  They can, for
4563   example, be used to implement local end-entity certificate or
4564   trust-anchor "pinning", where the "pin" data takes the form
4565   of TLSA records, which can augment or replace verification
4566   based on the usual WebPKI public certification authorities.
4567
4568   *Viktor Dukhovni*
4569
4570 * Revert default OPENSSL_NO_DEPRECATED setting.  Instead OpenSSL
4571   continues to support deprecated interfaces in default builds.
4572   However, applications are strongly advised to compile their
4573   source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides
4574   the declarations of all interfaces deprecated in 0.9.8, 1.0.0
4575   or the 1.1.0 releases.
4576
4577   In environments in which all applications have been ported to
4578   not use any deprecated interfaces OpenSSL's Configure script
4579   should be used with the --api=1.1.0 option to entirely remove
4580   support for the deprecated features from the library and
4581   unconditionally disable them in the installed headers.
4582   Essentially the same effect can be achieved with the "no-deprecated"
4583   argument to Configure, except that this will always restrict
4584   the build to just the latest API, rather than a fixed API
4585   version.
4586
4587   As applications are ported to future revisions of the API,
4588   they should update their compile-time OPENSSL_API_COMPAT define
4589   accordingly, but in most cases should be able to continue to
4590   compile with later releases.
4591
4592   The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are
4593   0x10000000L and 0x00908000L, respectively.  However those
4594   versions did not support the OPENSSL_API_COMPAT feature, and
4595   so applications are not typically tested for explicit support
4596   of just the undeprecated features of either release.
4597
4598   *Viktor Dukhovni*
4599
4600 * Add support for setting the minimum and maximum supported protocol.
4601   It can bet set via the SSL_set_min_proto_version() and
4602   SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and
4603   MaxProtocol.  It's recommended to use the new APIs to disable
4604   protocols instead of disabling individual protocols using
4605   SSL_set_options() or SSL_CONF's Protocol.  This change also
4606   removes support for disabling TLS 1.2 in the OpenSSL TLS
4607   client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT.
4608
4609   *Kurt Roeckx*
4610
4611 * Support for ChaCha20 and Poly1305 added to libcrypto and libssl.
4612
4613   *Andy Polyakov*
4614
4615 * New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD
4616   and integrates ECDSA and ECDH functionality into EC. Implementations can
4617   now redirect key generation and no longer need to convert to or from
4618   ECDSA_SIG format.
4619
4620   Note: the ecdsa.h and ecdh.h headers are now no longer needed and just
4621   include the ec.h header file instead.
4622
4623   *Steve Henson*
4624
4625 * Remove support for all 40 and 56 bit ciphers.  This includes all the export
4626   ciphers who are no longer supported and drops support the ephemeral RSA key
4627   exchange. The LOW ciphers currently doesn't have any ciphers in it.
4628
4629   *Kurt Roeckx*
4630
4631 * Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX
4632   opaque.  For HMAC_CTX, the following constructors and destructors
4633   were added:
4634
4635       HMAC_CTX *HMAC_CTX_new(void);
4636       void HMAC_CTX_free(HMAC_CTX *ctx);
4637
4638   For EVP_MD and EVP_CIPHER, complete APIs to create, fill and
4639   destroy such methods has been added.  See EVP_MD_meth_new(3) and
4640   EVP_CIPHER_meth_new(3) for documentation.
4641
4642   Additional changes:
4643   1) `EVP_MD_CTX_cleanup()`, `EVP_CIPHER_CTX_cleanup()` and
4644      `HMAC_CTX_cleanup()` were removed. `HMAC_CTX_reset()` and
4645      `EVP_MD_CTX_reset()` should be called instead to reinitialise
4646      an already created structure.
4647   2) For consistency with the majority of our object creators and
4648      destructors, `EVP_MD_CTX_(create|destroy)` were renamed to
4649      `EVP_MD_CTX_(new|free)`.  The old names are retained as macros
4650      for deprecated builds.
4651
4652   *Richard Levitte*
4653
4654 * Added ASYNC support. Libcrypto now includes the async sub-library to enable
4655   cryptographic operations to be performed asynchronously as long as an
4656   asynchronous capable engine is used. See the ASYNC_start_job() man page for
4657   further details. Libssl has also had this capability integrated with the
4658   introduction of the new mode SSL_MODE_ASYNC and associated error
4659   SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man
4660   pages. This work was developed in partnership with Intel Corp.
4661
4662   *Matt Caswell*
4663
4664 * SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is
4665   always enabled now.  If you want to disable the support you should
4666   exclude it using the list of supported ciphers. This also means that the
4667   "-no_ecdhe" option has been removed from s_server.
4668
4669   *Kurt Roeckx*
4670
4671 * SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls
4672   SSL_{CTX_}set1_curves() which can set a list.
4673
4674   *Kurt Roeckx*
4675
4676 * Remove support for SSL_{CTX_}set_tmp_ecdh_callback().  You should set the
4677   curve you want to support using SSL_{CTX_}set1_curves().
4678
4679   *Kurt Roeckx*
4680
4681 * State machine rewrite. The state machine code has been significantly
4682   refactored in order to remove much duplication of code and solve issues
4683   with the old code (see [ssl/statem/README.md](ssl/statem/README.md) for
4684   further details). This change does have some associated API changes.
4685   Notably the SSL_state() function has been removed and replaced by
4686   SSL_get_state which now returns an "OSSL_HANDSHAKE_STATE" instead of an int.
4687   SSL_set_state() has been removed altogether. The previous handshake states
4688   defined in ssl.h and ssl3.h have also been removed.
4689
4690   *Matt Caswell*
4691
4692 * All instances of the string "ssleay" in the public API were replaced
4693   with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's)
4694   Some error codes related to internal RSA_eay API's were renamed.
4695
4696   *Rich Salz*
4697
4698 * The demo files in crypto/threads were moved to demo/threads.
4699
4700   *Rich Salz*
4701
4702 * Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp,
4703   sureware and ubsec.
4704
4705   *Matt Caswell, Rich Salz*
4706
4707 * New ASN.1 embed macro.
4708
4709   New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the
4710   structure is not allocated: it is part of the parent. That is instead of
4711
4712           FOO *x;
4713
4714   it must be:
4715
4716           FOO x;
4717
4718   This reduces memory fragmentation and make it impossible to accidentally
4719   set a mandatory field to NULL.
4720
4721   This currently only works for some fields specifically a SEQUENCE, CHOICE,
4722   or ASN1_STRING type which is part of a parent SEQUENCE. Since it is
4723   equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or
4724   SEQUENCE OF.
4725
4726   *Steve Henson*
4727
4728 * Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled.
4729
4730   *Emilia Käsper*
4731
4732 * Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although
4733   in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also
4734   an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add
4735   DES and RC4 ciphersuites.
4736
4737   *Matt Caswell*
4738
4739 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
4740   This changes the decoding behaviour for some invalid messages,
4741   though the change is mostly in the more lenient direction, and
4742   legacy behaviour is preserved as much as possible.
4743
4744   *Emilia Käsper*
4745
4746 * Fix no-stdio build.
4747   *David Woodhouse <David.Woodhouse@intel.com> and also*
4748   *Ivan Nestlerode <ivan.nestlerode@sonos.com>*
4749
4750 * New testing framework
4751   The testing framework has been largely rewritten and is now using
4752   perl and the perl modules Test::Harness and an extended variant of
4753   Test::More called OpenSSL::Test to do its work.  All test scripts in
4754   test/ have been rewritten into test recipes, and all direct calls to
4755   executables in test/Makefile have become individual recipes using the
4756   simplified testing OpenSSL::Test::Simple.
4757
4758   For documentation on our testing modules, do:
4759
4760           perldoc test/testlib/OpenSSL/Test/Simple.pm
4761           perldoc test/testlib/OpenSSL/Test.pm
4762
4763   *Richard Levitte*
4764
4765 * Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT
4766   are used; the latter aborts on memory leaks (usually checked on exit).
4767   Some undocumented "set malloc, etc., hooks" functions were removed
4768   and others were changed.  All are now documented.
4769
4770   *Rich Salz*
4771
4772 * In DSA_generate_parameters_ex, if the provided seed is too short,
4773   return an error
4774
4775   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
4776
4777 * Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites
4778   from RFC4279, RFC4785, RFC5487, RFC5489.
4779
4780   Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the
4781   original RSA_PSK patch.
4782
4783   *Steve Henson*
4784
4785 * Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay
4786   era flag was never set throughout the codebase (only read). Also removed
4787   SSL3_FLAGS_POP_BUFFER which was only used if
4788   SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set.
4789
4790   *Matt Caswell*
4791
4792 * Changed the default name options in the "ca", "crl", "req" and "x509"
4793   to be "oneline" instead of "compat".
4794
4795   *Richard Levitte*
4796
4797 * Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're
4798   not aware of clients that still exhibit this bug, and the workaround
4799   hasn't been working properly for a while.
4800
4801   *Emilia Käsper*
4802
4803 * The return type of BIO_number_read() and BIO_number_written() as well as
4804   the corresponding num_read and num_write members in the BIO structure has
4805   changed from unsigned long to uint64_t. On platforms where an unsigned
4806   long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is
4807   transferred.
4808
4809   *Matt Caswell*
4810
4811 * Given the pervasive nature of TLS extensions it is inadvisable to run
4812   OpenSSL without support for them. It also means that maintaining
4813   the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably
4814   not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed.
4815
4816   *Matt Caswell*
4817
4818 * Removed support for the two export grade static DH ciphersuites
4819   EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites
4820   were newly added (along with a number of other static DH ciphersuites) to
4821   1.0.2. However the two export ones have *never* worked since they were
4822   introduced. It seems strange in any case to be adding new export
4823   ciphersuites, and given "logjam" it also does not seem correct to fix them.
4824
4825   *Matt Caswell*
4826
4827 * Version negotiation has been rewritten. In particular SSLv23_method(),
4828   SSLv23_client_method() and SSLv23_server_method() have been deprecated,
4829   and turned into macros which simply call the new preferred function names
4830   TLS_method(), TLS_client_method() and TLS_server_method(). All new code
4831   should use the new names instead. Also as part of this change the ssl23.h
4832   header file has been removed.
4833
4834   *Matt Caswell*
4835
4836 * Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This
4837   code and the associated standard is no longer considered fit-for-purpose.
4838
4839   *Matt Caswell*
4840
4841 * RT2547 was closed.  When generating a private key, try to make the
4842   output file readable only by the owner.  This behavior change might
4843   be noticeable when interacting with other software.
4844
4845 * Documented all exdata functions.  Added CRYPTO_free_ex_index.
4846   Added a test.
4847
4848   *Rich Salz*
4849
4850 * Added HTTP GET support to the ocsp command.
4851
4852   *Rich Salz*
4853
4854 * Changed default digest for the dgst and enc commands from MD5 to
4855   sha256
4856
4857   *Rich Salz*
4858
4859 * RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead.
4860
4861   *Matt Caswell*
4862
4863 * Added support for TLS extended master secret from
4864   draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an
4865   initial patch which was a great help during development.
4866
4867   *Steve Henson*
4868
4869 * All libssl internal structures have been removed from the public header
4870   files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is
4871   now redundant). Users should not attempt to access internal structures
4872   directly. Instead they should use the provided API functions.
4873
4874   *Matt Caswell*
4875
4876 * config has been changed so that by default OPENSSL_NO_DEPRECATED is used.
4877   Access to deprecated functions can be re-enabled by running config with
4878   "enable-deprecated". In addition applications wishing to use deprecated
4879   functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour
4880   will, by default, disable some transitive includes that previously existed
4881   in the header files (e.g. ec.h will no longer, by default, include bn.h)
4882
4883   *Matt Caswell*
4884
4885 * Added support for OCB mode. OpenSSL has been granted a patent license
4886   compatible with the OpenSSL license for use of OCB. Details are available
4887   at <https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf>. Support
4888   for OCB can be removed by calling config with no-ocb.
4889
4890   *Matt Caswell*
4891
4892 * SSLv2 support has been removed.  It still supports receiving a SSLv2
4893   compatible client hello.
4894
4895   *Kurt Roeckx*
4896
4897 * Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz],
4898   done while fixing the error code for the key-too-small case.
4899
4900   *Annie Yousar <a.yousar@informatik.hu-berlin.de>*
4901
4902 * CA.sh has been removed; use CA.pl instead.
4903
4904   *Rich Salz*
4905
4906 * Removed old DES API.
4907
4908   *Rich Salz*
4909
4910 * Remove various unsupported platforms:
4911      Sony NEWS4
4912      BEOS and BEOS_R5
4913      NeXT
4914      SUNOS
4915      MPE/iX
4916      Sinix/ReliantUNIX RM400
4917      DGUX
4918      NCR
4919      Tandem
4920      Cray
4921      16-bit platforms such as WIN16
4922
4923   *Rich Salz*
4924
4925 * Clean up OPENSSL_NO_xxx #define's
4926   - Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF
4927   - Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx
4928   - OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC
4929   - OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160
4930   - OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO
4931   - Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY
4932     OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP
4933     OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK
4934     OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY
4935   - Remove MS_STATIC; it's a relic from platforms <32 bits.
4936
4937   *Rich Salz*
4938
4939 * Cleaned up dead code
4940     Remove all but one '#ifdef undef' which is to be looked at.
4941
4942   *Rich Salz*
4943
4944 * Clean up calling of xxx_free routines.
4945      Just like free(), fix most of the xxx_free routines to accept
4946      NULL.  Remove the non-null checks from callers.  Save much code.
4947
4948   *Rich Salz*
4949
4950 * Add secure heap for storage of private keys (when possible).
4951   Add BIO_s_secmem(), CBIGNUM, etc.
4952   Contributed by Akamai Technologies under our Corporate CLA.
4953
4954   *Rich Salz*
4955
4956 * Experimental support for a new, fast, unbiased prime candidate generator,
4957   bn_probable_prime_dh_coprime(). Not currently used by any prime generator.
4958
4959   *Felix Laurie von Massenbach <felix@erbridge.co.uk>*
4960
4961 * New output format NSS in the sess_id command line tool. This allows
4962   exporting the session id and the master key in NSS keylog format.
4963
4964   *Martin Kaiser <martin@kaiser.cx>*
4965
4966 * Harmonize version and its documentation. -f flag is used to display
4967   compilation flags.
4968
4969   *mancha <mancha1@zoho.com>*
4970
4971 * Fix eckey_priv_encode so it immediately returns an error upon a failure
4972   in i2d_ECPrivateKey.  Thanks to Ted Unangst for feedback on this issue.
4973
4974   *mancha <mancha1@zoho.com>*
4975
4976 * Fix some double frees. These are not thought to be exploitable.
4977
4978   *mancha <mancha1@zoho.com>*
4979
4980 * A missing bounds check in the handling of the TLS heartbeat extension
4981   can be used to reveal up to 64k of memory to a connected client or
4982   server.
4983
4984   Thanks for Neel Mehta of Google Security for discovering this bug and to
4985   Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
4986   preparing the fix ([CVE-2014-0160])
4987
4988   *Adam Langley, Bodo Moeller*
4989
4990 * Fix for the attack described in the paper "Recovering OpenSSL
4991   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
4992   by Yuval Yarom and Naomi Benger. Details can be obtained from:
4993   <http://eprint.iacr.org/2014/140>
4994
4995   Thanks to Yuval Yarom and Naomi Benger for discovering this
4996   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
4997
4998   *Yuval Yarom and Naomi Benger*
4999
5000 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
5001   this fixes a limitation in previous versions of OpenSSL.
5002
5003   *Steve Henson*
5004
5005 * Experimental encrypt-then-mac support.
5006
5007   Experimental support for encrypt then mac from
5008   draft-gutmann-tls-encrypt-then-mac-02.txt
5009
5010   To enable it set the appropriate extension number (0x42 for the test
5011   server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42
5012
5013   For non-compliant peers (i.e. just about everything) this should have no
5014   effect.
5015
5016   WARNING: EXPERIMENTAL, SUBJECT TO CHANGE.
5017
5018   *Steve Henson*
5019
5020 * Add EVP support for key wrapping algorithms, to avoid problems with
5021   existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
5022   the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
5023   algorithms and include tests cases.
5024
5025   *Steve Henson*
5026
5027 * Extend CMS code to support RSA-PSS signatures and RSA-OAEP for
5028   enveloped data.
5029
5030   *Steve Henson*
5031
5032 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
5033   MGF1 digest and OAEP label.
5034
5035   *Steve Henson*
5036
5037 * Make openssl verify return errors.
5038
5039   *Chris Palmer <palmer@google.com> and Ben Laurie*
5040
5041 * New function ASN1_TIME_diff to calculate the difference between two
5042   ASN1_TIME structures or one structure and the current time.
5043
5044   *Steve Henson*
5045
5046 * Update fips_test_suite to support multiple command line options. New
5047   test to induce all self test errors in sequence and check expected
5048   failures.
5049
5050   *Steve Henson*
5051
5052 * Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and
5053   sign or verify all in one operation.
5054
5055   *Steve Henson*
5056
5057 * Add fips_algvs: a multicall fips utility incorporating all the algorithm
5058   test programs and fips_test_suite. Includes functionality to parse
5059   the minimal script output of fipsalgest.pl directly.
5060
5061   *Steve Henson*
5062
5063 * Add authorisation parameter to FIPS_module_mode_set().
5064
5065   *Steve Henson*
5066
5067 * Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves.
5068
5069   *Steve Henson*
5070
5071 * Use separate DRBG fields for internal and external flags. New function
5072   FIPS_drbg_health_check() to perform on demand health checking. Add
5073   generation tests to fips_test_suite with reduced health check interval to
5074   demonstrate periodic health checking. Add "nodh" option to
5075   fips_test_suite to skip very slow DH test.
5076
5077   *Steve Henson*
5078
5079 * New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers
5080   based on NID.
5081
5082   *Steve Henson*
5083
5084 * More extensive health check for DRBG checking many more failure modes.
5085   New function FIPS_selftest_drbg_all() to handle every possible DRBG
5086   combination: call this in fips_test_suite.
5087
5088   *Steve Henson*
5089
5090 * Add support for canonical generation of DSA parameter 'g'. See
5091   FIPS 186-3 A.2.3.
5092
5093 * Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and
5094   POST to handle HMAC cases.
5095
5096   *Steve Henson*
5097
5098 * Add functions FIPS_module_version() and FIPS_module_version_text()
5099   to return numerical and string versions of the FIPS module number.
5100
5101   *Steve Henson*
5102
5103 * Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and
5104   FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented
5105   outside the validated module in the FIPS capable OpenSSL.
5106
5107   *Steve Henson*
5108
5109 * Minor change to DRBG entropy callback semantics. In some cases
5110   there is no multiple of the block length between min_len and
5111   max_len. Allow the callback to return more than max_len bytes
5112   of entropy but discard any extra: it is the callback's responsibility
5113   to ensure that the extra data discarded does not impact the
5114   requested amount of entropy.
5115
5116   *Steve Henson*
5117
5118 * Add PRNG security strength checks to RSA, DSA and ECDSA using
5119   information in FIPS186-3, SP800-57 and SP800-131A.
5120
5121   *Steve Henson*
5122
5123 * CCM support via EVP. Interface is very similar to GCM case except we
5124   must supply all data in one chunk (i.e. no update, final) and the
5125   message length must be supplied if AAD is used. Add algorithm test
5126   support.
5127
5128   *Steve Henson*
5129
5130 * Initial version of POST overhaul. Add POST callback to allow the status
5131   of POST to be monitored and/or failures induced. Modify fips_test_suite
5132   to use callback. Always run all selftests even if one fails.
5133
5134   *Steve Henson*
5135
5136 * XTS support including algorithm test driver in the fips_gcmtest program.
5137   Note: this does increase the maximum key length from 32 to 64 bytes but
5138   there should be no binary compatibility issues as existing applications
5139   will never use XTS mode.
5140
5141   *Steve Henson*
5142
5143 * Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies
5144   to OpenSSL RAND code and replace with a tiny FIPS RAND API which also
5145   performs algorithm blocking for unapproved PRNG types. Also do not
5146   set PRNG type in FIPS_mode_set(): leave this to the application.
5147   Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with
5148   the standard OpenSSL PRNG: set additional data to a date time vector.
5149
5150   *Steve Henson*
5151
5152 * Rename old X9.31 PRNG functions of the form `FIPS_rand*` to `FIPS_x931*`.
5153   This shouldn't present any incompatibility problems because applications
5154   shouldn't be using these directly and any that are will need to rethink
5155   anyway as the X9.31 PRNG is now deprecated by FIPS 140-2
5156
5157   *Steve Henson*
5158
5159 * Extensive self tests and health checking required by SP800-90 DRBG.
5160   Remove strength parameter from FIPS_drbg_instantiate and always
5161   instantiate at maximum supported strength.
5162
5163   *Steve Henson*
5164
5165 * Add ECDH code to fips module and fips_ecdhvs for primitives only testing.
5166
5167   *Steve Henson*
5168
5169 * New algorithm test program fips_dhvs to handle DH primitives only testing.
5170
5171   *Steve Henson*
5172
5173 * New function DH_compute_key_padded() to compute a DH key and pad with
5174   leading zeroes if needed: this complies with SP800-56A et al.
5175
5176   *Steve Henson*
5177
5178 * Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by
5179   anything, incomplete, subject to change and largely untested at present.
5180
5181   *Steve Henson*
5182
5183 * Modify fipscanisteronly build option to only build the necessary object
5184   files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile.
5185
5186   *Steve Henson*
5187
5188 * Add experimental option FIPSSYMS to give all symbols in
5189   fipscanister.o and FIPS or fips prefix. This will avoid
5190   conflicts with future versions of OpenSSL. Add perl script
5191   util/fipsas.pl to preprocess assembly language source files
5192   and rename any affected symbols.
5193
5194   *Steve Henson*
5195
5196 * Add selftest checks and algorithm block of non-fips algorithms in
5197   FIPS mode. Remove DES2 from selftests.
5198
5199   *Steve Henson*
5200
5201 * Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just
5202   return internal method without any ENGINE dependencies. Add new
5203   tiny fips sign and verify functions.
5204
5205   *Steve Henson*
5206
5207 * New build option no-ec2m to disable characteristic 2 code.
5208
5209   *Steve Henson*
5210
5211 * New build option "fipscanisteronly". This only builds fipscanister.o
5212   and (currently) associated fips utilities. Uses the file Makefile.fips
5213   instead of Makefile.org as the prototype.
5214
5215   *Steve Henson*
5216
5217 * Add some FIPS mode restrictions to GCM. Add internal IV generator.
5218   Update fips_gcmtest to use IV generator.
5219
5220   *Steve Henson*
5221
5222 * Initial, experimental EVP support for AES-GCM. AAD can be input by
5223   setting output buffer to NULL. The `*Final` function must be
5224   called although it will not retrieve any additional data. The tag
5225   can be set or retrieved with a ctrl. The IV length is by default 12
5226   bytes (96 bits) but can be set to an alternative value. If the IV
5227   length exceeds the maximum IV length (currently 16 bytes) it cannot be
5228   set before the key.
5229
5230   *Steve Henson*
5231
5232 * New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the
5233   underlying do_cipher function handles all cipher semantics itself
5234   including padding and finalisation. This is useful if (for example)
5235   an ENGINE cipher handles block padding itself. The behaviour of
5236   do_cipher is subtly changed if this flag is set: the return value
5237   is the number of characters written to the output buffer (zero is
5238   no longer an error code) or a negative error code. Also if the
5239   input buffer is NULL and length 0 finalisation should be performed.
5240
5241   *Steve Henson*
5242
5243 * If a candidate issuer certificate is already part of the constructed
5244   path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case.
5245
5246   *Steve Henson*
5247
5248 * Improve forward-security support: add functions
5249
5250           void SSL_CTX_set_not_resumable_session_callback(
5251                    SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure))
5252           void SSL_set_not_resumable_session_callback(
5253                    SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure))
5254
5255   for use by SSL/TLS servers; the callback function will be called whenever a
5256   new session is created, and gets to decide whether the session may be
5257   cached to make it resumable (return 0) or not (return 1).  (As by the
5258   SSL/TLS protocol specifications, the session_id sent by the server will be
5259   empty to indicate that the session is not resumable; also, the server will
5260   not generate RFC 4507 (RFC 5077) session tickets.)
5261
5262   A simple reasonable callback implementation is to return is_forward_secure.
5263   This parameter will be set to 1 or 0 depending on the ciphersuite selected
5264   by the SSL/TLS server library, indicating whether it can provide forward
5265   security.
5266
5267   *Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)*
5268
5269 * New -verify_name option in command line utilities to set verification
5270   parameters by name.
5271
5272   *Steve Henson*
5273
5274 * Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE.
5275   Add CMAC pkey methods.
5276
5277   *Steve Henson*
5278
5279 * Experimental renegotiation in s_server -www mode. If the client
5280   browses /reneg connection is renegotiated. If /renegcert it is
5281   renegotiated requesting a certificate.
5282
5283   *Steve Henson*
5284
5285 * Add an "external" session cache for debugging purposes to s_server. This
5286   should help trace issues which normally are only apparent in deployed
5287   multi-process servers.
5288
5289   *Steve Henson*
5290
5291 * Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where
5292   return value is ignored. NB. The functions RAND_add(), RAND_seed(),
5293   BIO_set_cipher() and some obscure PEM functions were changed so they
5294   can now return an error. The RAND changes required a change to the
5295   RAND_METHOD structure.
5296
5297   *Steve Henson*
5298
5299 * New macro `__owur` for "OpenSSL Warn Unused Result". This makes use of
5300   a gcc attribute to warn if the result of a function is ignored. This
5301   is enable if DEBUG_UNUSED is set. Add to several functions in evp.h
5302   whose return value is often ignored.
5303
5304   *Steve Henson*
5305
5306 * New -noct, -requestct, -requirect and -ctlogfile options for s_client.
5307   These allow SCTs (signed certificate timestamps) to be requested and
5308   validated when establishing a connection.
5309
5310   *Rob Percival <robpercival@google.com>*
5311
5312OpenSSL 1.0.2
5313-------------
5314
5315### Changes between 1.0.2s and 1.0.2t [10 Sep 2019]
5316
5317 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
5318   used even when parsing explicit parameters, when loading a encoded key
5319   or calling `EC_GROUP_new_from_ecpkparameters()`/
5320   `EC_GROUP_new_from_ecparameters()`.
5321   This prevents bypass of security hardening and performance gains,
5322   especially for curves with specialized EC_METHODs.
5323   By default, if a key encoded with explicit parameters is loaded and later
5324   encoded, the output is still encoded with explicit parameters, even if
5325   internally a "named" EC_GROUP is used for computation.
5326
5327   *Nicola Tuveri*
5328
5329 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
5330   this change, EC_GROUP_set_generator would accept order and/or cofactor as
5331   NULL. After this change, only the cofactor parameter can be NULL. It also
5332   does some minimal sanity checks on the passed order.
5333   ([CVE-2019-1547])
5334
5335   *Billy Bob Brumley*
5336
5337 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
5338   An attack is simple, if the first CMS_recipientInfo is valid but the
5339   second CMS_recipientInfo is chosen ciphertext. If the second
5340   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
5341   encryption key will be replaced by garbage, and the message cannot be
5342   decoded, but if the RSA decryption fails, the correct encryption key is
5343   used and the recipient will not notice the attack.
5344   As a work around for this potential attack the length of the decrypted
5345   key must be equal to the cipher default key length, in case the
5346   certifiate is not given and all recipientInfo are tried out.
5347   The old behaviour can be re-enabled in the CMS code by setting the
5348   CMS_DEBUG_DECRYPT flag.
5349   ([CVE-2019-1563])
5350
5351   *Bernd Edlinger*
5352
5353 * Document issue with installation paths in diverse Windows builds
5354
5355   '/usr/local/ssl' is an unsafe prefix for location to install OpenSSL
5356   binaries and run-time config file.
5357   ([CVE-2019-1552])
5358
5359   *Richard Levitte*
5360
5361### Changes between 1.0.2r and 1.0.2s [28 May 2019]
5362
5363 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
5364   This changes the size when using the `genpkey` command when no size is given.
5365   It fixes an omission in earlier changes that changed all RSA, DSA and DH
5366   generation commands to use 2048 bits by default.
5367
5368   *Kurt Roeckx*
5369
5370 * Add FIPS support for Android Arm 64-bit
5371
5372   Support for Android Arm 64-bit was added to the OpenSSL FIPS Object
5373   Module in Version 2.0.10. For some reason, the corresponding target
5374   'android64-aarch64' was missing OpenSSL 1.0.2, whence it could not be
5375   built with FIPS support on Android Arm 64-bit. This omission has been
5376   fixed.
5377
5378   *Matthias St. Pierre*
5379
5380### Changes between 1.0.2q and 1.0.2r [26 Feb 2019]
5381
5382 * 0-byte record padding oracle
5383
5384   If an application encounters a fatal protocol error and then calls
5385   SSL_shutdown() twice (once to send a close_notify, and once to receive one)
5386   then OpenSSL can respond differently to the calling application if a 0 byte
5387   record is received with invalid padding compared to if a 0 byte record is
5388   received with an invalid MAC. If the application then behaves differently
5389   based on that in a way that is detectable to the remote peer, then this
5390   amounts to a padding oracle that could be used to decrypt data.
5391
5392   In order for this to be exploitable "non-stitched" ciphersuites must be in
5393   use. Stitched ciphersuites are optimised implementations of certain
5394   commonly used ciphersuites. Also the application must call SSL_shutdown()
5395   twice even if a protocol error has occurred (applications should not do
5396   this but some do anyway).
5397
5398   This issue was discovered by Juraj Somorovsky, Robert Merget and Nimrod
5399   Aviram, with additional investigation by Steven Collison and Andrew
5400   Hourselt. It was reported to OpenSSL on 10th December 2018.
5401   ([CVE-2019-1559])
5402
5403   *Matt Caswell*
5404
5405 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
5406
5407   *Richard Levitte*
5408
5409### Changes between 1.0.2p and 1.0.2q [20 Nov 2018]
5410
5411 * Microarchitecture timing vulnerability in ECC scalar multiplication
5412
5413   OpenSSL ECC scalar multiplication, used in e.g. ECDSA and ECDH, has been
5414   shown to be vulnerable to a microarchitecture timing side channel attack.
5415   An attacker with sufficient access to mount local timing attacks during
5416   ECDSA signature generation could recover the private key.
5417
5418   This issue was reported to OpenSSL on 26th October 2018 by Alejandro
5419   Cabrera Aldaya, Billy Brumley, Sohaib ul Hassan, Cesar Pereida Garcia and
5420   Nicola Tuveri.
5421   ([CVE-2018-5407])
5422
5423   *Billy Brumley*
5424
5425 * Timing vulnerability in DSA signature generation
5426
5427   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
5428   timing side channel attack. An attacker could use variations in the signing
5429   algorithm to recover the private key.
5430
5431   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
5432   ([CVE-2018-0734])
5433
5434   *Paul Dale*
5435
5436 * Resolve a compatibility issue in EC_GROUP handling with the FIPS Object
5437   Module, accidentally introduced while backporting security fixes from the
5438   development branch and hindering the use of ECC in FIPS mode.
5439
5440   *Nicola Tuveri*
5441
5442### Changes between 1.0.2o and 1.0.2p [14 Aug 2018]
5443
5444 * Client DoS due to large DH parameter
5445
5446   During key agreement in a TLS handshake using a DH(E) based ciphersuite a
5447   malicious server can send a very large prime value to the client. This will
5448   cause the client to spend an unreasonably long period of time generating a
5449   key for this prime resulting in a hang until the client has finished. This
5450   could be exploited in a Denial Of Service attack.
5451
5452   This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
5453   ([CVE-2018-0732])
5454
5455   *Guido Vranken*
5456
5457 * Cache timing vulnerability in RSA Key Generation
5458
5459   The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
5460   a cache timing side channel attack. An attacker with sufficient access to
5461   mount cache timing attacks during the RSA key generation process could
5462   recover the private key.
5463
5464   This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
5465   Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
5466   ([CVE-2018-0737])
5467
5468   *Billy Brumley*
5469
5470 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
5471   parameter is no longer accepted, as it leads to a corrupt table.  NULL
5472   pem_str is reserved for alias entries only.
5473
5474   *Richard Levitte*
5475
5476 * Revert blinding in ECDSA sign and instead make problematic addition
5477   length-invariant. Switch even to fixed-length Montgomery multiplication.
5478
5479   *Andy Polyakov*
5480
5481 * Change generating and checking of primes so that the error rate of not
5482   being prime depends on the intended use based on the size of the input.
5483   For larger primes this will result in more rounds of Miller-Rabin.
5484   The maximal error rate for primes with more than 1080 bits is lowered
5485   to 2^-128.
5486
5487   *Kurt Roeckx, Annie Yousar*
5488
5489 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
5490
5491   *Kurt Roeckx*
5492
5493 * Add blinding to ECDSA and DSA signatures to protect against side channel
5494   attacks discovered by Keegan Ryan (NCC Group).
5495
5496   *Matt Caswell*
5497
5498 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
5499   now allow empty (zero character) pass phrases.
5500
5501   *Richard Levitte*
5502
5503 * Certificate time validation (X509_cmp_time) enforces stricter
5504   compliance with RFC 5280. Fractional seconds and timezone offsets
5505   are no longer allowed.
5506
5507   *Emilia Käsper*
5508
5509### Changes between 1.0.2n and 1.0.2o [27 Mar 2018]
5510
5511 * Constructed ASN.1 types with a recursive definition could exceed the stack
5512
5513   Constructed ASN.1 types with a recursive definition (such as can be found
5514   in PKCS7) could eventually exceed the stack given malicious input with
5515   excessive recursion. This could result in a Denial Of Service attack. There
5516   are no such structures used within SSL/TLS that come from untrusted sources
5517   so this is considered safe.
5518
5519   This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
5520   project.
5521   ([CVE-2018-0739])
5522
5523   *Matt Caswell*
5524
5525### Changes between 1.0.2m and 1.0.2n [7 Dec 2017]
5526
5527 * Read/write after SSL object in error state
5528
5529   OpenSSL 1.0.2 (starting from version 1.0.2b) introduced an "error state"
5530   mechanism. The intent was that if a fatal error occurred during a handshake
5531   then OpenSSL would move into the error state and would immediately fail if
5532   you attempted to continue the handshake. This works as designed for the
5533   explicit handshake functions (SSL_do_handshake(), SSL_accept() and
5534   SSL_connect()), however due to a bug it does not work correctly if
5535   SSL_read() or SSL_write() is called directly. In that scenario, if the
5536   handshake fails then a fatal error will be returned in the initial function
5537   call. If SSL_read()/SSL_write() is subsequently called by the application
5538   for the same SSL object then it will succeed and the data is passed without
5539   being decrypted/encrypted directly from the SSL/TLS record layer.
5540
5541   In order to exploit this issue an application bug would have to be present
5542   that resulted in a call to SSL_read()/SSL_write() being issued after having
5543   already received a fatal error.
5544
5545   This issue was reported to OpenSSL by David Benjamin (Google).
5546   ([CVE-2017-3737])
5547
5548   *Matt Caswell*
5549
5550 * rsaz_1024_mul_avx2 overflow bug on x86_64
5551
5552   There is an overflow bug in the AVX2 Montgomery multiplication procedure
5553   used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
5554   Analysis suggests that attacks against RSA and DSA as a result of this
5555   defect would be very difficult to perform and are not believed likely.
5556   Attacks against DH1024 are considered just feasible, because most of the
5557   work necessary to deduce information about a private key may be performed
5558   offline. The amount of resources required for such an attack would be
5559   significant. However, for an attack on TLS to be meaningful, the server
5560   would have to share the DH1024 private key among multiple clients, which is
5561   no longer an option since CVE-2016-0701.
5562
5563   This only affects processors that support the AVX2 but not ADX extensions
5564   like Intel Haswell (4th generation).
5565
5566   This issue was reported to OpenSSL by David Benjamin (Google). The issue
5567   was originally found via the OSS-Fuzz project.
5568   ([CVE-2017-3738])
5569
5570   *Andy Polyakov*
5571
5572### Changes between 1.0.2l and 1.0.2m [2 Nov 2017]
5573
5574 * bn_sqrx8x_internal carry bug on x86_64
5575
5576   There is a carry propagating bug in the x86_64 Montgomery squaring
5577   procedure. No EC algorithms are affected. Analysis suggests that attacks
5578   against RSA and DSA as a result of this defect would be very difficult to
5579   perform and are not believed likely. Attacks against DH are considered just
5580   feasible (although very difficult) because most of the work necessary to
5581   deduce information about a private key may be performed offline. The amount
5582   of resources required for such an attack would be very significant and
5583   likely only accessible to a limited number of attackers. An attacker would
5584   additionally need online access to an unpatched system using the target
5585   private key in a scenario with persistent DH parameters and a private
5586   key that is shared between multiple clients.
5587
5588   This only affects processors that support the BMI1, BMI2 and ADX extensions
5589   like Intel Broadwell (5th generation) and later or AMD Ryzen.
5590
5591   This issue was reported to OpenSSL by the OSS-Fuzz project.
5592   ([CVE-2017-3736])
5593
5594   *Andy Polyakov*
5595
5596 * Malformed X.509 IPAddressFamily could cause OOB read
5597
5598   If an X.509 certificate has a malformed IPAddressFamily extension,
5599   OpenSSL could do a one-byte buffer overread. The most likely result
5600   would be an erroneous display of the certificate in text format.
5601
5602   This issue was reported to OpenSSL by the OSS-Fuzz project.
5603
5604   *Rich Salz*
5605
5606### Changes between 1.0.2k and 1.0.2l [25 May 2017]
5607
5608 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
5609   platform rather than 'mingw'.
5610
5611   *Richard Levitte*
5612
5613### Changes between 1.0.2j and 1.0.2k [26 Jan 2017]
5614
5615 * Truncated packet could crash via OOB read
5616
5617   If one side of an SSL/TLS path is running on a 32-bit host and a specific
5618   cipher is being used, then a truncated packet can cause that host to
5619   perform an out-of-bounds read, usually resulting in a crash.
5620
5621   This issue was reported to OpenSSL by Robert Święcki of Google.
5622   ([CVE-2017-3731])
5623
5624   *Andy Polyakov*
5625
5626 * BN_mod_exp may produce incorrect results on x86_64
5627
5628   There is a carry propagating bug in the x86_64 Montgomery squaring
5629   procedure. No EC algorithms are affected. Analysis suggests that attacks
5630   against RSA and DSA as a result of this defect would be very difficult to
5631   perform and are not believed likely. Attacks against DH are considered just
5632   feasible (although very difficult) because most of the work necessary to
5633   deduce information about a private key may be performed offline. The amount
5634   of resources required for such an attack would be very significant and
5635   likely only accessible to a limited number of attackers. An attacker would
5636   additionally need online access to an unpatched system using the target
5637   private key in a scenario with persistent DH parameters and a private
5638   key that is shared between multiple clients. For example this can occur by
5639   default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
5640   similar to CVE-2015-3193 but must be treated as a separate problem.
5641
5642   This issue was reported to OpenSSL by the OSS-Fuzz project.
5643   ([CVE-2017-3732])
5644
5645   *Andy Polyakov*
5646
5647 * Montgomery multiplication may produce incorrect results
5648
5649   There is a carry propagating bug in the Broadwell-specific Montgomery
5650   multiplication procedure that handles input lengths divisible by, but
5651   longer than 256 bits. Analysis suggests that attacks against RSA, DSA
5652   and DH private keys are impossible. This is because the subroutine in
5653   question is not used in operations with the private key itself and an input
5654   of the attacker's direct choice. Otherwise the bug can manifest itself as
5655   transient authentication and key negotiation failures or reproducible
5656   erroneous outcome of public-key operations with specially crafted input.
5657   Among EC algorithms only Brainpool P-512 curves are affected and one
5658   presumably can attack ECDH key negotiation. Impact was not analyzed in
5659   detail, because pre-requisites for attack are considered unlikely. Namely
5660   multiple clients have to choose the curve in question and the server has to
5661   share the private key among them, neither of which is default behaviour.
5662   Even then only clients that chose the curve will be affected.
5663
5664   This issue was publicly reported as transient failures and was not
5665   initially recognized as a security issue. Thanks to Richard Morgan for
5666   providing reproducible case.
5667   ([CVE-2016-7055])
5668
5669   *Andy Polyakov*
5670
5671 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0
5672   or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
5673   prevent issues where no progress is being made and the peer continually
5674   sends unrecognised record types, using up resources processing them.
5675
5676   *Matt Caswell*
5677
5678### Changes between 1.0.2i and 1.0.2j [26 Sep 2016]
5679
5680 * Missing CRL sanity check
5681
5682   A bug fix which included a CRL sanity check was added to OpenSSL 1.1.0
5683   but was omitted from OpenSSL 1.0.2i. As a result any attempt to use
5684   CRLs in OpenSSL 1.0.2i will crash with a null pointer exception.
5685
5686   This issue only affects the OpenSSL 1.0.2i
5687   ([CVE-2016-7052])
5688
5689   *Matt Caswell*
5690
5691### Changes between 1.0.2h and 1.0.2i [22 Sep 2016]
5692
5693 * OCSP Status Request extension unbounded memory growth
5694
5695   A malicious client can send an excessively large OCSP Status Request
5696   extension. If that client continually requests renegotiation, sending a
5697   large OCSP Status Request extension each time, then there will be unbounded
5698   memory growth on the server. This will eventually lead to a Denial Of
5699   Service attack through memory exhaustion. Servers with a default
5700   configuration are vulnerable even if they do not support OCSP. Builds using
5701   the "no-ocsp" build time option are not affected.
5702
5703   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5704   ([CVE-2016-6304])
5705
5706   *Matt Caswell*
5707
5708 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from
5709   HIGH to MEDIUM.
5710
5711   This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan
5712   Leurent (INRIA)
5713   ([CVE-2016-2183])
5714
5715   *Rich Salz*
5716
5717 * OOB write in MDC2_Update()
5718
5719   An overflow can occur in MDC2_Update() either if called directly or
5720   through the EVP_DigestUpdate() function using MDC2. If an attacker
5721   is able to supply very large amounts of input data after a previous
5722   call to EVP_EncryptUpdate() with a partial block then a length check
5723   can overflow resulting in a heap corruption.
5724
5725   The amount of data needed is comparable to SIZE_MAX which is impractical
5726   on most platforms.
5727
5728   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5729   ([CVE-2016-6303])
5730
5731   *Stephen Henson*
5732
5733 * Malformed SHA512 ticket DoS
5734
5735   If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a
5736   DoS attack where a malformed ticket will result in an OOB read which will
5737   ultimately crash.
5738
5739   The use of SHA512 in TLS session tickets is comparatively rare as it requires
5740   a custom server callback and ticket lookup mechanism.
5741
5742   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5743   ([CVE-2016-6302])
5744
5745   *Stephen Henson*
5746
5747 * OOB write in BN_bn2dec()
5748
5749   The function BN_bn2dec() does not check the return value of BN_div_word().
5750   This can cause an OOB write if an application uses this function with an
5751   overly large BIGNUM. This could be a problem if an overly large certificate
5752   or CRL is printed out from an untrusted source. TLS is not affected because
5753   record limits will reject an oversized certificate before it is parsed.
5754
5755   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5756   ([CVE-2016-2182])
5757
5758   *Stephen Henson*
5759
5760 * OOB read in TS_OBJ_print_bio()
5761
5762   The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is
5763   the total length the OID text representation would use and not the amount
5764   of data written. This will result in OOB reads when large OIDs are
5765   presented.
5766
5767   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5768   ([CVE-2016-2180])
5769
5770   *Stephen Henson*
5771
5772 * Pointer arithmetic undefined behaviour
5773
5774   Avoid some undefined pointer arithmetic
5775
5776   A common idiom in the codebase is to check limits in the following manner:
5777   "p + len > limit"
5778
5779   Where "p" points to some malloc'd data of SIZE bytes and
5780   limit == p + SIZE
5781
5782   "len" here could be from some externally supplied data (e.g. from a TLS
5783   message).
5784
5785   The rules of C pointer arithmetic are such that "p + len" is only well
5786   defined where len <= SIZE. Therefore the above idiom is actually
5787   undefined behaviour.
5788
5789   For example this could cause problems if some malloc implementation
5790   provides an address for "p" such that "p + len" actually overflows for
5791   values of len that are too big and therefore p + len < limit.
5792
5793   This issue was reported to OpenSSL by Guido Vranken
5794   ([CVE-2016-2177])
5795
5796   *Matt Caswell*
5797
5798 * Constant time flag not preserved in DSA signing
5799
5800   Operations in the DSA signing algorithm should run in constant time in
5801   order to avoid side channel attacks. A flaw in the OpenSSL DSA
5802   implementation means that a non-constant time codepath is followed for
5803   certain operations. This has been demonstrated through a cache-timing
5804   attack to be sufficient for an attacker to recover the private DSA key.
5805
5806   This issue was reported by César Pereida (Aalto University), Billy Brumley
5807   (Tampere University of Technology), and Yuval Yarom (The University of
5808   Adelaide and NICTA).
5809   ([CVE-2016-2178])
5810
5811   *César Pereida*
5812
5813 * DTLS buffered message DoS
5814
5815   In a DTLS connection where handshake messages are delivered out-of-order
5816   those messages that OpenSSL is not yet ready to process will be buffered
5817   for later use. Under certain circumstances, a flaw in the logic means that
5818   those messages do not get removed from the buffer even though the handshake
5819   has been completed. An attacker could force up to approx. 15 messages to
5820   remain in the buffer when they are no longer required. These messages will
5821   be cleared when the DTLS connection is closed. The default maximum size for
5822   a message is 100k. Therefore the attacker could force an additional 1500k
5823   to be consumed per connection. By opening many simulataneous connections an
5824   attacker could cause a DoS attack through memory exhaustion.
5825
5826   This issue was reported to OpenSSL by Quan Luo.
5827   ([CVE-2016-2179])
5828
5829   *Matt Caswell*
5830
5831 * DTLS replay protection DoS
5832
5833   A flaw in the DTLS replay attack protection mechanism means that records
5834   that arrive for future epochs update the replay protection "window" before
5835   the MAC for the record has been validated. This could be exploited by an
5836   attacker by sending a record for the next epoch (which does not have to
5837   decrypt or have a valid MAC), with a very large sequence number. This means
5838   that all subsequent legitimate packets are dropped causing a denial of
5839   service for a specific DTLS connection.
5840
5841   This issue was reported to OpenSSL by the OCAP audit team.
5842   ([CVE-2016-2181])
5843
5844   *Matt Caswell*
5845
5846 * Certificate message OOB reads
5847
5848   In OpenSSL 1.0.2 and earlier some missing message length checks can result
5849   in OOB reads of up to 2 bytes beyond an allocated buffer. There is a
5850   theoretical DoS risk but this has not been observed in practice on common
5851   platforms.
5852
5853   The messages affected are client certificate, client certificate request
5854   and server certificate. As a result the attack can only be performed
5855   against a client or a server which enables client authentication.
5856
5857   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5858   ([CVE-2016-6306])
5859
5860   *Stephen Henson*
5861
5862### Changes between 1.0.2g and 1.0.2h [3 May 2016]
5863
5864 * Prevent padding oracle in AES-NI CBC MAC check
5865
5866   A MITM attacker can use a padding oracle attack to decrypt traffic
5867   when the connection uses an AES CBC cipher and the server support
5868   AES-NI.
5869
5870   This issue was introduced as part of the fix for Lucky 13 padding
5871   attack ([CVE-2013-0169]). The padding check was rewritten to be in
5872   constant time by making sure that always the same bytes are read and
5873   compared against either the MAC or padding bytes. But it no longer
5874   checked that there was enough data to have both the MAC and padding
5875   bytes.
5876
5877   This issue was reported by Juraj Somorovsky using TLS-Attacker.
5878
5879   *Kurt Roeckx*
5880
5881 * Fix EVP_EncodeUpdate overflow
5882
5883   An overflow can occur in the EVP_EncodeUpdate() function which is used for
5884   Base64 encoding of binary data. If an attacker is able to supply very large
5885   amounts of input data then a length check can overflow resulting in a heap
5886   corruption.
5887
5888   Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by
5889   the `PEM_write_bio*` family of functions. These are mainly used within the
5890   OpenSSL command line applications, so any application which processes data
5891   from an untrusted source and outputs it as a PEM file should be considered
5892   vulnerable to this issue. User applications that call these APIs directly
5893   with large amounts of untrusted data may also be vulnerable.
5894
5895   This issue was reported by Guido Vranken.
5896   ([CVE-2016-2105])
5897
5898   *Matt Caswell*
5899
5900 * Fix EVP_EncryptUpdate overflow
5901
5902   An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
5903   is able to supply very large amounts of input data after a previous call to
5904   EVP_EncryptUpdate() with a partial block then a length check can overflow
5905   resulting in a heap corruption. Following an analysis of all OpenSSL
5906   internal usage of the EVP_EncryptUpdate() function all usage is one of two
5907   forms. The first form is where the EVP_EncryptUpdate() call is known to be
5908   the first called function after an EVP_EncryptInit(), and therefore that
5909   specific call must be safe. The second form is where the length passed to
5910   EVP_EncryptUpdate() can be seen from the code to be some small value and
5911   therefore there is no possibility of an overflow. Since all instances are
5912   one of these two forms, it is believed that there can be no overflows in
5913   internal code due to this problem. It should be noted that
5914   EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
5915   Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
5916   of these calls have also been analysed too and it is believed there are no
5917   instances in internal usage where an overflow could occur.
5918
5919   This issue was reported by Guido Vranken.
5920   ([CVE-2016-2106])
5921
5922   *Matt Caswell*
5923
5924 * Prevent ASN.1 BIO excessive memory allocation
5925
5926   When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
5927   a short invalid encoding can cause allocation of large amounts of memory
5928   potentially consuming excessive resources or exhausting memory.
5929
5930   Any application parsing untrusted data through d2i BIO functions is
5931   affected. The memory based functions such as d2i_X509() are *not* affected.
5932   Since the memory based functions are used by the TLS library, TLS
5933   applications are not affected.
5934
5935   This issue was reported by Brian Carpenter.
5936   ([CVE-2016-2109])
5937
5938   *Stephen Henson*
5939
5940 * EBCDIC overread
5941
5942   ASN1 Strings that are over 1024 bytes can cause an overread in applications
5943   using the X509_NAME_oneline() function on EBCDIC systems. This could result
5944   in arbitrary stack data being returned in the buffer.
5945
5946   This issue was reported by Guido Vranken.
5947   ([CVE-2016-2176])
5948
5949   *Matt Caswell*
5950
5951 * Modify behavior of ALPN to invoke callback after SNI/servername
5952   callback, such that updates to the SSL_CTX affect ALPN.
5953
5954   *Todd Short*
5955
5956 * Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
5957   default.
5958
5959   *Kurt Roeckx*
5960
5961 * Only remove the SSLv2 methods with the no-ssl2-method option. When the
5962   methods are enabled and ssl2 is disabled the methods return NULL.
5963
5964   *Kurt Roeckx*
5965
5966### Changes between 1.0.2f and 1.0.2g [1 Mar 2016]
5967
5968* Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
5969  Builds that are not configured with "enable-weak-ssl-ciphers" will not
5970  provide any "EXPORT" or "LOW" strength ciphers.
5971
5972  *Viktor Dukhovni*
5973
5974* Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
5975  is by default disabled at build-time.  Builds that are not configured with
5976  "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
5977  users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
5978  will need to explicitly call either of:
5979
5980      SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
5981  or
5982      SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
5983
5984  as appropriate.  Even if either of those is used, or the application
5985  explicitly uses the version-specific SSLv2_method() or its client and
5986  server variants, SSLv2 ciphers vulnerable to exhaustive search key
5987  recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
5988  ciphers, and SSLv2 56-bit DES are no longer available.
5989  ([CVE-2016-0800])
5990
5991   *Viktor Dukhovni*
5992
5993 * Fix a double-free in DSA code
5994
5995   A double free bug was discovered when OpenSSL parses malformed DSA private
5996   keys and could lead to a DoS attack or memory corruption for applications
5997   that receive DSA private keys from untrusted sources.  This scenario is
5998   considered rare.
5999
6000   This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
6001   libFuzzer.
6002   ([CVE-2016-0705])
6003
6004   *Stephen Henson*
6005
6006 * Disable SRP fake user seed to address a server memory leak.
6007
6008   Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
6009
6010   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
6011   In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
6012   was changed to ignore the "fake user" SRP seed, even if the seed
6013   is configured.
6014
6015   Users should use SRP_VBASE_get1_by_user instead. Note that in
6016   SRP_VBASE_get1_by_user, caller must free the returned value. Note
6017   also that even though configuring the SRP seed attempts to hide
6018   invalid usernames by continuing the handshake with fake
6019   credentials, this behaviour is not constant time and no strong
6020   guarantees are made that the handshake is indistinguishable from
6021   that of a valid user.
6022   ([CVE-2016-0798])
6023
6024   *Emilia Käsper*
6025
6026 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
6027
6028   In the BN_hex2bn function the number of hex digits is calculated using an
6029   int value `i`. Later `bn_expand` is called with a value of `i * 4`. For
6030   large values of `i` this can result in `bn_expand` not allocating any
6031   memory because `i * 4` is negative. This can leave the internal BIGNUM data
6032   field as NULL leading to a subsequent NULL ptr deref. For very large values
6033   of `i`, the calculation `i * 4` could be a positive value smaller than `i`.
6034   In this case memory is allocated to the internal BIGNUM data field, but it
6035   is insufficiently sized leading to heap corruption. A similar issue exists
6036   in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
6037   is ever called by user applications with very large untrusted hex/dec data.
6038   This is anticipated to be a rare occurrence.
6039
6040   All OpenSSL internal usage of these functions use data that is not expected
6041   to be untrusted, e.g. config file data or application command line
6042   arguments. If user developed applications generate config file data based
6043   on untrusted data then it is possible that this could also lead to security
6044   consequences. This is also anticipated to be rare.
6045
6046   This issue was reported to OpenSSL by Guido Vranken.
6047   ([CVE-2016-0797])
6048
6049   *Matt Caswell*
6050
6051 * Fix memory issues in `BIO_*printf` functions
6052
6053   The internal `fmtstr` function used in processing a "%s" format string in
6054   the `BIO_*printf` functions could overflow while calculating the length of a
6055   string and cause an OOB read when printing very long strings.
6056
6057   Additionally the internal `doapr_outch` function can attempt to write to an
6058   OOB memory location (at an offset from the NULL pointer) in the event of a
6059   memory allocation failure. In 1.0.2 and below this could be caused where
6060   the size of a buffer to be allocated is greater than INT_MAX. E.g. this
6061   could be in processing a very long "%s" format string. Memory leaks can
6062   also occur.
6063
6064   The first issue may mask the second issue dependent on compiler behaviour.
6065   These problems could enable attacks where large amounts of untrusted data
6066   is passed to the `BIO_*printf` functions. If applications use these functions
6067   in this way then they could be vulnerable. OpenSSL itself uses these
6068   functions when printing out human-readable dumps of ASN.1 data. Therefore
6069   applications that print this data could be vulnerable if the data is from
6070   untrusted sources. OpenSSL command line applications could also be
6071   vulnerable where they print out ASN.1 data, or if untrusted data is passed
6072   as command line arguments.
6073
6074   Libssl is not considered directly vulnerable. Additionally certificates etc
6075   received via remote connections via libssl are also unlikely to be able to
6076   trigger these issues because of message size limits enforced within libssl.
6077
6078   This issue was reported to OpenSSL Guido Vranken.
6079   ([CVE-2016-0799])
6080
6081   *Matt Caswell*
6082
6083 * Side channel attack on modular exponentiation
6084
6085   A side-channel attack was found which makes use of cache-bank conflicts on
6086   the Intel Sandy-Bridge microarchitecture which could lead to the recovery
6087   of RSA keys.  The ability to exploit this issue is limited as it relies on
6088   an attacker who has control of code in a thread running on the same
6089   hyper-threaded core as the victim thread which is performing decryptions.
6090
6091   This issue was reported to OpenSSL by Yuval Yarom, The University of
6092   Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
6093   Nadia Heninger, University of Pennsylvania with more information at
6094   <http://cachebleed.info>.
6095   ([CVE-2016-0702])
6096
6097   *Andy Polyakov*
6098
6099 * Change the `req` command to generate a 2048-bit RSA/DSA key by default,
6100   if no keysize is specified with default_bits. This fixes an
6101   omission in an earlier change that changed all RSA/DSA key generation
6102   commands to use 2048 bits by default.
6103
6104   *Emilia Käsper*
6105
6106### Changes between 1.0.2e and 1.0.2f [28 Jan 2016]
6107
6108 * DH small subgroups
6109
6110   Historically OpenSSL only ever generated DH parameters based on "safe"
6111   primes. More recently (in version 1.0.2) support was provided for
6112   generating X9.42 style parameter files such as those required for RFC 5114
6113   support. The primes used in such files may not be "safe". Where an
6114   application is using DH configured with parameters based on primes that are
6115   not "safe" then an attacker could use this fact to find a peer's private
6116   DH exponent. This attack requires that the attacker complete multiple
6117   handshakes in which the peer uses the same private DH exponent. For example
6118   this could be used to discover a TLS server's private DH exponent if it's
6119   reusing the private DH exponent or it's using a static DH ciphersuite.
6120
6121   OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in
6122   TLS. It is not on by default. If the option is not set then the server
6123   reuses the same private DH exponent for the life of the server process and
6124   would be vulnerable to this attack. It is believed that many popular
6125   applications do set this option and would therefore not be at risk.
6126
6127   The fix for this issue adds an additional check where a "q" parameter is
6128   available (as is the case in X9.42 based parameters). This detects the
6129   only known attack, and is the only possible defense for static DH
6130   ciphersuites. This could have some performance impact.
6131
6132   Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by
6133   default and cannot be disabled. This could have some performance impact.
6134
6135   This issue was reported to OpenSSL by Antonio Sanso (Adobe).
6136   ([CVE-2016-0701])
6137
6138   *Matt Caswell*
6139
6140 * SSLv2 doesn't block disabled ciphers
6141
6142   A malicious client can negotiate SSLv2 ciphers that have been disabled on
6143   the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
6144   been disabled, provided that the SSLv2 protocol was not also disabled via
6145   SSL_OP_NO_SSLv2.
6146
6147   This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
6148   and Sebastian Schinzel.
6149   ([CVE-2015-3197])
6150
6151   *Viktor Dukhovni*
6152
6153### Changes between 1.0.2d and 1.0.2e [3 Dec 2015]
6154
6155 * BN_mod_exp may produce incorrect results on x86_64
6156
6157   There is a carry propagating bug in the x86_64 Montgomery squaring
6158   procedure. No EC algorithms are affected. Analysis suggests that attacks
6159   against RSA and DSA as a result of this defect would be very difficult to
6160   perform and are not believed likely. Attacks against DH are considered just
6161   feasible (although very difficult) because most of the work necessary to
6162   deduce information about a private key may be performed offline. The amount
6163   of resources required for such an attack would be very significant and
6164   likely only accessible to a limited number of attackers. An attacker would
6165   additionally need online access to an unpatched system using the target
6166   private key in a scenario with persistent DH parameters and a private
6167   key that is shared between multiple clients. For example this can occur by
6168   default in OpenSSL DHE based SSL/TLS ciphersuites.
6169
6170   This issue was reported to OpenSSL by Hanno Böck.
6171   ([CVE-2015-3193])
6172
6173   *Andy Polyakov*
6174
6175 * Certificate verify crash with missing PSS parameter
6176
6177   The signature verification routines will crash with a NULL pointer
6178   dereference if presented with an ASN.1 signature using the RSA PSS
6179   algorithm and absent mask generation function parameter. Since these
6180   routines are used to verify certificate signature algorithms this can be
6181   used to crash any certificate verification operation and exploited in a
6182   DoS attack. Any application which performs certificate verification is
6183   vulnerable including OpenSSL clients and servers which enable client
6184   authentication.
6185
6186   This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
6187   ([CVE-2015-3194])
6188
6189   *Stephen Henson*
6190
6191 * X509_ATTRIBUTE memory leak
6192
6193   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
6194   memory. This structure is used by the PKCS#7 and CMS routines so any
6195   application which reads PKCS#7 or CMS data from untrusted sources is
6196   affected. SSL/TLS is not affected.
6197
6198   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
6199   libFuzzer.
6200   ([CVE-2015-3195])
6201
6202   *Stephen Henson*
6203
6204 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
6205   This changes the decoding behaviour for some invalid messages,
6206   though the change is mostly in the more lenient direction, and
6207   legacy behaviour is preserved as much as possible.
6208
6209   *Emilia Käsper*
6210
6211 * In DSA_generate_parameters_ex, if the provided seed is too short,
6212   return an error
6213
6214   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
6215
6216### Changes between 1.0.2c and 1.0.2d [9 Jul 2015]
6217
6218 * Alternate chains certificate forgery
6219
6220   During certificate verification, OpenSSL will attempt to find an
6221   alternative certificate chain if the first attempt to build such a chain
6222   fails. An error in the implementation of this logic can mean that an
6223   attacker could cause certain checks on untrusted certificates to be
6224   bypassed, such as the CA flag, enabling them to use a valid leaf
6225   certificate to act as a CA and "issue" an invalid certificate.
6226
6227   This issue was reported to OpenSSL by Adam Langley/David Benjamin
6228   (Google/BoringSSL).
6229
6230   *Matt Caswell*
6231
6232### Changes between 1.0.2b and 1.0.2c [12 Jun 2015]
6233
6234 * Fix HMAC ABI incompatibility. The previous version introduced an ABI
6235   incompatibility in the handling of HMAC. The previous ABI has now been
6236   restored.
6237
6238   *Matt Caswell*
6239
6240### Changes between 1.0.2a and 1.0.2b [11 Jun 2015]
6241
6242 * Malformed ECParameters causes infinite loop
6243
6244   When processing an ECParameters structure OpenSSL enters an infinite loop
6245   if the curve specified is over a specially malformed binary polynomial
6246   field.
6247
6248   This can be used to perform denial of service against any
6249   system which processes public keys, certificate requests or
6250   certificates.  This includes TLS clients and TLS servers with
6251   client authentication enabled.
6252
6253   This issue was reported to OpenSSL by Joseph Barr-Pixton.
6254   ([CVE-2015-1788])
6255
6256   *Andy Polyakov*
6257
6258 * Exploitable out-of-bounds read in X509_cmp_time
6259
6260   X509_cmp_time does not properly check the length of the ASN1_TIME
6261   string and can read a few bytes out of bounds. In addition,
6262   X509_cmp_time accepts an arbitrary number of fractional seconds in the
6263   time string.
6264
6265   An attacker can use this to craft malformed certificates and CRLs of
6266   various sizes and potentially cause a segmentation fault, resulting in
6267   a DoS on applications that verify certificates or CRLs. TLS clients
6268   that verify CRLs are affected. TLS clients and servers with client
6269   authentication enabled may be affected if they use custom verification
6270   callbacks.
6271
6272   This issue was reported to OpenSSL by Robert Swiecki (Google), and
6273   independently by Hanno Böck.
6274   ([CVE-2015-1789])
6275
6276   *Emilia Käsper*
6277
6278 * PKCS7 crash with missing EnvelopedContent
6279
6280   The PKCS#7 parsing code does not handle missing inner EncryptedContent
6281   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
6282   with missing content and trigger a NULL pointer dereference on parsing.
6283
6284   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
6285   structures from untrusted sources are affected. OpenSSL clients and
6286   servers are not affected.
6287
6288   This issue was reported to OpenSSL by Michal Zalewski (Google).
6289   ([CVE-2015-1790])
6290
6291   *Emilia Käsper*
6292
6293 * CMS verify infinite loop with unknown hash function
6294
6295   When verifying a signedData message the CMS code can enter an infinite loop
6296   if presented with an unknown hash function OID. This can be used to perform
6297   denial of service against any system which verifies signedData messages using
6298   the CMS code.
6299   This issue was reported to OpenSSL by Johannes Bauer.
6300   ([CVE-2015-1792])
6301
6302   *Stephen Henson*
6303
6304 * Race condition handling NewSessionTicket
6305
6306   If a NewSessionTicket is received by a multi-threaded client when attempting to
6307   reuse a previous ticket then a race condition can occur potentially leading to
6308   a double free of the ticket data.
6309   ([CVE-2015-1791])
6310
6311   *Matt Caswell*
6312
6313 * Only support 256-bit or stronger elliptic curves with the
6314   'ecdh_auto' setting (server) or by default (client). Of supported
6315   curves, prefer P-256 (both).
6316
6317   *Emilia Kasper*
6318
6319### Changes between 1.0.2 and 1.0.2a [19 Mar 2015]
6320
6321 * ClientHello sigalgs DoS fix
6322
6323   If a client connects to an OpenSSL 1.0.2 server and renegotiates with an
6324   invalid signature algorithms extension a NULL pointer dereference will
6325   occur. This can be exploited in a DoS attack against the server.
6326
6327   This issue was was reported to OpenSSL by David Ramos of Stanford
6328   University.
6329   ([CVE-2015-0291])
6330
6331   *Stephen Henson and Matt Caswell*
6332
6333 * Multiblock corrupted pointer fix
6334
6335   OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This
6336   feature only applies on 64 bit x86 architecture platforms that support AES
6337   NI instructions. A defect in the implementation of "multiblock" can cause
6338   OpenSSL's internal write buffer to become incorrectly set to NULL when
6339   using non-blocking IO. Typically, when the user application is using a
6340   socket BIO for writing, this will only result in a failed connection.
6341   However if some other BIO is used then it is likely that a segmentation
6342   fault will be triggered, thus enabling a potential DoS attack.
6343
6344   This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller.
6345   ([CVE-2015-0290])
6346
6347   *Matt Caswell*
6348
6349 * Segmentation fault in DTLSv1_listen fix
6350
6351   The DTLSv1_listen function is intended to be stateless and processes the
6352   initial ClientHello from many peers. It is common for user code to loop
6353   over the call to DTLSv1_listen until a valid ClientHello is received with
6354   an associated cookie. A defect in the implementation of DTLSv1_listen means
6355   that state is preserved in the SSL object from one invocation to the next
6356   that can lead to a segmentation fault. Errors processing the initial
6357   ClientHello can trigger this scenario. An example of such an error could be
6358   that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only
6359   server.
6360
6361   This issue was reported to OpenSSL by Per Allansson.
6362   ([CVE-2015-0207])
6363
6364   *Matt Caswell*
6365
6366 * Segmentation fault in ASN1_TYPE_cmp fix
6367
6368   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
6369   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
6370   certificate signature algorithm consistency this can be used to crash any
6371   certificate verification operation and exploited in a DoS attack. Any
6372   application which performs certificate verification is vulnerable including
6373   OpenSSL clients and servers which enable client authentication.
6374   ([CVE-2015-0286])
6375
6376   *Stephen Henson*
6377
6378 * Segmentation fault for invalid PSS parameters fix
6379
6380   The signature verification routines will crash with a NULL pointer
6381   dereference if presented with an ASN.1 signature using the RSA PSS
6382   algorithm and invalid parameters. Since these routines are used to verify
6383   certificate signature algorithms this can be used to crash any
6384   certificate verification operation and exploited in a DoS attack. Any
6385   application which performs certificate verification is vulnerable including
6386   OpenSSL clients and servers which enable client authentication.
6387
6388   This issue was was reported to OpenSSL by Brian Carpenter.
6389   ([CVE-2015-0208])
6390
6391   *Stephen Henson*
6392
6393 * ASN.1 structure reuse memory corruption fix
6394
6395   Reusing a structure in ASN.1 parsing may allow an attacker to cause
6396   memory corruption via an invalid write. Such reuse is and has been
6397   strongly discouraged and is believed to be rare.
6398
6399   Applications that parse structures containing CHOICE or ANY DEFINED BY
6400   components may be affected. Certificate parsing (d2i_X509 and related
6401   functions) are however not affected. OpenSSL clients and servers are
6402   not affected.
6403   ([CVE-2015-0287])
6404
6405   *Stephen Henson*
6406
6407 * PKCS7 NULL pointer dereferences fix
6408
6409   The PKCS#7 parsing code does not handle missing outer ContentInfo
6410   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
6411   missing content and trigger a NULL pointer dereference on parsing.
6412
6413   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
6414   otherwise parse PKCS#7 structures from untrusted sources are
6415   affected. OpenSSL clients and servers are not affected.
6416
6417   This issue was reported to OpenSSL by Michal Zalewski (Google).
6418   ([CVE-2015-0289])
6419
6420   *Emilia Käsper*
6421
6422 * DoS via reachable assert in SSLv2 servers fix
6423
6424   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
6425   servers that both support SSLv2 and enable export cipher suites by sending
6426   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
6427
6428   This issue was discovered by Sean Burford (Google) and Emilia Käsper
6429   (OpenSSL development team).
6430   ([CVE-2015-0293])
6431
6432   *Emilia Käsper*
6433
6434 * Empty CKE with client auth and DHE fix
6435
6436   If client auth is used then a server can seg fault in the event of a DHE
6437   ciphersuite being selected and a zero length ClientKeyExchange message
6438   being sent by the client. This could be exploited in a DoS attack.
6439   ([CVE-2015-1787])
6440
6441   *Matt Caswell*
6442
6443 * Handshake with unseeded PRNG fix
6444
6445   Under certain conditions an OpenSSL 1.0.2 client can complete a handshake
6446   with an unseeded PRNG. The conditions are:
6447   - The client is on a platform where the PRNG has not been seeded
6448   automatically, and the user has not seeded manually
6449   - A protocol specific client method version has been used (i.e. not
6450   SSL_client_methodv23)
6451   - A ciphersuite is used that does not require additional random data from
6452   the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA).
6453
6454   If the handshake succeeds then the client random that has been used will
6455   have been generated from a PRNG with insufficient entropy and therefore the
6456   output may be predictable.
6457
6458   For example using the following command with an unseeded openssl will
6459   succeed on an unpatched platform:
6460
6461   openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA
6462   ([CVE-2015-0285])
6463
6464   *Matt Caswell*
6465
6466 * Use After Free following d2i_ECPrivatekey error fix
6467
6468   A malformed EC private key file consumed via the d2i_ECPrivateKey function
6469   could cause a use after free condition. This, in turn, could cause a double
6470   free in several private key parsing functions (such as d2i_PrivateKey
6471   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
6472   for applications that receive EC private keys from untrusted
6473   sources. This scenario is considered rare.
6474
6475   This issue was discovered by the BoringSSL project and fixed in their
6476   commit 517073cd4b.
6477   ([CVE-2015-0209])
6478
6479   *Matt Caswell*
6480
6481 * X509_to_X509_REQ NULL pointer deref fix
6482
6483   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
6484   the certificate key is invalid. This function is rarely used in practice.
6485
6486   This issue was discovered by Brian Carpenter.
6487   ([CVE-2015-0288])
6488
6489   *Stephen Henson*
6490
6491 * Removed the export ciphers from the DEFAULT ciphers
6492
6493   *Kurt Roeckx*
6494
6495### Changes between 1.0.1l and 1.0.2 [22 Jan 2015]
6496
6497 * Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g.
6498   ARMv5 through ARMv8, as opposite to "locking" it to single one.
6499   So far those who have to target multiple platforms would compromise
6500   and argue that binary targeting say ARMv5 would still execute on
6501   ARMv8. "Universal" build resolves this compromise by providing
6502   near-optimal performance even on newer platforms.
6503
6504   *Andy Polyakov*
6505
6506 * Accelerated NIST P-256 elliptic curve implementation for x86_64
6507   (other platforms pending).
6508
6509   *Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov*
6510
6511 * Add support for the SignedCertificateTimestampList certificate and
6512   OCSP response extensions from RFC6962.
6513
6514   *Rob Stradling*
6515
6516 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
6517   for corner cases. (Certain input points at infinity could lead to
6518   bogus results, with non-infinity inputs mapped to infinity too.)
6519
6520   *Bodo Moeller*
6521
6522 * Initial support for PowerISA 2.0.7, first implemented in POWER8.
6523   This covers AES, SHA256/512 and GHASH. "Initial" means that most
6524   common cases are optimized and there still is room for further
6525   improvements. Vector Permutation AES for Altivec is also added.
6526
6527   *Andy Polyakov*
6528
6529 * Add support for little-endian ppc64 Linux target.
6530
6531   *Marcelo Cerri (IBM)*
6532
6533 * Initial support for AMRv8 ISA crypto extensions. This covers AES,
6534   SHA1, SHA256 and GHASH. "Initial" means that most common cases
6535   are optimized and there still is room for further improvements.
6536   Both 32- and 64-bit modes are supported.
6537
6538   *Andy Polyakov, Ard Biesheuvel (Linaro)*
6539
6540 * Improved ARMv7 NEON support.
6541
6542   *Andy Polyakov*
6543
6544 * Support for SPARC Architecture 2011 crypto extensions, first
6545   implemented in SPARC T4. This covers AES, DES, Camellia, SHA1,
6546   SHA256/512, MD5, GHASH and modular exponentiation.
6547
6548   *Andy Polyakov, David Miller*
6549
6550 * Accelerated modular exponentiation for Intel processors, a.k.a.
6551   RSAZ.
6552
6553   *Shay Gueron & Vlad Krasnov (Intel Corp)*
6554
6555 * Support for new and upcoming Intel processors, including AVX2,
6556   BMI and SHA ISA extensions. This includes additional "stitched"
6557   implementations, AESNI-SHA256 and GCM, and multi-buffer support
6558   for TLS encrypt.
6559
6560   This work was sponsored by Intel Corp.
6561
6562   *Andy Polyakov*
6563
6564 * Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method()
6565   supports both DTLS 1.2 and 1.0 and should use whatever version the peer
6566   supports and DTLSv1_2_*_method() which supports DTLS 1.2 only.
6567
6568   *Steve Henson*
6569
6570 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
6571   this fixes a limitation in previous versions of OpenSSL.
6572
6573   *Steve Henson*
6574
6575 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
6576   MGF1 digest and OAEP label.
6577
6578   *Steve Henson*
6579
6580 * Add EVP support for key wrapping algorithms, to avoid problems with
6581   existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
6582   the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
6583   algorithms and include tests cases.
6584
6585   *Steve Henson*
6586
6587 * Add functions to allocate and set the fields of an ECDSA_METHOD
6588   structure.
6589
6590   *Douglas E. Engert, Steve Henson*
6591
6592 * New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the
6593   difference in days and seconds between two tm or ASN1_TIME structures.
6594
6595   *Steve Henson*
6596
6597 * Add -rev test option to s_server to just reverse order of characters
6598   received by client and send back to server. Also prints an abbreviated
6599   summary of the connection parameters.
6600
6601   *Steve Henson*
6602
6603 * New option -brief for s_client and s_server to print out a brief summary
6604   of connection parameters.
6605
6606   *Steve Henson*
6607
6608 * Add callbacks for arbitrary TLS extensions.
6609
6610   *Trevor Perrin <trevp@trevp.net> and Ben Laurie*
6611
6612 * New option -crl_download in several openssl utilities to download CRLs
6613   from CRLDP extension in certificates.
6614
6615   *Steve Henson*
6616
6617 * New options -CRL and -CRLform for s_client and s_server for CRLs.
6618
6619   *Steve Henson*
6620
6621 * New function X509_CRL_diff to generate a delta CRL from the difference
6622   of two full CRLs. Add support to "crl" utility.
6623
6624   *Steve Henson*
6625
6626 * New functions to set lookup_crls function and to retrieve
6627   X509_STORE from X509_STORE_CTX.
6628
6629   *Steve Henson*
6630
6631 * Print out deprecated issuer and subject unique ID fields in
6632   certificates.
6633
6634   *Steve Henson*
6635
6636 * Extend OCSP I/O functions so they can be used for simple general purpose
6637   HTTP as well as OCSP. New wrapper function which can be used to download
6638   CRLs using the OCSP API.
6639
6640   *Steve Henson*
6641
6642 * Delegate command line handling in s_client/s_server to SSL_CONF APIs.
6643
6644   *Steve Henson*
6645
6646 * `SSL_CONF*` functions. These provide a common framework for application
6647   configuration using configuration files or command lines.
6648
6649   *Steve Henson*
6650
6651 * SSL/TLS tracing code. This parses out SSL/TLS records using the
6652   message callback and prints the results. Needs compile time option
6653   "enable-ssl-trace". New options to s_client and s_server to enable
6654   tracing.
6655
6656   *Steve Henson*
6657
6658 * New ctrl and macro to retrieve supported points extensions.
6659   Print out extension in s_server and s_client.
6660
6661   *Steve Henson*
6662
6663 * New functions to retrieve certificate signature and signature
6664   OID NID.
6665
6666   *Steve Henson*
6667
6668 * Add functions to retrieve and manipulate the raw cipherlist sent by a
6669   client to OpenSSL.
6670
6671   *Steve Henson*
6672
6673 * New Suite B modes for TLS code. These use and enforce the requirements
6674   of RFC6460: restrict ciphersuites, only permit Suite B algorithms and
6675   only use Suite B curves. The Suite B modes can be set by using the
6676   strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring.
6677
6678   *Steve Henson*
6679
6680 * New chain verification flags for Suite B levels of security. Check
6681   algorithms are acceptable when flags are set in X509_verify_cert.
6682
6683   *Steve Henson*
6684
6685 * Make tls1_check_chain return a set of flags indicating checks passed
6686   by a certificate chain. Add additional tests to handle client
6687   certificates: checks for matching certificate type and issuer name
6688   comparison.
6689
6690   *Steve Henson*
6691
6692 * If an attempt is made to use a signature algorithm not in the peer
6693   preference list abort the handshake. If client has no suitable
6694   signature algorithms in response to a certificate request do not
6695   use the certificate.
6696
6697   *Steve Henson*
6698
6699 * If server EC tmp key is not in client preference list abort handshake.
6700
6701   *Steve Henson*
6702
6703 * Add support for certificate stores in CERT structure. This makes it
6704   possible to have different stores per SSL structure or one store in
6705   the parent SSL_CTX. Include distinct stores for certificate chain
6706   verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN
6707   to build and store a certificate chain in CERT structure: returning
6708   an error if the chain cannot be built: this will allow applications
6709   to test if a chain is correctly configured.
6710
6711   Note: if the CERT based stores are not set then the parent SSL_CTX
6712   store is used to retain compatibility with existing behaviour.
6713
6714   *Steve Henson*
6715
6716 * New function ssl_set_client_disabled to set a ciphersuite disabled
6717   mask based on the current session, check mask when sending client
6718   hello and checking the requested ciphersuite.
6719
6720   *Steve Henson*
6721
6722 * New ctrls to retrieve and set certificate types in a certificate
6723   request message. Print out received values in s_client. If certificate
6724   types is not set with custom values set sensible values based on
6725   supported signature algorithms.
6726
6727   *Steve Henson*
6728
6729 * Support for distinct client and server supported signature algorithms.
6730
6731   *Steve Henson*
6732
6733 * Add certificate callback. If set this is called whenever a certificate
6734   is required by client or server. An application can decide which
6735   certificate chain to present based on arbitrary criteria: for example
6736   supported signature algorithms. Add very simple example to s_server.
6737   This fixes many of the problems and restrictions of the existing client
6738   certificate callback: for example you can now clear an existing
6739   certificate and specify the whole chain.
6740
6741   *Steve Henson*
6742
6743 * Add new "valid_flags" field to CERT_PKEY structure which determines what
6744   the certificate can be used for (if anything). Set valid_flags field
6745   in new tls1_check_chain function. Simplify ssl_set_cert_masks which used
6746   to have similar checks in it.
6747
6748   Add new "cert_flags" field to CERT structure and include a "strict mode".
6749   This enforces some TLS certificate requirements (such as only permitting
6750   certificate signature algorithms contained in the supported algorithms
6751   extension) which some implementations ignore: this option should be used
6752   with caution as it could cause interoperability issues.
6753
6754   *Steve Henson*
6755
6756 * Update and tidy signature algorithm extension processing. Work out
6757   shared signature algorithms based on preferences and peer algorithms
6758   and print them out in s_client and s_server. Abort handshake if no
6759   shared signature algorithms.
6760
6761   *Steve Henson*
6762
6763 * Add new functions to allow customised supported signature algorithms
6764   for SSL and SSL_CTX structures. Add options to s_client and s_server
6765   to support them.
6766
6767   *Steve Henson*
6768
6769 * New function SSL_certs_clear() to delete all references to certificates
6770   from an SSL structure. Before this once a certificate had been added
6771   it couldn't be removed.
6772
6773   *Steve Henson*
6774
6775 * Integrate hostname, email address and IP address checking with certificate
6776   verification. New verify options supporting checking in openssl utility.
6777
6778   *Steve Henson*
6779
6780 * Fixes and wildcard matching support to hostname and email checking
6781   functions. Add manual page.
6782
6783   *Florian Weimer (Red Hat Product Security Team)*
6784
6785 * New functions to check a hostname email or IP address against a
6786   certificate. Add options x509 utility to print results of checks against
6787   a certificate.
6788
6789   *Steve Henson*
6790
6791 * Fix OCSP checking.
6792
6793   *Rob Stradling <rob.stradling@comodo.com> and Ben Laurie*
6794
6795 * Initial experimental support for explicitly trusted non-root CAs.
6796   OpenSSL still tries to build a complete chain to a root but if an
6797   intermediate CA has a trust setting included that is used. The first
6798   setting is used: whether to trust (e.g., -addtrust option to the x509
6799   utility) or reject.
6800
6801   *Steve Henson*
6802
6803 * Add -trusted_first option which attempts to find certificates in the
6804   trusted store even if an untrusted chain is also supplied.
6805
6806   *Steve Henson*
6807
6808 * MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE,
6809   platform support for Linux and Android.
6810
6811   *Andy Polyakov*
6812
6813 * Support for linux-x32, ILP32 environment in x86_64 framework.
6814
6815   *Andy Polyakov*
6816
6817 * Experimental multi-implementation support for FIPS capable OpenSSL.
6818   When in FIPS mode the approved implementations are used as normal,
6819   when not in FIPS mode the internal unapproved versions are used instead.
6820   This means that the FIPS capable OpenSSL isn't forced to use the
6821   (often lower performance) FIPS implementations outside FIPS mode.
6822
6823   *Steve Henson*
6824
6825 * Transparently support X9.42 DH parameters when calling
6826   PEM_read_bio_DHparameters. This means existing applications can handle
6827   the new parameter format automatically.
6828
6829   *Steve Henson*
6830
6831 * Initial experimental support for X9.42 DH parameter format: mainly
6832   to support use of 'q' parameter for RFC5114 parameters.
6833
6834   *Steve Henson*
6835
6836 * Add DH parameters from RFC5114 including test data to dhtest.
6837
6838   *Steve Henson*
6839
6840 * Support for automatic EC temporary key parameter selection. If enabled
6841   the most preferred EC parameters are automatically used instead of
6842   hardcoded fixed parameters. Now a server just has to call:
6843   SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically
6844   support ECDH and use the most appropriate parameters.
6845
6846   *Steve Henson*
6847
6848 * Enhance and tidy EC curve and point format TLS extension code. Use
6849   static structures instead of allocation if default values are used.
6850   New ctrls to set curves we wish to support and to retrieve shared curves.
6851   Print out shared curves in s_server. New options to s_server and s_client
6852   to set list of supported curves.
6853
6854   *Steve Henson*
6855
6856 * New ctrls to retrieve supported signature algorithms and
6857   supported curve values as an array of NIDs. Extend openssl utility
6858   to print out received values.
6859
6860   *Steve Henson*
6861
6862 * Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert
6863   between NIDs and the more common NIST names such as "P-256". Enhance
6864   ecparam utility and ECC method to recognise the NIST names for curves.
6865
6866   *Steve Henson*
6867
6868 * Enhance SSL/TLS certificate chain handling to support different
6869   chains for each certificate instead of one chain in the parent SSL_CTX.
6870
6871   *Steve Henson*
6872
6873 * Support for fixed DH ciphersuite client authentication: where both
6874   server and client use DH certificates with common parameters.
6875
6876   *Steve Henson*
6877
6878 * Support for fixed DH ciphersuites: those requiring DH server
6879   certificates.
6880
6881   *Steve Henson*
6882
6883 * New function i2d_re_X509_tbs for re-encoding the TBS portion of
6884   the certificate.
6885   Note: Related 1.0.2-beta specific macros X509_get_cert_info,
6886   X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and
6887   X509_CINF_get_signature were reverted post internal team review.
6888
6889OpenSSL 1.0.1
6890-------------
6891
6892### Changes between 1.0.1t and 1.0.1u [22 Sep 2016]
6893
6894 * OCSP Status Request extension unbounded memory growth
6895
6896   A malicious client can send an excessively large OCSP Status Request
6897   extension. If that client continually requests renegotiation, sending a
6898   large OCSP Status Request extension each time, then there will be unbounded
6899   memory growth on the server. This will eventually lead to a Denial Of
6900   Service attack through memory exhaustion. Servers with a default
6901   configuration are vulnerable even if they do not support OCSP. Builds using
6902   the "no-ocsp" build time option are not affected.
6903
6904   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6905   ([CVE-2016-6304])
6906
6907   *Matt Caswell*
6908
6909 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from
6910   HIGH to MEDIUM.
6911
6912   This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan
6913   Leurent (INRIA)
6914   ([CVE-2016-2183])
6915
6916   *Rich Salz*
6917
6918 * OOB write in MDC2_Update()
6919
6920   An overflow can occur in MDC2_Update() either if called directly or
6921   through the EVP_DigestUpdate() function using MDC2. If an attacker
6922   is able to supply very large amounts of input data after a previous
6923   call to EVP_EncryptUpdate() with a partial block then a length check
6924   can overflow resulting in a heap corruption.
6925
6926   The amount of data needed is comparable to SIZE_MAX which is impractical
6927   on most platforms.
6928
6929   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6930   ([CVE-2016-6303])
6931
6932   *Stephen Henson*
6933
6934 * Malformed SHA512 ticket DoS
6935
6936   If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a
6937   DoS attack where a malformed ticket will result in an OOB read which will
6938   ultimately crash.
6939
6940   The use of SHA512 in TLS session tickets is comparatively rare as it requires
6941   a custom server callback and ticket lookup mechanism.
6942
6943   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6944   ([CVE-2016-6302])
6945
6946   *Stephen Henson*
6947
6948 * OOB write in BN_bn2dec()
6949
6950   The function BN_bn2dec() does not check the return value of BN_div_word().
6951   This can cause an OOB write if an application uses this function with an
6952   overly large BIGNUM. This could be a problem if an overly large certificate
6953   or CRL is printed out from an untrusted source. TLS is not affected because
6954   record limits will reject an oversized certificate before it is parsed.
6955
6956   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6957   ([CVE-2016-2182])
6958
6959   *Stephen Henson*
6960
6961 * OOB read in TS_OBJ_print_bio()
6962
6963   The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is
6964   the total length the OID text representation would use and not the amount
6965   of data written. This will result in OOB reads when large OIDs are
6966   presented.
6967
6968   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6969   ([CVE-2016-2180])
6970
6971   *Stephen Henson*
6972
6973 * Pointer arithmetic undefined behaviour
6974
6975   Avoid some undefined pointer arithmetic
6976
6977   A common idiom in the codebase is to check limits in the following manner:
6978   "p + len > limit"
6979
6980   Where "p" points to some malloc'd data of SIZE bytes and
6981   limit == p + SIZE
6982
6983   "len" here could be from some externally supplied data (e.g. from a TLS
6984   message).
6985
6986   The rules of C pointer arithmetic are such that "p + len" is only well
6987   defined where len <= SIZE. Therefore the above idiom is actually
6988   undefined behaviour.
6989
6990   For example this could cause problems if some malloc implementation
6991   provides an address for "p" such that "p + len" actually overflows for
6992   values of len that are too big and therefore p + len < limit.
6993
6994   This issue was reported to OpenSSL by Guido Vranken
6995   ([CVE-2016-2177])
6996
6997   *Matt Caswell*
6998
6999 * Constant time flag not preserved in DSA signing
7000
7001   Operations in the DSA signing algorithm should run in constant time in
7002   order to avoid side channel attacks. A flaw in the OpenSSL DSA
7003   implementation means that a non-constant time codepath is followed for
7004   certain operations. This has been demonstrated through a cache-timing
7005   attack to be sufficient for an attacker to recover the private DSA key.
7006
7007   This issue was reported by César Pereida (Aalto University), Billy Brumley
7008   (Tampere University of Technology), and Yuval Yarom (The University of
7009   Adelaide and NICTA).
7010   ([CVE-2016-2178])
7011
7012   *César Pereida*
7013
7014 * DTLS buffered message DoS
7015
7016   In a DTLS connection where handshake messages are delivered out-of-order
7017   those messages that OpenSSL is not yet ready to process will be buffered
7018   for later use. Under certain circumstances, a flaw in the logic means that
7019   those messages do not get removed from the buffer even though the handshake
7020   has been completed. An attacker could force up to approx. 15 messages to
7021   remain in the buffer when they are no longer required. These messages will
7022   be cleared when the DTLS connection is closed. The default maximum size for
7023   a message is 100k. Therefore the attacker could force an additional 1500k
7024   to be consumed per connection. By opening many simulataneous connections an
7025   attacker could cause a DoS attack through memory exhaustion.
7026
7027   This issue was reported to OpenSSL by Quan Luo.
7028   ([CVE-2016-2179])
7029
7030   *Matt Caswell*
7031
7032 * DTLS replay protection DoS
7033
7034   A flaw in the DTLS replay attack protection mechanism means that records
7035   that arrive for future epochs update the replay protection "window" before
7036   the MAC for the record has been validated. This could be exploited by an
7037   attacker by sending a record for the next epoch (which does not have to
7038   decrypt or have a valid MAC), with a very large sequence number. This means
7039   that all subsequent legitimate packets are dropped causing a denial of
7040   service for a specific DTLS connection.
7041
7042   This issue was reported to OpenSSL by the OCAP audit team.
7043   ([CVE-2016-2181])
7044
7045   *Matt Caswell*
7046
7047 * Certificate message OOB reads
7048
7049   In OpenSSL 1.0.2 and earlier some missing message length checks can result
7050   in OOB reads of up to 2 bytes beyond an allocated buffer. There is a
7051   theoretical DoS risk but this has not been observed in practice on common
7052   platforms.
7053
7054   The messages affected are client certificate, client certificate request
7055   and server certificate. As a result the attack can only be performed
7056   against a client or a server which enables client authentication.
7057
7058   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
7059   ([CVE-2016-6306])
7060
7061   *Stephen Henson*
7062
7063### Changes between 1.0.1s and 1.0.1t [3 May 2016]
7064
7065 * Prevent padding oracle in AES-NI CBC MAC check
7066
7067   A MITM attacker can use a padding oracle attack to decrypt traffic
7068   when the connection uses an AES CBC cipher and the server support
7069   AES-NI.
7070
7071   This issue was introduced as part of the fix for Lucky 13 padding
7072   attack ([CVE-2013-0169]). The padding check was rewritten to be in
7073   constant time by making sure that always the same bytes are read and
7074   compared against either the MAC or padding bytes. But it no longer
7075   checked that there was enough data to have both the MAC and padding
7076   bytes.
7077
7078   This issue was reported by Juraj Somorovsky using TLS-Attacker.
7079   ([CVE-2016-2107])
7080
7081   *Kurt Roeckx*
7082
7083 * Fix EVP_EncodeUpdate overflow
7084
7085   An overflow can occur in the EVP_EncodeUpdate() function which is used for
7086   Base64 encoding of binary data. If an attacker is able to supply very large
7087   amounts of input data then a length check can overflow resulting in a heap
7088   corruption.
7089
7090   Internally to OpenSSL the EVP_EncodeUpdate() function is primarly used by
7091   the `PEM_write_bio*` family of functions. These are mainly used within the
7092   OpenSSL command line applications, so any application which processes data
7093   from an untrusted source and outputs it as a PEM file should be considered
7094   vulnerable to this issue. User applications that call these APIs directly
7095   with large amounts of untrusted data may also be vulnerable.
7096
7097   This issue was reported by Guido Vranken.
7098   ([CVE-2016-2105])
7099
7100   *Matt Caswell*
7101
7102 * Fix EVP_EncryptUpdate overflow
7103
7104   An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
7105   is able to supply very large amounts of input data after a previous call to
7106   EVP_EncryptUpdate() with a partial block then a length check can overflow
7107   resulting in a heap corruption. Following an analysis of all OpenSSL
7108   internal usage of the EVP_EncryptUpdate() function all usage is one of two
7109   forms. The first form is where the EVP_EncryptUpdate() call is known to be
7110   the first called function after an EVP_EncryptInit(), and therefore that
7111   specific call must be safe. The second form is where the length passed to
7112   EVP_EncryptUpdate() can be seen from the code to be some small value and
7113   therefore there is no possibility of an overflow. Since all instances are
7114   one of these two forms, it is believed that there can be no overflows in
7115   internal code due to this problem. It should be noted that
7116   EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
7117   Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
7118   of these calls have also been analysed too and it is believed there are no
7119   instances in internal usage where an overflow could occur.
7120
7121   This issue was reported by Guido Vranken.
7122   ([CVE-2016-2106])
7123
7124   *Matt Caswell*
7125
7126 * Prevent ASN.1 BIO excessive memory allocation
7127
7128   When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
7129   a short invalid encoding can casuse allocation of large amounts of memory
7130   potentially consuming excessive resources or exhausting memory.
7131
7132   Any application parsing untrusted data through d2i BIO functions is
7133   affected. The memory based functions such as d2i_X509() are *not* affected.
7134   Since the memory based functions are used by the TLS library, TLS
7135   applications are not affected.
7136
7137   This issue was reported by Brian Carpenter.
7138   ([CVE-2016-2109])
7139
7140   *Stephen Henson*
7141
7142 * EBCDIC overread
7143
7144   ASN1 Strings that are over 1024 bytes can cause an overread in applications
7145   using the X509_NAME_oneline() function on EBCDIC systems. This could result
7146   in arbitrary stack data being returned in the buffer.
7147
7148   This issue was reported by Guido Vranken.
7149   ([CVE-2016-2176])
7150
7151   *Matt Caswell*
7152
7153 * Modify behavior of ALPN to invoke callback after SNI/servername
7154   callback, such that updates to the SSL_CTX affect ALPN.
7155
7156   *Todd Short*
7157
7158 * Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
7159   default.
7160
7161   *Kurt Roeckx*
7162
7163 * Only remove the SSLv2 methods with the no-ssl2-method option. When the
7164   methods are enabled and ssl2 is disabled the methods return NULL.
7165
7166   *Kurt Roeckx*
7167
7168### Changes between 1.0.1r and 1.0.1s [1 Mar 2016]
7169
7170* Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
7171  Builds that are not configured with "enable-weak-ssl-ciphers" will not
7172  provide any "EXPORT" or "LOW" strength ciphers.
7173
7174  *Viktor Dukhovni*
7175
7176* Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
7177  is by default disabled at build-time.  Builds that are not configured with
7178  "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
7179  users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
7180  will need to explicitly call either of:
7181
7182      SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
7183  or
7184      SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
7185
7186  as appropriate.  Even if either of those is used, or the application
7187  explicitly uses the version-specific SSLv2_method() or its client and
7188  server variants, SSLv2 ciphers vulnerable to exhaustive search key
7189  recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
7190  ciphers, and SSLv2 56-bit DES are no longer available.
7191  ([CVE-2016-0800])
7192
7193  *Viktor Dukhovni*
7194
7195 * Fix a double-free in DSA code
7196
7197   A double free bug was discovered when OpenSSL parses malformed DSA private
7198   keys and could lead to a DoS attack or memory corruption for applications
7199   that receive DSA private keys from untrusted sources.  This scenario is
7200   considered rare.
7201
7202   This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
7203   libFuzzer.
7204   ([CVE-2016-0705])
7205
7206   *Stephen Henson*
7207
7208 * Disable SRP fake user seed to address a server memory leak.
7209
7210   Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
7211
7212   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
7213   In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
7214   was changed to ignore the "fake user" SRP seed, even if the seed
7215   is configured.
7216
7217   Users should use SRP_VBASE_get1_by_user instead. Note that in
7218   SRP_VBASE_get1_by_user, caller must free the returned value. Note
7219   also that even though configuring the SRP seed attempts to hide
7220   invalid usernames by continuing the handshake with fake
7221   credentials, this behaviour is not constant time and no strong
7222   guarantees are made that the handshake is indistinguishable from
7223   that of a valid user.
7224   ([CVE-2016-0798])
7225
7226   *Emilia Käsper*
7227
7228 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
7229
7230   In the BN_hex2bn function the number of hex digits is calculated using an
7231   int value `i`. Later `bn_expand` is called with a value of `i * 4`. For
7232   large values of `i` this can result in `bn_expand` not allocating any
7233   memory because `i * 4` is negative. This can leave the internal BIGNUM data
7234   field as NULL leading to a subsequent NULL ptr deref. For very large values
7235   of `i`, the calculation `i * 4` could be a positive value smaller than `i`.
7236   In this case memory is allocated to the internal BIGNUM data field, but it
7237   is insufficiently sized leading to heap corruption. A similar issue exists
7238   in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
7239   is ever called by user applications with very large untrusted hex/dec data.
7240   This is anticipated to be a rare occurrence.
7241
7242   All OpenSSL internal usage of these functions use data that is not expected
7243   to be untrusted, e.g. config file data or application command line
7244   arguments. If user developed applications generate config file data based
7245   on untrusted data then it is possible that this could also lead to security
7246   consequences. This is also anticipated to be rare.
7247
7248   This issue was reported to OpenSSL by Guido Vranken.
7249   ([CVE-2016-0797])
7250
7251   *Matt Caswell*
7252
7253 * Fix memory issues in `BIO_*printf` functions
7254
7255   The internal `fmtstr` function used in processing a "%s" format string in
7256   the `BIO_*printf` functions could overflow while calculating the length of a
7257   string and cause an OOB read when printing very long strings.
7258
7259   Additionally the internal `doapr_outch` function can attempt to write to an
7260   OOB memory location (at an offset from the NULL pointer) in the event of a
7261   memory allocation failure. In 1.0.2 and below this could be caused where
7262   the size of a buffer to be allocated is greater than INT_MAX. E.g. this
7263   could be in processing a very long "%s" format string. Memory leaks can
7264   also occur.
7265
7266   The first issue may mask the second issue dependent on compiler behaviour.
7267   These problems could enable attacks where large amounts of untrusted data
7268   is passed to the `BIO_*printf` functions. If applications use these functions
7269   in this way then they could be vulnerable. OpenSSL itself uses these
7270   functions when printing out human-readable dumps of ASN.1 data. Therefore
7271   applications that print this data could be vulnerable if the data is from
7272   untrusted sources. OpenSSL command line applications could also be
7273   vulnerable where they print out ASN.1 data, or if untrusted data is passed
7274   as command line arguments.
7275
7276   Libssl is not considered directly vulnerable. Additionally certificates etc
7277   received via remote connections via libssl are also unlikely to be able to
7278   trigger these issues because of message size limits enforced within libssl.
7279
7280   This issue was reported to OpenSSL Guido Vranken.
7281   ([CVE-2016-0799])
7282
7283   *Matt Caswell*
7284
7285 * Side channel attack on modular exponentiation
7286
7287   A side-channel attack was found which makes use of cache-bank conflicts on
7288   the Intel Sandy-Bridge microarchitecture which could lead to the recovery
7289   of RSA keys.  The ability to exploit this issue is limited as it relies on
7290   an attacker who has control of code in a thread running on the same
7291   hyper-threaded core as the victim thread which is performing decryptions.
7292
7293   This issue was reported to OpenSSL by Yuval Yarom, The University of
7294   Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
7295   Nadia Heninger, University of Pennsylvania with more information at
7296   <http://cachebleed.info>.
7297   ([CVE-2016-0702])
7298
7299   *Andy Polyakov*
7300
7301 * Change the req command to generate a 2048-bit RSA/DSA key by default,
7302   if no keysize is specified with default_bits. This fixes an
7303   omission in an earlier change that changed all RSA/DSA key generation
7304   commands to use 2048 bits by default.
7305
7306   *Emilia Käsper*
7307
7308### Changes between 1.0.1q and 1.0.1r [28 Jan 2016]
7309
7310 * Protection for DH small subgroup attacks
7311
7312   As a precautionary measure the SSL_OP_SINGLE_DH_USE option has been
7313   switched on by default and cannot be disabled. This could have some
7314   performance impact.
7315
7316   *Matt Caswell*
7317
7318 * SSLv2 doesn't block disabled ciphers
7319
7320   A malicious client can negotiate SSLv2 ciphers that have been disabled on
7321   the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
7322   been disabled, provided that the SSLv2 protocol was not also disabled via
7323   SSL_OP_NO_SSLv2.
7324
7325   This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
7326   and Sebastian Schinzel.
7327   ([CVE-2015-3197])
7328
7329   *Viktor Dukhovni*
7330
7331 * Reject DH handshakes with parameters shorter than 1024 bits.
7332
7333   *Kurt Roeckx*
7334
7335### Changes between 1.0.1p and 1.0.1q [3 Dec 2015]
7336
7337 * Certificate verify crash with missing PSS parameter
7338
7339   The signature verification routines will crash with a NULL pointer
7340   dereference if presented with an ASN.1 signature using the RSA PSS
7341   algorithm and absent mask generation function parameter. Since these
7342   routines are used to verify certificate signature algorithms this can be
7343   used to crash any certificate verification operation and exploited in a
7344   DoS attack. Any application which performs certificate verification is
7345   vulnerable including OpenSSL clients and servers which enable client
7346   authentication.
7347
7348   This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
7349   ([CVE-2015-3194])
7350
7351   *Stephen Henson*
7352
7353 * X509_ATTRIBUTE memory leak
7354
7355   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
7356   memory. This structure is used by the PKCS#7 and CMS routines so any
7357   application which reads PKCS#7 or CMS data from untrusted sources is
7358   affected. SSL/TLS is not affected.
7359
7360   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
7361   libFuzzer.
7362   ([CVE-2015-3195])
7363
7364   *Stephen Henson*
7365
7366 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
7367   This changes the decoding behaviour for some invalid messages,
7368   though the change is mostly in the more lenient direction, and
7369   legacy behaviour is preserved as much as possible.
7370
7371   *Emilia Käsper*
7372
7373 * In DSA_generate_parameters_ex, if the provided seed is too short,
7374   use a random seed, as already documented.
7375
7376   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
7377
7378### Changes between 1.0.1o and 1.0.1p [9 Jul 2015]
7379
7380 * Alternate chains certificate forgery
7381
7382   During certificate verfification, OpenSSL will attempt to find an
7383   alternative certificate chain if the first attempt to build such a chain
7384   fails. An error in the implementation of this logic can mean that an
7385   attacker could cause certain checks on untrusted certificates to be
7386   bypassed, such as the CA flag, enabling them to use a valid leaf
7387   certificate to act as a CA and "issue" an invalid certificate.
7388
7389   This issue was reported to OpenSSL by Adam Langley/David Benjamin
7390   (Google/BoringSSL).
7391   ([CVE-2015-1793])
7392
7393   *Matt Caswell*
7394
7395 * Race condition handling PSK identify hint
7396
7397   If PSK identity hints are received by a multi-threaded client then
7398   the values are wrongly updated in the parent SSL_CTX structure. This can
7399   result in a race condition potentially leading to a double free of the
7400   identify hint data.
7401   ([CVE-2015-3196])
7402
7403   *Stephen Henson*
7404
7405### Changes between 1.0.1n and 1.0.1o [12 Jun 2015]
7406
7407 * Fix HMAC ABI incompatibility. The previous version introduced an ABI
7408   incompatibility in the handling of HMAC. The previous ABI has now been
7409   restored.
7410
7411### Changes between 1.0.1m and 1.0.1n [11 Jun 2015]
7412
7413 * Malformed ECParameters causes infinite loop
7414
7415   When processing an ECParameters structure OpenSSL enters an infinite loop
7416   if the curve specified is over a specially malformed binary polynomial
7417   field.
7418
7419   This can be used to perform denial of service against any
7420   system which processes public keys, certificate requests or
7421   certificates.  This includes TLS clients and TLS servers with
7422   client authentication enabled.
7423
7424   This issue was reported to OpenSSL by Joseph Barr-Pixton.
7425   ([CVE-2015-1788])
7426
7427   *Andy Polyakov*
7428
7429 * Exploitable out-of-bounds read in X509_cmp_time
7430
7431   X509_cmp_time does not properly check the length of the ASN1_TIME
7432   string and can read a few bytes out of bounds. In addition,
7433   X509_cmp_time accepts an arbitrary number of fractional seconds in the
7434   time string.
7435
7436   An attacker can use this to craft malformed certificates and CRLs of
7437   various sizes and potentially cause a segmentation fault, resulting in
7438   a DoS on applications that verify certificates or CRLs. TLS clients
7439   that verify CRLs are affected. TLS clients and servers with client
7440   authentication enabled may be affected if they use custom verification
7441   callbacks.
7442
7443   This issue was reported to OpenSSL by Robert Swiecki (Google), and
7444   independently by Hanno Böck.
7445   ([CVE-2015-1789])
7446
7447   *Emilia Käsper*
7448
7449 * PKCS7 crash with missing EnvelopedContent
7450
7451   The PKCS#7 parsing code does not handle missing inner EncryptedContent
7452   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
7453   with missing content and trigger a NULL pointer dereference on parsing.
7454
7455   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
7456   structures from untrusted sources are affected. OpenSSL clients and
7457   servers are not affected.
7458
7459   This issue was reported to OpenSSL by Michal Zalewski (Google).
7460   ([CVE-2015-1790])
7461
7462   *Emilia Käsper*
7463
7464 * CMS verify infinite loop with unknown hash function
7465
7466   When verifying a signedData message the CMS code can enter an infinite loop
7467   if presented with an unknown hash function OID. This can be used to perform
7468   denial of service against any system which verifies signedData messages using
7469   the CMS code.
7470   This issue was reported to OpenSSL by Johannes Bauer.
7471   ([CVE-2015-1792])
7472
7473   *Stephen Henson*
7474
7475 * Race condition handling NewSessionTicket
7476
7477   If a NewSessionTicket is received by a multi-threaded client when attempting to
7478   reuse a previous ticket then a race condition can occur potentially leading to
7479   a double free of the ticket data.
7480   ([CVE-2015-1791])
7481
7482   *Matt Caswell*
7483
7484 * Reject DH handshakes with parameters shorter than 768 bits.
7485
7486   *Kurt Roeckx and Emilia Kasper*
7487
7488 * dhparam: generate 2048-bit parameters by default.
7489
7490   *Kurt Roeckx and Emilia Kasper*
7491
7492### Changes between 1.0.1l and 1.0.1m [19 Mar 2015]
7493
7494 * Segmentation fault in ASN1_TYPE_cmp fix
7495
7496   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
7497   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
7498   certificate signature algorithm consistency this can be used to crash any
7499   certificate verification operation and exploited in a DoS attack. Any
7500   application which performs certificate verification is vulnerable including
7501   OpenSSL clients and servers which enable client authentication.
7502   ([CVE-2015-0286])
7503
7504   *Stephen Henson*
7505
7506 * ASN.1 structure reuse memory corruption fix
7507
7508   Reusing a structure in ASN.1 parsing may allow an attacker to cause
7509   memory corruption via an invalid write. Such reuse is and has been
7510   strongly discouraged and is believed to be rare.
7511
7512   Applications that parse structures containing CHOICE or ANY DEFINED BY
7513   components may be affected. Certificate parsing (d2i_X509 and related
7514   functions) are however not affected. OpenSSL clients and servers are
7515   not affected.
7516   ([CVE-2015-0287])
7517
7518   *Stephen Henson*
7519
7520 * PKCS7 NULL pointer dereferences fix
7521
7522   The PKCS#7 parsing code does not handle missing outer ContentInfo
7523   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
7524   missing content and trigger a NULL pointer dereference on parsing.
7525
7526   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
7527   otherwise parse PKCS#7 structures from untrusted sources are
7528   affected. OpenSSL clients and servers are not affected.
7529
7530   This issue was reported to OpenSSL by Michal Zalewski (Google).
7531   ([CVE-2015-0289])
7532
7533   *Emilia Käsper*
7534
7535 * DoS via reachable assert in SSLv2 servers fix
7536
7537   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
7538   servers that both support SSLv2 and enable export cipher suites by sending
7539   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
7540
7541   This issue was discovered by Sean Burford (Google) and Emilia Käsper
7542   (OpenSSL development team).
7543   ([CVE-2015-0293])
7544
7545   *Emilia Käsper*
7546
7547 * Use After Free following d2i_ECPrivatekey error fix
7548
7549   A malformed EC private key file consumed via the d2i_ECPrivateKey function
7550   could cause a use after free condition. This, in turn, could cause a double
7551   free in several private key parsing functions (such as d2i_PrivateKey
7552   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
7553   for applications that receive EC private keys from untrusted
7554   sources. This scenario is considered rare.
7555
7556   This issue was discovered by the BoringSSL project and fixed in their
7557   commit 517073cd4b.
7558   ([CVE-2015-0209])
7559
7560   *Matt Caswell*
7561
7562 * X509_to_X509_REQ NULL pointer deref fix
7563
7564   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
7565   the certificate key is invalid. This function is rarely used in practice.
7566
7567   This issue was discovered by Brian Carpenter.
7568   ([CVE-2015-0288])
7569
7570   *Stephen Henson*
7571
7572 * Removed the export ciphers from the DEFAULT ciphers
7573
7574   *Kurt Roeckx*
7575
7576### Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
7577
7578 * Build fixes for the Windows and OpenVMS platforms
7579
7580   *Matt Caswell and Richard Levitte*
7581
7582### Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
7583
7584 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
7585   message can cause a segmentation fault in OpenSSL due to a NULL pointer
7586   dereference. This could lead to a Denial Of Service attack. Thanks to
7587   Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
7588   ([CVE-2014-3571])
7589
7590   *Steve Henson*
7591
7592 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
7593   dtls1_buffer_record function under certain conditions. In particular this
7594   could occur if an attacker sent repeated DTLS records with the same
7595   sequence number but for the next epoch. The memory leak could be exploited
7596   by an attacker in a Denial of Service attack through memory exhaustion.
7597   Thanks to Chris Mueller for reporting this issue.
7598   ([CVE-2015-0206])
7599
7600   *Matt Caswell*
7601
7602 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
7603   built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
7604   method would be set to NULL which could later result in a NULL pointer
7605   dereference. Thanks to Frank Schmirler for reporting this issue.
7606   ([CVE-2014-3569])
7607
7608   *Kurt Roeckx*
7609
7610 * Abort handshake if server key exchange message is omitted for ephemeral
7611   ECDH ciphersuites.
7612
7613   Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
7614   reporting this issue.
7615   ([CVE-2014-3572])
7616
7617   *Steve Henson*
7618
7619 * Remove non-export ephemeral RSA code on client and server. This code
7620   violated the TLS standard by allowing the use of temporary RSA keys in
7621   non-export ciphersuites and could be used by a server to effectively
7622   downgrade the RSA key length used to a value smaller than the server
7623   certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
7624   INRIA or reporting this issue.
7625   ([CVE-2015-0204])
7626
7627   *Steve Henson*
7628
7629 * Fixed issue where DH client certificates are accepted without verification.
7630   An OpenSSL server will accept a DH certificate for client authentication
7631   without the certificate verify message. This effectively allows a client to
7632   authenticate without the use of a private key. This only affects servers
7633   which trust a client certificate authority which issues certificates
7634   containing DH keys: these are extremely rare and hardly ever encountered.
7635   Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
7636   this issue.
7637   ([CVE-2015-0205])
7638
7639   *Steve Henson*
7640
7641 * Ensure that the session ID context of an SSL is updated when its
7642   SSL_CTX is updated via SSL_set_SSL_CTX.
7643
7644   The session ID context is typically set from the parent SSL_CTX,
7645   and can vary with the CTX.
7646
7647   *Adam Langley*
7648
7649 * Fix various certificate fingerprint issues.
7650
7651   By using non-DER or invalid encodings outside the signed portion of a
7652   certificate the fingerprint can be changed without breaking the signature.
7653   Although no details of the signed portion of the certificate can be changed
7654   this can cause problems with some applications: e.g. those using the
7655   certificate fingerprint for blacklists.
7656
7657   1. Reject signatures with non zero unused bits.
7658
7659   If the BIT STRING containing the signature has non zero unused bits reject
7660   the signature. All current signature algorithms require zero unused bits.
7661
7662   2. Check certificate algorithm consistency.
7663
7664   Check the AlgorithmIdentifier inside TBS matches the one in the
7665   certificate signature. NB: this will result in signature failure
7666   errors for some broken certificates.
7667
7668   Thanks to Konrad Kraszewski from Google for reporting this issue.
7669
7670   3. Check DSA/ECDSA signatures use DER.
7671
7672   Re-encode DSA/ECDSA signatures and compare with the original received
7673   signature. Return an error if there is a mismatch.
7674
7675   This will reject various cases including garbage after signature
7676   (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
7677   program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
7678   (negative or with leading zeroes).
7679
7680   Further analysis was conducted and fixes were developed by Stephen Henson
7681   of the OpenSSL core team.
7682
7683   ([CVE-2014-8275])
7684
7685   *Steve Henson*
7686
7687 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
7688   results on some platforms, including x86_64. This bug occurs at random
7689   with a very low probability, and is not known to be exploitable in any
7690   way, though its exact impact is difficult to determine. Thanks to Pieter
7691   Wuille (Blockstream) who reported this issue and also suggested an initial
7692   fix. Further analysis was conducted by the OpenSSL development team and
7693   Adam Langley of Google. The final fix was developed by Andy Polyakov of
7694   the OpenSSL core team.
7695   ([CVE-2014-3570])
7696
7697   *Andy Polyakov*
7698
7699 * Do not resume sessions on the server if the negotiated protocol
7700   version does not match the session's version. Resuming with a different
7701   version, while not strictly forbidden by the RFC, is of questionable
7702   sanity and breaks all known clients.
7703
7704   *David Benjamin, Emilia Käsper*
7705
7706 * Tighten handling of the ChangeCipherSpec (CCS) message: reject
7707   early CCS messages during renegotiation. (Note that because
7708   renegotiation is encrypted, this early CCS was not exploitable.)
7709
7710   *Emilia Käsper*
7711
7712 * Tighten client-side session ticket handling during renegotiation:
7713   ensure that the client only accepts a session ticket if the server sends
7714   the extension anew in the ServerHello. Previously, a TLS client would
7715   reuse the old extension state and thus accept a session ticket if one was
7716   announced in the initial ServerHello.
7717
7718   Similarly, ensure that the client requires a session ticket if one
7719   was advertised in the ServerHello. Previously, a TLS client would
7720   ignore a missing NewSessionTicket message.
7721
7722   *Emilia Käsper*
7723
7724### Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
7725
7726 * SRTP Memory Leak.
7727
7728   A flaw in the DTLS SRTP extension parsing code allows an attacker, who
7729   sends a carefully crafted handshake message, to cause OpenSSL to fail
7730   to free up to 64k of memory causing a memory leak. This could be
7731   exploited in a Denial Of Service attack. This issue affects OpenSSL
7732   1.0.1 server implementations for both SSL/TLS and DTLS regardless of
7733   whether SRTP is used or configured. Implementations of OpenSSL that
7734   have been compiled with OPENSSL_NO_SRTP defined are not affected.
7735
7736   The fix was developed by the OpenSSL team.
7737   ([CVE-2014-3513])
7738
7739   *OpenSSL team*
7740
7741 * Session Ticket Memory Leak.
7742
7743   When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
7744   integrity of that ticket is first verified. In the event of a session
7745   ticket integrity check failing, OpenSSL will fail to free memory
7746   causing a memory leak. By sending a large number of invalid session
7747   tickets an attacker could exploit this issue in a Denial Of Service
7748   attack.
7749   ([CVE-2014-3567])
7750
7751   *Steve Henson*
7752
7753 * Build option no-ssl3 is incomplete.
7754
7755   When OpenSSL is configured with "no-ssl3" as a build option, servers
7756   could accept and complete a SSL 3.0 handshake, and clients could be
7757   configured to send them.
7758   ([CVE-2014-3568])
7759
7760   *Akamai and the OpenSSL team*
7761
7762 * Add support for TLS_FALLBACK_SCSV.
7763   Client applications doing fallback retries should call
7764   SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
7765   ([CVE-2014-3566])
7766
7767   *Adam Langley, Bodo Moeller*
7768
7769 * Add additional DigestInfo checks.
7770
7771   Re-encode DigestInto in DER and check against the original when
7772   verifying RSA signature: this will reject any improperly encoded
7773   DigestInfo structures.
7774
7775   Note: this is a precautionary measure and no attacks are currently known.
7776
7777   *Steve Henson*
7778
7779### Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
7780
7781 * Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
7782   SRP code can be overrun an internal buffer. Add sanity check that
7783   g, A, B < N to SRP code.
7784
7785   Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
7786   Group for discovering this issue.
7787   ([CVE-2014-3512])
7788
7789   *Steve Henson*
7790
7791 * A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
7792   TLS 1.0 instead of higher protocol versions when the ClientHello message
7793   is badly fragmented. This allows a man-in-the-middle attacker to force a
7794   downgrade to TLS 1.0 even if both the server and the client support a
7795   higher protocol version, by modifying the client's TLS records.
7796
7797   Thanks to David Benjamin and Adam Langley (Google) for discovering and
7798   researching this issue.
7799   ([CVE-2014-3511])
7800
7801   *David Benjamin*
7802
7803 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
7804   to a denial of service attack. A malicious server can crash the client
7805   with a null pointer dereference (read) by specifying an anonymous (EC)DH
7806   ciphersuite and sending carefully crafted handshake messages.
7807
7808   Thanks to Felix Gröbert (Google) for discovering and researching this
7809   issue.
7810   ([CVE-2014-3510])
7811
7812   *Emilia Käsper*
7813
7814 * By sending carefully crafted DTLS packets an attacker could cause openssl
7815   to leak memory. This can be exploited through a Denial of Service attack.
7816   Thanks to Adam Langley for discovering and researching this issue.
7817   ([CVE-2014-3507])
7818
7819   *Adam Langley*
7820
7821 * An attacker can force openssl to consume large amounts of memory whilst
7822   processing DTLS handshake messages. This can be exploited through a
7823   Denial of Service attack.
7824   Thanks to Adam Langley for discovering and researching this issue.
7825   ([CVE-2014-3506])
7826
7827   *Adam Langley*
7828
7829 * An attacker can force an error condition which causes openssl to crash
7830   whilst processing DTLS packets due to memory being freed twice. This
7831   can be exploited through a Denial of Service attack.
7832   Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
7833   this issue.
7834   ([CVE-2014-3505])
7835
7836   *Adam Langley*
7837
7838 * If a multithreaded client connects to a malicious server using a resumed
7839   session and the server sends an ec point format extension it could write
7840   up to 255 bytes to freed memory.
7841
7842   Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
7843   issue.
7844   ([CVE-2014-3509])
7845
7846   *Gabor Tyukasz*
7847
7848 * A malicious server can crash an OpenSSL client with a null pointer
7849   dereference (read) by specifying an SRP ciphersuite even though it was not
7850   properly negotiated with the client. This can be exploited through a
7851   Denial of Service attack.
7852
7853   Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for
7854   discovering and researching this issue.
7855   ([CVE-2014-5139])
7856
7857   *Steve Henson*
7858
7859 * A flaw in OBJ_obj2txt may cause pretty printing functions such as
7860   X509_name_oneline, X509_name_print_ex et al. to leak some information
7861   from the stack. Applications may be affected if they echo pretty printing
7862   output to the attacker.
7863
7864   Thanks to Ivan Fratric (Google) for discovering this issue.
7865   ([CVE-2014-3508])
7866
7867   *Emilia Käsper, and Steve Henson*
7868
7869 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
7870   for corner cases. (Certain input points at infinity could lead to
7871   bogus results, with non-infinity inputs mapped to infinity too.)
7872
7873   *Bodo Moeller*
7874
7875### Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
7876
7877 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
7878   handshake can force the use of weak keying material in OpenSSL
7879   SSL/TLS clients and servers.
7880
7881   Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
7882   researching this issue. ([CVE-2014-0224])
7883
7884   *KIKUCHI Masashi, Steve Henson*
7885
7886 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
7887   OpenSSL DTLS client the code can be made to recurse eventually crashing
7888   in a DoS attack.
7889
7890   Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
7891   ([CVE-2014-0221])
7892
7893   *Imre Rad, Steve Henson*
7894
7895 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
7896   be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
7897   client or server. This is potentially exploitable to run arbitrary
7898   code on a vulnerable client or server.
7899
7900   Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195])
7901
7902   *Jüri Aedla, Steve Henson*
7903
7904 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
7905   are subject to a denial of service attack.
7906
7907   Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
7908   this issue. ([CVE-2014-3470])
7909
7910   *Felix Gröbert, Ivan Fratric, Steve Henson*
7911
7912 * Harmonize version and its documentation. -f flag is used to display
7913   compilation flags.
7914
7915   *mancha <mancha1@zoho.com>*
7916
7917 * Fix eckey_priv_encode so it immediately returns an error upon a failure
7918   in i2d_ECPrivateKey.
7919
7920   *mancha <mancha1@zoho.com>*
7921
7922 * Fix some double frees. These are not thought to be exploitable.
7923
7924   *mancha <mancha1@zoho.com>*
7925
7926### Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
7927
7928 * A missing bounds check in the handling of the TLS heartbeat extension
7929   can be used to reveal up to 64k of memory to a connected client or
7930   server.
7931
7932   Thanks for Neel Mehta of Google Security for discovering this bug and to
7933   Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
7934   preparing the fix ([CVE-2014-0160])
7935
7936   *Adam Langley, Bodo Moeller*
7937
7938 * Fix for the attack described in the paper "Recovering OpenSSL
7939   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
7940   by Yuval Yarom and Naomi Benger. Details can be obtained from:
7941   <http://eprint.iacr.org/2014/140>
7942
7943   Thanks to Yuval Yarom and Naomi Benger for discovering this
7944   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
7945
7946   *Yuval Yarom and Naomi Benger*
7947
7948 * TLS pad extension: draft-agl-tls-padding-03
7949
7950   Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
7951   TLS client Hello record length value would otherwise be > 255 and
7952   less that 512 pad with a dummy extension containing zeroes so it
7953   is at least 512 bytes long.
7954
7955   *Adam Langley, Steve Henson*
7956
7957### Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
7958
7959 * Fix for TLS record tampering bug. A carefully crafted invalid
7960   handshake could crash OpenSSL with a NULL pointer exception.
7961   Thanks to Anton Johansson for reporting this issues.
7962   ([CVE-2013-4353])
7963
7964 * Keep original DTLS digest and encryption contexts in retransmission
7965   structures so we can use the previous session parameters if they need
7966   to be resent. ([CVE-2013-6450])
7967
7968   *Steve Henson*
7969
7970 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
7971   avoids preferring ECDHE-ECDSA ciphers when the client appears to be
7972   Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
7973   several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
7974   is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
7975   10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
7976
7977   *Rob Stradling, Adam Langley*
7978
7979### Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
7980
7981 * Correct fix for CVE-2013-0169. The original didn't work on AES-NI
7982   supporting platforms or when small records were transferred.
7983
7984   *Andy Polyakov, Steve Henson*
7985
7986### Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
7987
7988 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
7989
7990   This addresses the flaw in CBC record processing discovered by
7991   Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
7992   at: <http://www.isg.rhul.ac.uk/tls/>
7993
7994   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
7995   Security Group at Royal Holloway, University of London
7996   (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
7997   Emilia Käsper for the initial patch.
7998   ([CVE-2013-0169])
7999
8000   *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson*
8001
8002 * Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
8003   ciphersuites which can be exploited in a denial of service attack.
8004   Thanks go to and to Adam Langley <agl@chromium.org> for discovering
8005   and detecting this bug and to Wolfgang Ettlinger
8006   <wolfgang.ettlinger@gmail.com> for independently discovering this issue.
8007   ([CVE-2012-2686])
8008
8009   *Adam Langley*
8010
8011 * Return an error when checking OCSP signatures when key is NULL.
8012   This fixes a DoS attack. ([CVE-2013-0166])
8013
8014   *Steve Henson*
8015
8016 * Make openssl verify return errors.
8017
8018   *Chris Palmer <palmer@google.com> and Ben Laurie*
8019
8020 * Call OCSP Stapling callback after ciphersuite has been chosen, so
8021   the right response is stapled. Also change SSL_get_certificate()
8022   so it returns the certificate actually sent.
8023   See <http://rt.openssl.org/Ticket/Display.html?id=2836>.
8024
8025   *Rob Stradling <rob.stradling@comodo.com>*
8026
8027 * Fix possible deadlock when decoding public keys.
8028
8029   *Steve Henson*
8030
8031 * Don't use TLS 1.0 record version number in initial client hello
8032   if renegotiating.
8033
8034   *Steve Henson*
8035
8036### Changes between 1.0.1b and 1.0.1c [10 May 2012]
8037
8038 * Sanity check record length before skipping explicit IV in TLS
8039   1.2, 1.1 and DTLS to fix DoS attack.
8040
8041   Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
8042   fuzzing as a service testing platform.
8043   ([CVE-2012-2333])
8044
8045   *Steve Henson*
8046
8047 * Initialise tkeylen properly when encrypting CMS messages.
8048   Thanks to Solar Designer of Openwall for reporting this issue.
8049
8050   *Steve Henson*
8051
8052 * In FIPS mode don't try to use composite ciphers as they are not
8053   approved.
8054
8055   *Steve Henson*
8056
8057### Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
8058
8059 * OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
8060   1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
8061   mean any application compiled against OpenSSL 1.0.0 headers setting
8062   SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling
8063   TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
8064   0x10000000L Any application which was previously compiled against
8065   OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
8066   will need to be recompiled as a result. Letting be results in
8067   inability to disable specifically TLS 1.1 and in client context,
8068   in unlike event, limit maximum offered version to TLS 1.0 [see below].
8069
8070   *Steve Henson*
8071
8072 * In order to ensure interoperability SSL_OP_NO_protocolX does not
8073   disable just protocol X, but all protocols above X *if* there are
8074   protocols *below* X still enabled. In more practical terms it means
8075   that if application wants to disable TLS1.0 in favor of TLS1.1 and
8076   above, it's not sufficient to pass `SSL_OP_NO_TLSv1`, one has to pass
8077   `SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2`. This applies to
8078   client side.
8079
8080   *Andy Polyakov*
8081
8082### Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
8083
8084 * Check for potentially exploitable overflows in asn1_d2i_read_bio
8085   BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
8086   in CRYPTO_realloc_clean.
8087
8088   Thanks to Tavis Ormandy, Google Security Team, for discovering this
8089   issue and to Adam Langley <agl@chromium.org> for fixing it.
8090   ([CVE-2012-2110])
8091
8092   *Adam Langley (Google), Tavis Ormandy, Google Security Team*
8093
8094 * Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
8095
8096   *Adam Langley*
8097
8098 * Workarounds for some broken servers that "hang" if a client hello
8099   record length exceeds 255 bytes.
8100
8101   1. Do not use record version number > TLS 1.0 in initial client
8102      hello: some (but not all) hanging servers will now work.
8103   2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
8104      the number of ciphers sent in the client hello. This should be
8105      set to an even number, such as 50, for example by passing:
8106      -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
8107      Most broken servers should now work.
8108   3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
8109      TLS 1.2 client support entirely.
8110
8111   *Steve Henson*
8112
8113 * Fix SEGV in Vector Permutation AES module observed in OpenSSH.
8114
8115   *Andy Polyakov*
8116
8117### Changes between 1.0.0h and 1.0.1  [14 Mar 2012]
8118
8119 * Add compatibility with old MDC2 signatures which use an ASN1 OCTET
8120   STRING form instead of a DigestInfo.
8121
8122   *Steve Henson*
8123
8124 * The format used for MDC2 RSA signatures is inconsistent between EVP
8125   and the RSA_sign/RSA_verify functions. This was made more apparent when
8126   OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
8127   those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect
8128   the correct format in RSA_verify so both forms transparently work.
8129
8130   *Steve Henson*
8131
8132 * Some servers which support TLS 1.0 can choke if we initially indicate
8133   support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
8134   encrypted premaster secret. As a workaround use the maximum permitted
8135   client version in client hello, this should keep such servers happy
8136   and still work with previous versions of OpenSSL.
8137
8138   *Steve Henson*
8139
8140 * Add support for TLS/DTLS heartbeats.
8141
8142   *Robin Seggelmann <seggelmann@fh-muenster.de>*
8143
8144 * Add support for SCTP.
8145
8146   *Robin Seggelmann <seggelmann@fh-muenster.de>*
8147
8148 * Improved PRNG seeding for VOS.
8149
8150   *Paul Green <Paul.Green@stratus.com>*
8151
8152 * Extensive assembler packs updates, most notably:
8153
8154   - x86[_64]:     AES-NI, PCLMULQDQ, RDRAND support;
8155   - x86[_64]:     SSSE3 support (SHA1, vector-permutation AES);
8156   - x86_64:       bit-sliced AES implementation;
8157   - ARM:          NEON support, contemporary platforms optimizations;
8158   - s390x:        z196 support;
8159   - `*`:            GHASH and GF(2^m) multiplication implementations;
8160
8161   *Andy Polyakov*
8162
8163 * Make TLS-SRP code conformant with RFC 5054 API cleanup
8164   (removal of unnecessary code)
8165
8166   *Peter Sylvester <peter.sylvester@edelweb.fr>*
8167
8168 * Add TLS key material exporter from RFC 5705.
8169
8170   *Eric Rescorla*
8171
8172 * Add DTLS-SRTP negotiation from RFC 5764.
8173
8174   *Eric Rescorla*
8175
8176 * Add Next Protocol Negotiation,
8177   <http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00>. Can be
8178   disabled with a no-npn flag to config or Configure. Code donated
8179   by Google.
8180
8181   *Adam Langley <agl@google.com> and Ben Laurie*
8182
8183 * Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
8184   NIST-P256, NIST-P521, with constant-time single point multiplication on
8185   typical inputs. Compiler support for the nonstandard type `__uint128_t` is
8186   required to use this (present in gcc 4.4 and later, for 64-bit builds).
8187   Code made available under Apache License version 2.0.
8188
8189   Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
8190   line to include this in your build of OpenSSL, and run "make depend" (or
8191   "make update"). This enables the following EC_METHODs:
8192
8193           EC_GFp_nistp224_method()
8194           EC_GFp_nistp256_method()
8195           EC_GFp_nistp521_method()
8196
8197   EC_GROUP_new_by_curve_name() will automatically use these (while
8198   EC_GROUP_new_curve_GFp() currently prefers the more flexible
8199   implementations).
8200
8201   *Emilia Käsper, Adam Langley, Bodo Moeller (Google)*
8202
8203 * Use type ossl_ssize_t instead of ssize_t which isn't available on
8204   all platforms. Move ssize_t definition from e_os.h to the public
8205   header file e_os2.h as it now appears in public header file cms.h
8206
8207   *Steve Henson*
8208
8209 * New -sigopt option to the ca, req and x509 utilities. Additional
8210   signature parameters can be passed using this option and in
8211   particular PSS.
8212
8213   *Steve Henson*
8214
8215 * Add RSA PSS signing function. This will generate and set the
8216   appropriate AlgorithmIdentifiers for PSS based on those in the
8217   corresponding EVP_MD_CTX structure. No application support yet.
8218
8219   *Steve Henson*
8220
8221 * Support for companion algorithm specific ASN1 signing routines.
8222   New function ASN1_item_sign_ctx() signs a pre-initialised
8223   EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
8224   the appropriate parameters.
8225
8226   *Steve Henson*
8227
8228 * Add new algorithm specific ASN1 verification initialisation function
8229   to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
8230   handling will be the same no matter what EVP_PKEY_METHOD is used.
8231   Add a PSS handler to support verification of PSS signatures: checked
8232   against a number of sample certificates.
8233
8234   *Steve Henson*
8235
8236 * Add signature printing for PSS. Add PSS OIDs.
8237
8238   *Steve Henson, Martin Kaiser <lists@kaiser.cx>*
8239
8240 * Add algorithm specific signature printing. An individual ASN1 method
8241   can now print out signatures instead of the standard hex dump.
8242
8243   More complex signatures (e.g. PSS) can print out more meaningful
8244   information. Include DSA version that prints out the signature
8245   parameters r, s.
8246
8247   *Steve Henson*
8248
8249 * Password based recipient info support for CMS library: implementing
8250   RFC3211.
8251
8252   *Steve Henson*
8253
8254 * Split password based encryption into PBES2 and PBKDF2 functions. This
8255   neatly separates the code into cipher and PBE sections and is required
8256   for some algorithms that split PBES2 into separate pieces (such as
8257   password based CMS).
8258
8259   *Steve Henson*
8260
8261 * Session-handling fixes:
8262   - Fix handling of connections that are resuming with a session ID,
8263     but also support Session Tickets.
8264   - Fix a bug that suppressed issuing of a new ticket if the client
8265     presented a ticket with an expired session.
8266   - Try to set the ticket lifetime hint to something reasonable.
8267   - Make tickets shorter by excluding irrelevant information.
8268   - On the client side, don't ignore renewed tickets.
8269
8270   *Adam Langley, Bodo Moeller (Google)*
8271
8272 * Fix PSK session representation.
8273
8274   *Bodo Moeller*
8275
8276 * Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
8277
8278   This work was sponsored by Intel.
8279
8280   *Andy Polyakov*
8281
8282 * Add GCM support to TLS library. Some custom code is needed to split
8283   the IV between the fixed (from PRF) and explicit (from TLS record)
8284   portions. This adds all GCM ciphersuites supported by RFC5288 and
8285   RFC5289. Generalise some `AES*` cipherstrings to include GCM and
8286   add a special AESGCM string for GCM only.
8287
8288   *Steve Henson*
8289
8290 * Expand range of ctrls for AES GCM. Permit setting invocation
8291   field on decrypt and retrieval of invocation field only on encrypt.
8292
8293   *Steve Henson*
8294
8295 * Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
8296   As required by RFC5289 these ciphersuites cannot be used if for
8297   versions of TLS earlier than 1.2.
8298
8299   *Steve Henson*
8300
8301 * For FIPS capable OpenSSL interpret a NULL default public key method
8302   as unset and return the appropriate default but do *not* set the default.
8303   This means we can return the appropriate method in applications that
8304   switch between FIPS and non-FIPS modes.
8305
8306   *Steve Henson*
8307
8308 * Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
8309   ENGINE is used then we cannot handle that in the FIPS module so we
8310   keep original code iff non-FIPS operations are allowed.
8311
8312   *Steve Henson*
8313
8314 * Add -attime option to openssl utilities.
8315
8316   *Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson*
8317
8318 * Redirect DSA and DH operations to FIPS module in FIPS mode.
8319
8320   *Steve Henson*
8321
8322 * Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
8323   FIPS EC methods unconditionally for now.
8324
8325   *Steve Henson*
8326
8327 * New build option no-ec2m to disable characteristic 2 code.
8328
8329   *Steve Henson*
8330
8331 * Backport libcrypto audit of return value checking from 1.1.0-dev; not
8332   all cases can be covered as some introduce binary incompatibilities.
8333
8334   *Steve Henson*
8335
8336 * Redirect RSA operations to FIPS module including keygen,
8337   encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
8338
8339   *Steve Henson*
8340
8341 * Add similar low-level API blocking to ciphers.
8342
8343   *Steve Henson*
8344
8345 * low-level digest APIs are not approved in FIPS mode: any attempt
8346   to use these will cause a fatal error. Applications that *really* want
8347   to use them can use the `private_*` version instead.
8348
8349   *Steve Henson*
8350
8351 * Redirect cipher operations to FIPS module for FIPS builds.
8352
8353   *Steve Henson*
8354
8355 * Redirect digest operations to FIPS module for FIPS builds.
8356
8357   *Steve Henson*
8358
8359 * Update build system to add "fips" flag which will link in fipscanister.o
8360   for static and shared library builds embedding a signature if needed.
8361
8362   *Steve Henson*
8363
8364 * Output TLS supported curves in preference order instead of numerical
8365   order. This is currently hardcoded for the highest order curves first.
8366   This should be configurable so applications can judge speed vs strength.
8367
8368   *Steve Henson*
8369
8370 * Add TLS v1.2 server support for client authentication.
8371
8372   *Steve Henson*
8373
8374 * Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
8375   and enable MD5.
8376
8377   *Steve Henson*
8378
8379 * Functions FIPS_mode_set() and FIPS_mode() which call the underlying
8380   FIPS modules versions.
8381
8382   *Steve Henson*
8383
8384 * Add TLS v1.2 client side support for client authentication. Keep cache
8385   of handshake records longer as we don't know the hash algorithm to use
8386   until after the certificate request message is received.
8387
8388   *Steve Henson*
8389
8390 * Initial TLS v1.2 client support. Add a default signature algorithms
8391   extension including all the algorithms we support. Parse new signature
8392   format in client key exchange. Relax some ECC signing restrictions for
8393   TLS v1.2 as indicated in RFC5246.
8394
8395   *Steve Henson*
8396
8397 * Add server support for TLS v1.2 signature algorithms extension. Switch
8398   to new signature format when needed using client digest preference.
8399   All server ciphersuites should now work correctly in TLS v1.2. No client
8400   support yet and no support for client certificates.
8401
8402   *Steve Henson*
8403
8404 * Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
8405   to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
8406   ciphersuites. At present only RSA key exchange ciphersuites work with
8407   TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
8408   SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
8409   and version checking.
8410
8411   *Steve Henson*
8412
8413 * New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
8414   with this defined it will not be affected by any changes to ssl internal
8415   structures. Add several utility functions to allow openssl application
8416   to work with OPENSSL_NO_SSL_INTERN defined.
8417
8418   *Steve Henson*
8419
8420 * A long standing patch to add support for SRP from EdelWeb (Peter
8421   Sylvester and Christophe Renou) was integrated.
8422   *Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester
8423   <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and
8424   Ben Laurie*
8425
8426 * Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
8427
8428   *Steve Henson*
8429
8430 * Permit abbreviated handshakes when renegotiating using the function
8431   SSL_renegotiate_abbreviated().
8432
8433   *Robin Seggelmann <seggelmann@fh-muenster.de>*
8434
8435 * Add call to ENGINE_register_all_complete() to
8436   ENGINE_load_builtin_engines(), so some implementations get used
8437   automatically instead of needing explicit application support.
8438
8439   *Steve Henson*
8440
8441 * Add support for TLS key exporter as described in RFC5705.
8442
8443   *Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson*
8444
8445 * Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
8446   a few changes are required:
8447
8448     Add SSL_OP_NO_TLSv1_1 flag.
8449     Add TLSv1_1 methods.
8450     Update version checking logic to handle version 1.1.
8451     Add explicit IV handling (ported from DTLS code).
8452     Add command line options to s_client/s_server.
8453
8454   *Steve Henson*
8455
8456OpenSSL 1.0.0
8457-------------
8458
8459### Changes between 1.0.0s and 1.0.0t [3 Dec 2015]
8460
8461 * X509_ATTRIBUTE memory leak
8462
8463   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
8464   memory. This structure is used by the PKCS#7 and CMS routines so any
8465   application which reads PKCS#7 or CMS data from untrusted sources is
8466   affected. SSL/TLS is not affected.
8467
8468   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
8469   libFuzzer.
8470   ([CVE-2015-3195])
8471
8472   *Stephen Henson*
8473
8474 * Race condition handling PSK identify hint
8475
8476   If PSK identity hints are received by a multi-threaded client then
8477   the values are wrongly updated in the parent SSL_CTX structure. This can
8478   result in a race condition potentially leading to a double free of the
8479   identify hint data.
8480   ([CVE-2015-3196])
8481
8482   *Stephen Henson*
8483
8484### Changes between 1.0.0r and 1.0.0s [11 Jun 2015]
8485
8486 * Malformed ECParameters causes infinite loop
8487
8488   When processing an ECParameters structure OpenSSL enters an infinite loop
8489   if the curve specified is over a specially malformed binary polynomial
8490   field.
8491
8492   This can be used to perform denial of service against any
8493   system which processes public keys, certificate requests or
8494   certificates.  This includes TLS clients and TLS servers with
8495   client authentication enabled.
8496
8497   This issue was reported to OpenSSL by Joseph Barr-Pixton.
8498   ([CVE-2015-1788])
8499
8500   *Andy Polyakov*
8501
8502 * Exploitable out-of-bounds read in X509_cmp_time
8503
8504   X509_cmp_time does not properly check the length of the ASN1_TIME
8505   string and can read a few bytes out of bounds. In addition,
8506   X509_cmp_time accepts an arbitrary number of fractional seconds in the
8507   time string.
8508
8509   An attacker can use this to craft malformed certificates and CRLs of
8510   various sizes and potentially cause a segmentation fault, resulting in
8511   a DoS on applications that verify certificates or CRLs. TLS clients
8512   that verify CRLs are affected. TLS clients and servers with client
8513   authentication enabled may be affected if they use custom verification
8514   callbacks.
8515
8516   This issue was reported to OpenSSL by Robert Swiecki (Google), and
8517   independently by Hanno Böck.
8518   ([CVE-2015-1789])
8519
8520   *Emilia Käsper*
8521
8522 * PKCS7 crash with missing EnvelopedContent
8523
8524   The PKCS#7 parsing code does not handle missing inner EncryptedContent
8525   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
8526   with missing content and trigger a NULL pointer dereference on parsing.
8527
8528   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
8529   structures from untrusted sources are affected. OpenSSL clients and
8530   servers are not affected.
8531
8532   This issue was reported to OpenSSL by Michal Zalewski (Google).
8533   ([CVE-2015-1790])
8534
8535   *Emilia Käsper*
8536
8537 * CMS verify infinite loop with unknown hash function
8538
8539   When verifying a signedData message the CMS code can enter an infinite loop
8540   if presented with an unknown hash function OID. This can be used to perform
8541   denial of service against any system which verifies signedData messages using
8542   the CMS code.
8543   This issue was reported to OpenSSL by Johannes Bauer.
8544   ([CVE-2015-1792])
8545
8546   *Stephen Henson*
8547
8548 * Race condition handling NewSessionTicket
8549
8550   If a NewSessionTicket is received by a multi-threaded client when attempting to
8551   reuse a previous ticket then a race condition can occur potentially leading to
8552   a double free of the ticket data.
8553   ([CVE-2015-1791])
8554
8555   *Matt Caswell*
8556
8557### Changes between 1.0.0q and 1.0.0r [19 Mar 2015]
8558
8559 * Segmentation fault in ASN1_TYPE_cmp fix
8560
8561   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
8562   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
8563   certificate signature algorithm consistency this can be used to crash any
8564   certificate verification operation and exploited in a DoS attack. Any
8565   application which performs certificate verification is vulnerable including
8566   OpenSSL clients and servers which enable client authentication.
8567   ([CVE-2015-0286])
8568
8569   *Stephen Henson*
8570
8571 * ASN.1 structure reuse memory corruption fix
8572
8573   Reusing a structure in ASN.1 parsing may allow an attacker to cause
8574   memory corruption via an invalid write. Such reuse is and has been
8575   strongly discouraged and is believed to be rare.
8576
8577   Applications that parse structures containing CHOICE or ANY DEFINED BY
8578   components may be affected. Certificate parsing (d2i_X509 and related
8579   functions) are however not affected. OpenSSL clients and servers are
8580   not affected.
8581   ([CVE-2015-0287])
8582
8583   *Stephen Henson*
8584
8585 * PKCS7 NULL pointer dereferences fix
8586
8587   The PKCS#7 parsing code does not handle missing outer ContentInfo
8588   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
8589   missing content and trigger a NULL pointer dereference on parsing.
8590
8591   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
8592   otherwise parse PKCS#7 structures from untrusted sources are
8593   affected. OpenSSL clients and servers are not affected.
8594
8595   This issue was reported to OpenSSL by Michal Zalewski (Google).
8596   ([CVE-2015-0289])
8597
8598   *Emilia Käsper*
8599
8600 * DoS via reachable assert in SSLv2 servers fix
8601
8602   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
8603   servers that both support SSLv2 and enable export cipher suites by sending
8604   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
8605
8606   This issue was discovered by Sean Burford (Google) and Emilia Käsper
8607   (OpenSSL development team).
8608   ([CVE-2015-0293])
8609
8610   *Emilia Käsper*
8611
8612 * Use After Free following d2i_ECPrivatekey error fix
8613
8614   A malformed EC private key file consumed via the d2i_ECPrivateKey function
8615   could cause a use after free condition. This, in turn, could cause a double
8616   free in several private key parsing functions (such as d2i_PrivateKey
8617   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
8618   for applications that receive EC private keys from untrusted
8619   sources. This scenario is considered rare.
8620
8621   This issue was discovered by the BoringSSL project and fixed in their
8622   commit 517073cd4b.
8623   ([CVE-2015-0209])
8624
8625   *Matt Caswell*
8626
8627 * X509_to_X509_REQ NULL pointer deref fix
8628
8629   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
8630   the certificate key is invalid. This function is rarely used in practice.
8631
8632   This issue was discovered by Brian Carpenter.
8633   ([CVE-2015-0288])
8634
8635   *Stephen Henson*
8636
8637 * Removed the export ciphers from the DEFAULT ciphers
8638
8639   *Kurt Roeckx*
8640
8641### Changes between 1.0.0p and 1.0.0q [15 Jan 2015]
8642
8643 * Build fixes for the Windows and OpenVMS platforms
8644
8645   *Matt Caswell and Richard Levitte*
8646
8647### Changes between 1.0.0o and 1.0.0p [8 Jan 2015]
8648
8649 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
8650   message can cause a segmentation fault in OpenSSL due to a NULL pointer
8651   dereference. This could lead to a Denial Of Service attack. Thanks to
8652   Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
8653   ([CVE-2014-3571])
8654
8655   *Steve Henson*
8656
8657 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
8658   dtls1_buffer_record function under certain conditions. In particular this
8659   could occur if an attacker sent repeated DTLS records with the same
8660   sequence number but for the next epoch. The memory leak could be exploited
8661   by an attacker in a Denial of Service attack through memory exhaustion.
8662   Thanks to Chris Mueller for reporting this issue.
8663   ([CVE-2015-0206])
8664
8665   *Matt Caswell*
8666
8667 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
8668   built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
8669   method would be set to NULL which could later result in a NULL pointer
8670   dereference. Thanks to Frank Schmirler for reporting this issue.
8671   ([CVE-2014-3569])
8672
8673   *Kurt Roeckx*
8674
8675 * Abort handshake if server key exchange message is omitted for ephemeral
8676   ECDH ciphersuites.
8677
8678   Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
8679   reporting this issue.
8680   ([CVE-2014-3572])
8681
8682   *Steve Henson*
8683
8684 * Remove non-export ephemeral RSA code on client and server. This code
8685   violated the TLS standard by allowing the use of temporary RSA keys in
8686   non-export ciphersuites and could be used by a server to effectively
8687   downgrade the RSA key length used to a value smaller than the server
8688   certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
8689   INRIA or reporting this issue.
8690   ([CVE-2015-0204])
8691
8692   *Steve Henson*
8693
8694 * Fixed issue where DH client certificates are accepted without verification.
8695   An OpenSSL server will accept a DH certificate for client authentication
8696   without the certificate verify message. This effectively allows a client to
8697   authenticate without the use of a private key. This only affects servers
8698   which trust a client certificate authority which issues certificates
8699   containing DH keys: these are extremely rare and hardly ever encountered.
8700   Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
8701   this issue.
8702   ([CVE-2015-0205])
8703
8704   *Steve Henson*
8705
8706 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
8707   results on some platforms, including x86_64. This bug occurs at random
8708   with a very low probability, and is not known to be exploitable in any
8709   way, though its exact impact is difficult to determine. Thanks to Pieter
8710   Wuille (Blockstream) who reported this issue and also suggested an initial
8711   fix. Further analysis was conducted by the OpenSSL development team and
8712   Adam Langley of Google. The final fix was developed by Andy Polyakov of
8713   the OpenSSL core team.
8714   ([CVE-2014-3570])
8715
8716   *Andy Polyakov*
8717
8718 * Fix various certificate fingerprint issues.
8719
8720   By using non-DER or invalid encodings outside the signed portion of a
8721   certificate the fingerprint can be changed without breaking the signature.
8722   Although no details of the signed portion of the certificate can be changed
8723   this can cause problems with some applications: e.g. those using the
8724   certificate fingerprint for blacklists.
8725
8726   1. Reject signatures with non zero unused bits.
8727
8728   If the BIT STRING containing the signature has non zero unused bits reject
8729   the signature. All current signature algorithms require zero unused bits.
8730
8731   2. Check certificate algorithm consistency.
8732
8733   Check the AlgorithmIdentifier inside TBS matches the one in the
8734   certificate signature. NB: this will result in signature failure
8735   errors for some broken certificates.
8736
8737   Thanks to Konrad Kraszewski from Google for reporting this issue.
8738
8739   3. Check DSA/ECDSA signatures use DER.
8740
8741   Reencode DSA/ECDSA signatures and compare with the original received
8742   signature. Return an error if there is a mismatch.
8743
8744   This will reject various cases including garbage after signature
8745   (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
8746   program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
8747   (negative or with leading zeroes).
8748
8749   Further analysis was conducted and fixes were developed by Stephen Henson
8750   of the OpenSSL core team.
8751
8752   ([CVE-2014-8275])
8753
8754   *Steve Henson*
8755
8756### Changes between 1.0.0n and 1.0.0o [15 Oct 2014]
8757
8758 * Session Ticket Memory Leak.
8759
8760   When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
8761   integrity of that ticket is first verified. In the event of a session
8762   ticket integrity check failing, OpenSSL will fail to free memory
8763   causing a memory leak. By sending a large number of invalid session
8764   tickets an attacker could exploit this issue in a Denial Of Service
8765   attack.
8766   ([CVE-2014-3567])
8767
8768   *Steve Henson*
8769
8770 * Build option no-ssl3 is incomplete.
8771
8772   When OpenSSL is configured with "no-ssl3" as a build option, servers
8773   could accept and complete a SSL 3.0 handshake, and clients could be
8774   configured to send them.
8775   ([CVE-2014-3568])
8776
8777   *Akamai and the OpenSSL team*
8778
8779 * Add support for TLS_FALLBACK_SCSV.
8780   Client applications doing fallback retries should call
8781   SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
8782   ([CVE-2014-3566])
8783
8784   *Adam Langley, Bodo Moeller*
8785
8786 * Add additional DigestInfo checks.
8787
8788   Reencode DigestInto in DER and check against the original when
8789   verifying RSA signature: this will reject any improperly encoded
8790   DigestInfo structures.
8791
8792   Note: this is a precautionary measure and no attacks are currently known.
8793
8794   *Steve Henson*
8795
8796### Changes between 1.0.0m and 1.0.0n [6 Aug 2014]
8797
8798 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
8799   to a denial of service attack. A malicious server can crash the client
8800   with a null pointer dereference (read) by specifying an anonymous (EC)DH
8801   ciphersuite and sending carefully crafted handshake messages.
8802
8803   Thanks to Felix Gröbert (Google) for discovering and researching this
8804   issue.
8805   ([CVE-2014-3510])
8806
8807   *Emilia Käsper*
8808
8809 * By sending carefully crafted DTLS packets an attacker could cause openssl
8810   to leak memory. This can be exploited through a Denial of Service attack.
8811   Thanks to Adam Langley for discovering and researching this issue.
8812   ([CVE-2014-3507])
8813
8814   *Adam Langley*
8815
8816 * An attacker can force openssl to consume large amounts of memory whilst
8817   processing DTLS handshake messages. This can be exploited through a
8818   Denial of Service attack.
8819   Thanks to Adam Langley for discovering and researching this issue.
8820   ([CVE-2014-3506])
8821
8822   *Adam Langley*
8823
8824 * An attacker can force an error condition which causes openssl to crash
8825   whilst processing DTLS packets due to memory being freed twice. This
8826   can be exploited through a Denial of Service attack.
8827   Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
8828   this issue.
8829   ([CVE-2014-3505])
8830
8831   *Adam Langley*
8832
8833 * If a multithreaded client connects to a malicious server using a resumed
8834   session and the server sends an ec point format extension it could write
8835   up to 255 bytes to freed memory.
8836
8837   Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
8838   issue.
8839   ([CVE-2014-3509])
8840
8841   *Gabor Tyukasz*
8842
8843 * A flaw in OBJ_obj2txt may cause pretty printing functions such as
8844   X509_name_oneline, X509_name_print_ex et al. to leak some information
8845   from the stack. Applications may be affected if they echo pretty printing
8846   output to the attacker.
8847
8848   Thanks to Ivan Fratric (Google) for discovering this issue.
8849   ([CVE-2014-3508])
8850
8851   *Emilia Käsper, and Steve Henson*
8852
8853 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
8854   for corner cases. (Certain input points at infinity could lead to
8855   bogus results, with non-infinity inputs mapped to infinity too.)
8856
8857   *Bodo Moeller*
8858
8859### Changes between 1.0.0l and 1.0.0m [5 Jun 2014]
8860
8861 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
8862   handshake can force the use of weak keying material in OpenSSL
8863   SSL/TLS clients and servers.
8864
8865   Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
8866   researching this issue. ([CVE-2014-0224])
8867
8868   *KIKUCHI Masashi, Steve Henson*
8869
8870 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
8871   OpenSSL DTLS client the code can be made to recurse eventually crashing
8872   in a DoS attack.
8873
8874   Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
8875   ([CVE-2014-0221])
8876
8877   *Imre Rad, Steve Henson*
8878
8879 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
8880   be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
8881   client or server. This is potentially exploitable to run arbitrary
8882   code on a vulnerable client or server.
8883
8884   Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195])
8885
8886   *Jüri Aedla, Steve Henson*
8887
8888 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
8889   are subject to a denial of service attack.
8890
8891   Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
8892   this issue. ([CVE-2014-3470])
8893
8894   *Felix Gröbert, Ivan Fratric, Steve Henson*
8895
8896 * Harmonize version and its documentation. -f flag is used to display
8897   compilation flags.
8898
8899   *mancha <mancha1@zoho.com>*
8900
8901 * Fix eckey_priv_encode so it immediately returns an error upon a failure
8902   in i2d_ECPrivateKey.
8903
8904   *mancha <mancha1@zoho.com>*
8905
8906 * Fix some double frees. These are not thought to be exploitable.
8907
8908   *mancha <mancha1@zoho.com>*
8909
8910 * Fix for the attack described in the paper "Recovering OpenSSL
8911   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
8912   by Yuval Yarom and Naomi Benger. Details can be obtained from:
8913   <http://eprint.iacr.org/2014/140>
8914
8915   Thanks to Yuval Yarom and Naomi Benger for discovering this
8916   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
8917
8918   *Yuval Yarom and Naomi Benger*
8919
8920### Changes between 1.0.0k and 1.0.0l [6 Jan 2014]
8921
8922 * Keep original DTLS digest and encryption contexts in retransmission
8923   structures so we can use the previous session parameters if they need
8924   to be resent. ([CVE-2013-6450])
8925
8926   *Steve Henson*
8927
8928 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
8929   avoids preferring ECDHE-ECDSA ciphers when the client appears to be
8930   Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
8931   several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
8932   is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
8933   10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
8934
8935   *Rob Stradling, Adam Langley*
8936
8937### Changes between 1.0.0j and 1.0.0k [5 Feb 2013]
8938
8939 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
8940
8941   This addresses the flaw in CBC record processing discovered by
8942   Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
8943   at: <http://www.isg.rhul.ac.uk/tls/>
8944
8945   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
8946   Security Group at Royal Holloway, University of London
8947   (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
8948   Emilia Käsper for the initial patch.
8949   ([CVE-2013-0169])
8950
8951   *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson*
8952
8953 * Return an error when checking OCSP signatures when key is NULL.
8954   This fixes a DoS attack. ([CVE-2013-0166])
8955
8956   *Steve Henson*
8957
8958 * Call OCSP Stapling callback after ciphersuite has been chosen, so
8959   the right response is stapled. Also change SSL_get_certificate()
8960   so it returns the certificate actually sent.
8961   See <http://rt.openssl.org/Ticket/Display.html?id=2836>.
8962   (This is a backport)
8963
8964   *Rob Stradling <rob.stradling@comodo.com>*
8965
8966 * Fix possible deadlock when decoding public keys.
8967
8968   *Steve Henson*
8969
8970### Changes between 1.0.0i and 1.0.0j [10 May 2012]
8971
8972[NB: OpenSSL 1.0.0i and later 1.0.0 patch levels were released after
8973OpenSSL 1.0.1.]
8974
8975 * Sanity check record length before skipping explicit IV in DTLS
8976   to fix DoS attack.
8977
8978   Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
8979   fuzzing as a service testing platform.
8980   ([CVE-2012-2333])
8981
8982   *Steve Henson*
8983
8984 * Initialise tkeylen properly when encrypting CMS messages.
8985   Thanks to Solar Designer of Openwall for reporting this issue.
8986
8987   *Steve Henson*
8988
8989### Changes between 1.0.0h and 1.0.0i [19 Apr 2012]
8990
8991 * Check for potentially exploitable overflows in asn1_d2i_read_bio
8992   BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
8993   in CRYPTO_realloc_clean.
8994
8995   Thanks to Tavis Ormandy, Google Security Team, for discovering this
8996   issue and to Adam Langley <agl@chromium.org> for fixing it.
8997   ([CVE-2012-2110])
8998
8999   *Adam Langley (Google), Tavis Ormandy, Google Security Team*
9000
9001### Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
9002
9003 * Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
9004   in CMS and PKCS7 code. When RSA decryption fails use a random key for
9005   content decryption and always return the same error. Note: this attack
9006   needs on average 2^20 messages so it only affects automated senders. The
9007   old behaviour can be re-enabled in the CMS code by setting the
9008   CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
9009   an MMA defence is not necessary.
9010   Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
9011   this issue. ([CVE-2012-0884])
9012
9013   *Steve Henson*
9014
9015 * Fix CVE-2011-4619: make sure we really are receiving a
9016   client hello before rejecting multiple SGC restarts. Thanks to
9017   Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
9018
9019   *Steve Henson*
9020
9021### Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
9022
9023 * Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
9024   Thanks to Antonio Martin, Enterprise Secure Access Research and
9025   Development, Cisco Systems, Inc. for discovering this bug and
9026   preparing a fix. ([CVE-2012-0050])
9027
9028   *Antonio Martin*
9029
9030### Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
9031
9032 * Nadhem Alfardan and Kenny Paterson have discovered an extension
9033   of the Vaudenay padding oracle attack on CBC mode encryption
9034   which enables an efficient plaintext recovery attack against
9035   the OpenSSL implementation of DTLS. Their attack exploits timing
9036   differences arising during decryption processing. A research
9037   paper describing this attack can be found at:
9038   <http://www.isg.rhul.ac.uk/~kp/dtls.pdf>
9039   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
9040   Security Group at Royal Holloway, University of London
9041   (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
9042   <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
9043   for preparing the fix. ([CVE-2011-4108])
9044
9045   *Robin Seggelmann, Michael Tuexen*
9046
9047 * Clear bytes used for block padding of SSL 3.0 records.
9048   ([CVE-2011-4576])
9049
9050   *Adam Langley (Google)*
9051
9052 * Only allow one SGC handshake restart for SSL/TLS. Thanks to George
9053   Kadianakis <desnacked@gmail.com> for discovering this issue and
9054   Adam Langley for preparing the fix. ([CVE-2011-4619])
9055
9056   *Adam Langley (Google)*
9057
9058 * Check parameters are not NULL in GOST ENGINE. ([CVE-2012-0027])
9059
9060   *Andrey Kulikov <amdeich@gmail.com>*
9061
9062 * Prevent malformed RFC3779 data triggering an assertion failure.
9063   Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
9064   and Rob Austein <sra@hactrn.net> for fixing it. ([CVE-2011-4577])
9065
9066   *Rob Austein <sra@hactrn.net>*
9067
9068 * Improved PRNG seeding for VOS.
9069
9070   *Paul Green <Paul.Green@stratus.com>*
9071
9072 * Fix ssl_ciph.c set-up race.
9073
9074   *Adam Langley (Google)*
9075
9076 * Fix spurious failures in ecdsatest.c.
9077
9078   *Emilia Käsper (Google)*
9079
9080 * Fix the BIO_f_buffer() implementation (which was mixing different
9081   interpretations of the `..._len` fields).
9082
9083   *Adam Langley (Google)*
9084
9085 * Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
9086   BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
9087   threads won't reuse the same blinding coefficients.
9088
9089   This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
9090   lock to call BN_BLINDING_invert_ex, and avoids one use of
9091   BN_BLINDING_update for each BN_BLINDING structure (previously,
9092   the last update always remained unused).
9093
9094   *Emilia Käsper (Google)*
9095
9096 * In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
9097
9098   *Bob Buckholz (Google)*
9099
9100### Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
9101
9102 * Fix bug where CRLs with nextUpdate in the past are sometimes accepted
9103   by initialising X509_STORE_CTX properly. ([CVE-2011-3207])
9104
9105   *Kaspar Brand <ossl@velox.ch>*
9106
9107 * Fix SSL memory handling for (EC)DH ciphersuites, in particular
9108   for multi-threaded use of ECDH. ([CVE-2011-3210])
9109
9110   *Adam Langley (Google)*
9111
9112 * Fix x509_name_ex_d2i memory leak on bad inputs.
9113
9114   *Bodo Moeller*
9115
9116 * Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
9117   signature public key algorithm by using OID xref utilities instead.
9118   Before this you could only use some ECC ciphersuites with SHA1 only.
9119
9120   *Steve Henson*
9121
9122 * Add protection against ECDSA timing attacks as mentioned in the paper
9123   by Billy Bob Brumley and Nicola Tuveri, see:
9124   <http://eprint.iacr.org/2011/232.pdf>
9125
9126   *Billy Bob Brumley and Nicola Tuveri*
9127
9128### Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
9129
9130 * Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
9131
9132   *Neel Mehta, Adam Langley, Bodo Moeller (Google)*
9133
9134 * Fix bug in string printing code: if *any* escaping is enabled we must
9135   escape the escape character (backslash) or the resulting string is
9136   ambiguous.
9137
9138   *Steve Henson*
9139
9140### Changes between 1.0.0b and 1.0.0c  [2 Dec 2010]
9141
9142 * Disable code workaround for ancient and obsolete Netscape browsers
9143   and servers: an attacker can use it in a ciphersuite downgrade attack.
9144   Thanks to Martin Rex for discovering this bug. CVE-2010-4180
9145
9146   *Steve Henson*
9147
9148 * Fixed J-PAKE implementation error, originally discovered by
9149   Sebastien Martini, further info and confirmation from Stefan
9150   Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
9151
9152   *Ben Laurie*
9153
9154### Changes between 1.0.0a and 1.0.0b  [16 Nov 2010]
9155
9156 * Fix extension code to avoid race conditions which can result in a buffer
9157   overrun vulnerability: resumed sessions must not be modified as they can
9158   be shared by multiple threads. CVE-2010-3864
9159
9160   *Steve Henson*
9161
9162 * Fix WIN32 build system to correctly link an ENGINE directory into
9163   a DLL.
9164
9165   *Steve Henson*
9166
9167### Changes between 1.0.0 and 1.0.0a  [01 Jun 2010]
9168
9169 * Check return value of int_rsa_verify in pkey_rsa_verifyrecover
9170   ([CVE-2010-1633])
9171
9172   *Steve Henson, Peter-Michael Hager <hager@dortmund.net>*
9173
9174### Changes between 0.9.8n and 1.0.0  [29 Mar 2010]
9175
9176 * Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
9177   context. The operation can be customised via the ctrl mechanism in
9178   case ENGINEs want to include additional functionality.
9179
9180   *Steve Henson*
9181
9182 * Tolerate yet another broken PKCS#8 key format: private key value negative.
9183
9184   *Steve Henson*
9185
9186 * Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
9187   output hashes compatible with older versions of OpenSSL.
9188
9189   *Willy Weisz <weisz@vcpc.univie.ac.at>*
9190
9191 * Fix compression algorithm handling: if resuming a session use the
9192   compression algorithm of the resumed session instead of determining
9193   it from client hello again. Don't allow server to change algorithm.
9194
9195   *Steve Henson*
9196
9197 * Add load_crls() function to commands tidying load_certs() too. Add option
9198   to verify utility to allow additional CRLs to be included.
9199
9200   *Steve Henson*
9201
9202 * Update OCSP request code to permit adding custom headers to the request:
9203   some responders need this.
9204
9205   *Steve Henson*
9206
9207 * The function EVP_PKEY_sign() returns <=0 on error: check return code
9208   correctly.
9209
9210   *Julia Lawall <julia@diku.dk>*
9211
9212 * Update verify callback code in `apps/s_cb.c` and `apps/verify.c`, it
9213   needlessly dereferenced structures, used obsolete functions and
9214   didn't handle all updated verify codes correctly.
9215
9216   *Steve Henson*
9217
9218 * Disable MD2 in the default configuration.
9219
9220   *Steve Henson*
9221
9222 * In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
9223   indicate the initial BIO being pushed or popped. This makes it possible
9224   to determine whether the BIO is the one explicitly called or as a result
9225   of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
9226   it handles reference counts correctly and doesn't zero out the I/O bio
9227   when it is not being explicitly popped. WARNING: applications which
9228   included workarounds for the old buggy behaviour will need to be modified
9229   or they could free up already freed BIOs.
9230
9231   *Steve Henson*
9232
9233 * Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
9234   renaming to all platforms (within the 0.9.8 branch, this was
9235   done conditionally on Netware platforms to avoid a name clash).
9236
9237   *Guenter <lists@gknw.net>*
9238
9239 * Add ECDHE and PSK support to DTLS.
9240
9241   *Michael Tuexen <tuexen@fh-muenster.de>*
9242
9243 * Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
9244   be used on C++.
9245
9246   *Steve Henson*
9247
9248 * Add "missing" function EVP_MD_flags() (without this the only way to
9249   retrieve a digest flags is by accessing the structure directly. Update
9250   `EVP_MD_do_all*()` and `EVP_CIPHER_do_all*()` to include the name a digest
9251   or cipher is registered as in the "from" argument. Print out all
9252   registered digests in the dgst usage message instead of manually
9253   attempting to work them out.
9254
9255   *Steve Henson*
9256
9257 * If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
9258   this allows the use of compression and extensions. Change default cipher
9259   string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
9260   by default unless an application cipher string requests it.
9261
9262   *Steve Henson*
9263
9264 * Alter match criteria in PKCS12_parse(). It used to try to use local
9265   key ids to find matching certificates and keys but some PKCS#12 files
9266   don't follow the (somewhat unwritten) rules and this strategy fails.
9267   Now just gather all certificates together and the first private key
9268   then look for the first certificate that matches the key.
9269
9270   *Steve Henson*
9271
9272 * Support use of registered digest and cipher names for dgst and cipher
9273   commands instead of having to add each one as a special case. So now
9274   you can do:
9275
9276           openssl sha256 foo
9277
9278   as well as:
9279
9280           openssl dgst -sha256 foo
9281
9282   and this works for ENGINE based algorithms too.
9283
9284   *Steve Henson*
9285
9286 * Update Gost ENGINE to support parameter files.
9287
9288   *Victor B. Wagner <vitus@cryptocom.ru>*
9289
9290 * Support GeneralizedTime in ca utility.
9291
9292   *Oliver Martin <oliver@volatilevoid.net>, Steve Henson*
9293
9294 * Enhance the hash format used for certificate directory links. The new
9295   form uses the canonical encoding (meaning equivalent names will work
9296   even if they aren't identical) and uses SHA1 instead of MD5. This form
9297   is incompatible with the older format and as a result c_rehash should
9298   be used to rebuild symbolic links.
9299
9300   *Steve Henson*
9301
9302 * Make PKCS#8 the default write format for private keys, replacing the
9303   traditional format. This form is standardised, more secure and doesn't
9304   include an implicit MD5 dependency.
9305
9306   *Steve Henson*
9307
9308 * Add a $gcc_devteam_warn option to Configure. The idea is that any code
9309   committed to OpenSSL should pass this lot as a minimum.
9310
9311   *Steve Henson*
9312
9313 * Add session ticket override functionality for use by EAP-FAST.
9314
9315   *Jouni Malinen <j@w1.fi>*
9316
9317 * Modify HMAC functions to return a value. Since these can be implemented
9318   in an ENGINE errors can occur.
9319
9320   *Steve Henson*
9321
9322 * Type-checked OBJ_bsearch_ex.
9323
9324   *Ben Laurie*
9325
9326 * Type-checked OBJ_bsearch. Also some constification necessitated
9327   by type-checking.  Still to come: TXT_DB, bsearch(?),
9328   OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
9329   CONF_VALUE.
9330
9331   *Ben Laurie*
9332
9333 * New function OPENSSL_gmtime_adj() to add a specific number of days and
9334   seconds to a tm structure directly, instead of going through OS
9335   specific date routines. This avoids any issues with OS routines such
9336   as the year 2038 bug. New `*_adj()` functions for ASN1 time structures
9337   and X509_time_adj_ex() to cover the extended range. The existing
9338   X509_time_adj() is still usable and will no longer have any date issues.
9339
9340   *Steve Henson*
9341
9342 * Delta CRL support. New use deltas option which will attempt to locate
9343   and search any appropriate delta CRLs available.
9344
9345   This work was sponsored by Google.
9346
9347   *Steve Henson*
9348
9349 * Support for CRLs partitioned by reason code. Reorganise CRL processing
9350   code and add additional score elements. Validate alternate CRL paths
9351   as part of the CRL checking and indicate a new error "CRL path validation
9352   error" in this case. Applications wanting additional details can use
9353   the verify callback and check the new "parent" field. If this is not
9354   NULL CRL path validation is taking place. Existing applications won't
9355   see this because it requires extended CRL support which is off by
9356   default.
9357
9358   This work was sponsored by Google.
9359
9360   *Steve Henson*
9361
9362 * Support for freshest CRL extension.
9363
9364   This work was sponsored by Google.
9365
9366   *Steve Henson*
9367
9368 * Initial indirect CRL support. Currently only supported in the CRLs
9369   passed directly and not via lookup. Process certificate issuer
9370   CRL entry extension and lookup CRL entries by bother issuer name
9371   and serial number. Check and process CRL issuer entry in IDP extension.
9372
9373   This work was sponsored by Google.
9374
9375   *Steve Henson*
9376
9377 * Add support for distinct certificate and CRL paths. The CRL issuer
9378   certificate is validated separately in this case. Only enabled if
9379   an extended CRL support flag is set: this flag will enable additional
9380   CRL functionality in future.
9381
9382   This work was sponsored by Google.
9383
9384   *Steve Henson*
9385
9386 * Add support for policy mappings extension.
9387
9388   This work was sponsored by Google.
9389
9390   *Steve Henson*
9391
9392 * Fixes to pathlength constraint, self issued certificate handling,
9393   policy processing to align with RFC3280 and PKITS tests.
9394
9395   This work was sponsored by Google.
9396
9397   *Steve Henson*
9398
9399 * Support for name constraints certificate extension. DN, email, DNS
9400   and URI types are currently supported.
9401
9402   This work was sponsored by Google.
9403
9404   *Steve Henson*
9405
9406 * To cater for systems that provide a pointer-based thread ID rather
9407   than numeric, deprecate the current numeric thread ID mechanism and
9408   replace it with a structure and associated callback type. This
9409   mechanism allows a numeric "hash" to be extracted from a thread ID in
9410   either case, and on platforms where pointers are larger than 'long',
9411   mixing is done to help ensure the numeric 'hash' is usable even if it
9412   can't be guaranteed unique. The default mechanism is to use "&errno"
9413   as a pointer-based thread ID to distinguish between threads.
9414
9415   Applications that want to provide their own thread IDs should now use
9416   CRYPTO_THREADID_set_callback() to register a callback that will call
9417   either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
9418
9419   Note that ERR_remove_state() is now deprecated, because it is tied
9420   to the assumption that thread IDs are numeric.  ERR_remove_state(0)
9421   to free the current thread's error state should be replaced by
9422   ERR_remove_thread_state(NULL).
9423
9424   (This new approach replaces the functions CRYPTO_set_idptr_callback(),
9425   CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
9426   OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
9427   application was previously providing a numeric thread callback that
9428   was inappropriate for distinguishing threads, then uniqueness might
9429   have been obtained with &errno that happened immediately in the
9430   intermediate development versions of OpenSSL; this is no longer the
9431   case, the numeric thread callback will now override the automatic use
9432   of &errno.)
9433
9434   *Geoff Thorpe, with help from Bodo Moeller*
9435
9436 * Initial support for different CRL issuing certificates. This covers a
9437   simple case where the self issued certificates in the chain exist and
9438   the real CRL issuer is higher in the existing chain.
9439
9440   This work was sponsored by Google.
9441
9442   *Steve Henson*
9443
9444 * Removed effectively defunct crypto/store from the build.
9445
9446   *Ben Laurie*
9447
9448 * Revamp of STACK to provide stronger type-checking. Still to come:
9449   TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
9450   ASN1_STRING, CONF_VALUE.
9451
9452   *Ben Laurie*
9453
9454 * Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
9455   RAM on SSL connections.  This option can save about 34k per idle SSL.
9456
9457   *Nick Mathewson*
9458
9459 * Revamp of LHASH to provide stronger type-checking. Still to come:
9460   STACK, TXT_DB, bsearch, qsort.
9461
9462   *Ben Laurie*
9463
9464 * Initial support for Cryptographic Message Syntax (aka CMS) based
9465   on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
9466   support for data, signedData, compressedData, digestedData and
9467   encryptedData, envelopedData types included. Scripts to check against
9468   RFC4134 examples draft and interop and consistency checks of many
9469   content types and variants.
9470
9471   *Steve Henson*
9472
9473 * Add options to enc utility to support use of zlib compression BIO.
9474
9475   *Steve Henson*
9476
9477 * Extend mk1mf to support importing of options and assembly language
9478   files from Configure script, currently only included in VC-WIN32.
9479   The assembly language rules can now optionally generate the source
9480   files from the associated perl scripts.
9481
9482   *Steve Henson*
9483
9484 * Implement remaining functionality needed to support GOST ciphersuites.
9485   Interop testing has been performed using CryptoPro implementations.
9486
9487   *Victor B. Wagner <vitus@cryptocom.ru>*
9488
9489 * s390x assembler pack.
9490
9491   *Andy Polyakov*
9492
9493 * ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
9494   "family."
9495
9496   *Andy Polyakov*
9497
9498 * Implement Opaque PRF Input TLS extension as specified in
9499   draft-rescorla-tls-opaque-prf-input-00.txt.  Since this is not an
9500   official specification yet and no extension type assignment by
9501   IANA exists, this extension (for now) will have to be explicitly
9502   enabled when building OpenSSL by providing the extension number
9503   to use.  For example, specify an option
9504
9505           -DTLSEXT_TYPE_opaque_prf_input=0x9527
9506
9507   to the "config" or "Configure" script to enable the extension,
9508   assuming extension number 0x9527 (which is a completely arbitrary
9509   and unofficial assignment based on the MD5 hash of the Internet
9510   Draft).  Note that by doing so, you potentially lose
9511   interoperability with other TLS implementations since these might
9512   be using the same extension number for other purposes.
9513
9514   SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
9515   opaque PRF input value to use in the handshake.  This will create
9516   an internal copy of the length-'len' string at 'src', and will
9517   return non-zero for success.
9518
9519   To get more control and flexibility, provide a callback function
9520   by using
9521
9522           SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
9523           SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
9524
9525   where
9526
9527           int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
9528           void *arg;
9529
9530   Callback function 'cb' will be called in handshakes, and is
9531   expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
9532   Argument 'arg' is for application purposes (the value as given to
9533   SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
9534   be provided to the callback function).  The callback function
9535   has to return non-zero to report success: usually 1 to use opaque
9536   PRF input just if possible, or 2 to enforce use of the opaque PRF
9537   input.  In the latter case, the library will abort the handshake
9538   if opaque PRF input is not successfully negotiated.
9539
9540   Arguments 'peerinput' and 'len' given to the callback function
9541   will always be NULL and 0 in the case of a client.  A server will
9542   see the client's opaque PRF input through these variables if
9543   available (NULL and 0 otherwise).  Note that if the server
9544   provides an opaque PRF input, the length must be the same as the
9545   length of the client's opaque PRF input.
9546
9547   Note that the callback function will only be called when creating
9548   a new session (session resumption can resume whatever was
9549   previously negotiated), and will not be called in SSL 2.0
9550   handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
9551   SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
9552   for applications that need to enforce opaque PRF input.
9553
9554   *Bodo Moeller*
9555
9556 * Update ssl code to support digests other than SHA1+MD5 for handshake
9557   MAC.
9558
9559   *Victor B. Wagner <vitus@cryptocom.ru>*
9560
9561 * Add RFC4507 support to OpenSSL. This includes the corrections in
9562   RFC4507bis. The encrypted ticket format is an encrypted encoded
9563   SSL_SESSION structure, that way new session features are automatically
9564   supported.
9565
9566   If a client application caches session in an SSL_SESSION structure
9567   support is transparent because tickets are now stored in the encoded
9568   SSL_SESSION.
9569
9570   The SSL_CTX structure automatically generates keys for ticket
9571   protection in servers so again support should be possible
9572   with no application modification.
9573
9574   If a client or server wishes to disable RFC4507 support then the option
9575   SSL_OP_NO_TICKET can be set.
9576
9577   Add a TLS extension debugging callback to allow the contents of any client
9578   or server extensions to be examined.
9579
9580   This work was sponsored by Google.
9581
9582   *Steve Henson*
9583
9584 * Final changes to avoid use of pointer pointer casts in OpenSSL.
9585   OpenSSL should now compile cleanly on gcc 4.2
9586
9587   *Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson*
9588
9589 * Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
9590   support including streaming MAC support: this is required for GOST
9591   ciphersuite support.
9592
9593   *Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson*
9594
9595 * Add option -stream to use PKCS#7 streaming in smime utility. New
9596   function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
9597   to output in BER and PEM format.
9598
9599   *Steve Henson*
9600
9601 * Experimental support for use of HMAC via EVP_PKEY interface. This
9602   allows HMAC to be handled via the `EVP_DigestSign*()` interface. The
9603   EVP_PKEY "key" in this case is the HMAC key, potentially allowing
9604   ENGINE support for HMAC keys which are unextractable. New -mac and
9605   -macopt options to dgst utility.
9606
9607   *Steve Henson*
9608
9609 * New option -sigopt to dgst utility. Update dgst to use
9610   `EVP_Digest{Sign,Verify}*`. These two changes make it possible to use
9611   alternative signing parameters such as X9.31 or PSS in the dgst
9612   utility.
9613
9614   *Steve Henson*
9615
9616 * Change ssl_cipher_apply_rule(), the internal function that does
9617   the work each time a ciphersuite string requests enabling
9618   ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
9619   removing ("!foo+bar") a class of ciphersuites: Now it maintains
9620   the order of disabled ciphersuites such that those ciphersuites
9621   that most recently went from enabled to disabled not only stay
9622   in order with respect to each other, but also have higher priority
9623   than other disabled ciphersuites the next time ciphersuites are
9624   enabled again.
9625
9626   This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
9627   the same ciphersuites as with "HIGH" alone, but in a specific
9628   order where the PSK ciphersuites come first (since they are the
9629   most recently disabled ciphersuites when "HIGH" is parsed).
9630
9631   Also, change ssl_create_cipher_list() (using this new
9632   functionality) such that between otherwise identical
9633   ciphersuites, ephemeral ECDH is preferred over ephemeral DH in
9634   the default order.
9635
9636   *Bodo Moeller*
9637
9638 * Change ssl_create_cipher_list() so that it automatically
9639   arranges the ciphersuites in reasonable order before starting
9640   to process the rule string.  Thus, the definition for "DEFAULT"
9641   (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
9642   remains equivalent to `"AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH"`.
9643   This makes it much easier to arrive at a reasonable default order
9644   in applications for which anonymous ciphers are OK (meaning
9645   that you can't actually use DEFAULT).
9646
9647   *Bodo Moeller; suggested by Victor Duchovni*
9648
9649 * Split the SSL/TLS algorithm mask (as used for ciphersuite string
9650   processing) into multiple integers instead of setting
9651   "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
9652   "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
9653   (These masks as well as the individual bit definitions are hidden
9654   away into the non-exported interface ssl/ssl_locl.h, so this
9655   change to the definition of the SSL_CIPHER structure shouldn't
9656   affect applications.)  This give us more bits for each of these
9657   categories, so there is no longer a need to coagulate AES128 and
9658   AES256 into a single algorithm bit, and to coagulate Camellia128
9659   and Camellia256 into a single algorithm bit, which has led to all
9660   kinds of kludges.
9661
9662   Thus, among other things, the kludge introduced in 0.9.7m and
9663   0.9.8e for masking out AES256 independently of AES128 or masking
9664   out Camellia256 independently of AES256 is not needed here in 0.9.9.
9665
9666   With the change, we also introduce new ciphersuite aliases that
9667   so far were missing: "AES128", "AES256", "CAMELLIA128", and
9668   "CAMELLIA256".
9669
9670   *Bodo Moeller*
9671
9672 * Add support for dsa-with-SHA224 and dsa-with-SHA256.
9673   Use the leftmost N bytes of the signature input if the input is
9674   larger than the prime q (with N being the size in bytes of q).
9675
9676   *Nils Larsch*
9677
9678 * Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
9679   it yet and it is largely untested.
9680
9681   *Steve Henson*
9682
9683 * Add support for the ecdsa-with-SHA224/256/384/512 signature types.
9684
9685   *Nils Larsch*
9686
9687 * Initial incomplete changes to avoid need for function casts in OpenSSL
9688   some compilers (gcc 4.2 and later) reject their use. Safestack is
9689   reimplemented.  Update ASN1 to avoid use of legacy functions.
9690
9691   *Steve Henson*
9692
9693 * Win32/64 targets are linked with Winsock2.
9694
9695   *Andy Polyakov*
9696
9697 * Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
9698   to external functions. This can be used to increase CRL handling
9699   efficiency especially when CRLs are very large by (for example) storing
9700   the CRL revoked certificates in a database.
9701
9702   *Steve Henson*
9703
9704 * Overhaul of by_dir code. Add support for dynamic loading of CRLs so
9705   new CRLs added to a directory can be used. New command line option
9706   -verify_return_error to s_client and s_server. This causes real errors
9707   to be returned by the verify callback instead of carrying on no matter
9708   what. This reflects the way a "real world" verify callback would behave.
9709
9710   *Steve Henson*
9711
9712 * GOST engine, supporting several GOST algorithms and public key formats.
9713   Kindly donated by Cryptocom.
9714
9715   *Cryptocom*
9716
9717 * Partial support for Issuing Distribution Point CRL extension. CRLs
9718   partitioned by DP are handled but no indirect CRL or reason partitioning
9719   (yet). Complete overhaul of CRL handling: now the most suitable CRL is
9720   selected via a scoring technique which handles IDP and AKID in CRLs.
9721
9722   *Steve Henson*
9723
9724 * New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
9725   will ultimately be used for all verify operations: this will remove the
9726   X509_STORE dependency on certificate verification and allow alternative
9727   lookup methods.  X509_STORE based implementations of these two callbacks.
9728
9729   *Steve Henson*
9730
9731 * Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
9732   Modify get_crl() to find a valid (unexpired) CRL if possible.
9733
9734   *Steve Henson*
9735
9736 * New function X509_CRL_match() to check if two CRLs are identical. Normally
9737   this would be called X509_CRL_cmp() but that name is already used by
9738   a function that just compares CRL issuer names. Cache several CRL
9739   extensions in X509_CRL structure and cache CRLDP in X509.
9740
9741   *Steve Henson*
9742
9743 * Store a "canonical" representation of X509_NAME structure (ASN1 Name)
9744   this maps equivalent X509_NAME structures into a consistent structure.
9745   Name comparison can then be performed rapidly using memcmp().
9746
9747   *Steve Henson*
9748
9749 * Non-blocking OCSP request processing. Add -timeout option to ocsp
9750   utility.
9751
9752   *Steve Henson*
9753
9754 * Allow digests to supply their own micalg string for S/MIME type using
9755   the ctrl EVP_MD_CTRL_MICALG.
9756
9757   *Steve Henson*
9758
9759 * During PKCS7 signing pass the PKCS7 SignerInfo structure to the
9760   EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
9761   ctrl. It can then customise the structure before and/or after signing
9762   if necessary.
9763
9764   *Steve Henson*
9765
9766 * New function OBJ_add_sigid() to allow application defined signature OIDs
9767   to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
9768   to free up any added signature OIDs.
9769
9770   *Steve Henson*
9771
9772 * New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
9773   EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
9774   digest and cipher tables. New options added to openssl utility:
9775   list-message-digest-algorithms and list-cipher-algorithms.
9776
9777   *Steve Henson*
9778
9779 * Change the array representation of binary polynomials: the list
9780   of degrees of non-zero coefficients is now terminated with -1.
9781   Previously it was terminated with 0, which was also part of the
9782   value; thus, the array representation was not applicable to
9783   polynomials where t^0 has coefficient zero.  This change makes
9784   the array representation useful in a more general context.
9785
9786   *Douglas Stebila*
9787
9788 * Various modifications and fixes to SSL/TLS cipher string
9789   handling.  For ECC, the code now distinguishes between fixed ECDH
9790   with RSA certificates on the one hand and with ECDSA certificates
9791   on the other hand, since these are separate ciphersuites.  The
9792   unused code for Fortezza ciphersuites has been removed.
9793
9794   For consistency with EDH, ephemeral ECDH is now called "EECDH"
9795   (not "ECDHE").  For consistency with the code for DH
9796   certificates, use of ECDH certificates is now considered ECDH
9797   authentication, not RSA or ECDSA authentication (the latter is
9798   merely the CA's signing algorithm and not actively used in the
9799   protocol).
9800
9801   The temporary ciphersuite alias "ECCdraft" is no longer
9802   available, and ECC ciphersuites are no longer excluded from "ALL"
9803   and "DEFAULT".  The following aliases now exist for RFC 4492
9804   ciphersuites, most of these by analogy with the DH case:
9805
9806           kECDHr   - ECDH cert, signed with RSA
9807           kECDHe   - ECDH cert, signed with ECDSA
9808           kECDH    - ECDH cert (signed with either RSA or ECDSA)
9809           kEECDH   - ephemeral ECDH
9810           ECDH     - ECDH cert or ephemeral ECDH
9811
9812           aECDH    - ECDH cert
9813           aECDSA   - ECDSA cert
9814           ECDSA    - ECDSA cert
9815
9816           AECDH    - anonymous ECDH
9817           EECDH    - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
9818
9819   *Bodo Moeller*
9820
9821 * Add additional S/MIME capabilities for AES and GOST ciphers if supported.
9822   Use correct micalg parameters depending on digest(s) in signed message.
9823
9824   *Steve Henson*
9825
9826 * Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
9827   an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
9828
9829   *Steve Henson*
9830
9831 * Initial engine support for EVP_PKEY_METHOD. New functions to permit
9832   an engine to register a method. Add ENGINE lookups for methods and
9833   functional reference processing.
9834
9835   *Steve Henson*
9836
9837 * New functions `EVP_Digest{Sign,Verify)*`. These are enhanced versions of
9838   `EVP_{Sign,Verify}*` which allow an application to customise the signature
9839   process.
9840
9841   *Steve Henson*
9842
9843 * New -resign option to smime utility. This adds one or more signers
9844   to an existing PKCS#7 signedData structure. Also -md option to use an
9845   alternative message digest algorithm for signing.
9846
9847   *Steve Henson*
9848
9849 * Tidy up PKCS#7 routines and add new functions to make it easier to
9850   create PKCS7 structures containing multiple signers. Update smime
9851   application to support multiple signers.
9852
9853   *Steve Henson*
9854
9855 * New -macalg option to pkcs12 utility to allow setting of an alternative
9856   digest MAC.
9857
9858   *Steve Henson*
9859
9860 * Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
9861   Reorganize PBE internals to lookup from a static table using NIDs,
9862   add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
9863   EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
9864   PRF which will be automatically used with PBES2.
9865
9866   *Steve Henson*
9867
9868 * Replace the algorithm specific calls to generate keys in "req" with the
9869   new API.
9870
9871   *Steve Henson*
9872
9873 * Update PKCS#7 enveloped data routines to use new API. This is now
9874   supported by any public key method supporting the encrypt operation. A
9875   ctrl is added to allow the public key algorithm to examine or modify
9876   the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
9877   a no op.
9878
9879   *Steve Henson*
9880
9881 * Add a ctrl to asn1 method to allow a public key algorithm to express
9882   a default digest type to use. In most cases this will be SHA1 but some
9883   algorithms (such as GOST) need to specify an alternative digest. The
9884   return value indicates how strong the preference is 1 means optional and
9885   2 is mandatory (that is it is the only supported type). Modify
9886   ASN1_item_sign() to accept a NULL digest argument to indicate it should
9887   use the default md. Update openssl utilities to use the default digest
9888   type for signing if it is not explicitly indicated.
9889
9890   *Steve Henson*
9891
9892 * Use OID cross reference table in ASN1_sign() and ASN1_verify(). New
9893   EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
9894   signing method from the key type. This effectively removes the link
9895   between digests and public key types.
9896
9897   *Steve Henson*
9898
9899 * Add an OID cross reference table and utility functions. Its purpose is to
9900   translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
9901   rsaEncryption. This will allow some of the algorithm specific hackery
9902   needed to use the correct OID to be removed.
9903
9904   *Steve Henson*
9905
9906 * Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
9907   structures for PKCS7_sign(). They are now set up by the relevant public
9908   key ASN1 method.
9909
9910   *Steve Henson*
9911
9912 * Add provisional EC pkey method with support for ECDSA and ECDH.
9913
9914   *Steve Henson*
9915
9916 * Add support for key derivation (agreement) in the API, DH method and
9917   pkeyutl.
9918
9919   *Steve Henson*
9920
9921 * Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
9922   public and private key formats. As a side effect these add additional
9923   command line functionality not previously available: DSA signatures can be
9924   generated and verified using pkeyutl and DH key support and generation in
9925   pkey, genpkey.
9926
9927   *Steve Henson*
9928
9929 * BeOS support.
9930
9931   *Oliver Tappe <zooey@hirschkaefer.de>*
9932
9933 * New make target "install_html_docs" installs HTML renditions of the
9934   manual pages.
9935
9936   *Oliver Tappe <zooey@hirschkaefer.de>*
9937
9938 * New utility "genpkey" this is analogous to "genrsa" etc except it can
9939   generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
9940   support key and parameter generation and add initial key generation
9941   functionality for RSA.
9942
9943   *Steve Henson*
9944
9945 * Add functions for main EVP_PKEY_method operations. The undocumented
9946   functions `EVP_PKEY_{encrypt,decrypt}` have been renamed to
9947   `EVP_PKEY_{encrypt,decrypt}_old`.
9948
9949   *Steve Henson*
9950
9951 * Initial definitions for EVP_PKEY_METHOD. This will be a high level public
9952   key API, doesn't do much yet.
9953
9954   *Steve Henson*
9955
9956 * New function EVP_PKEY_asn1_get0_info() to retrieve information about
9957   public key algorithms. New option to openssl utility:
9958   "list-public-key-algorithms" to print out info.
9959
9960   *Steve Henson*
9961
9962 * Implement the Supported Elliptic Curves Extension for
9963   ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
9964
9965   *Douglas Stebila*
9966
9967 * Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
9968   EVP_CIPHER structures to avoid later problems in EVP_cleanup().
9969
9970   *Steve Henson*
9971
9972 * New utilities pkey and pkeyparam. These are similar to algorithm specific
9973   utilities such as rsa, dsa, dsaparam etc except they process any key
9974   type.
9975
9976   *Steve Henson*
9977
9978 * Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New
9979   functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
9980   EVP_PKEY_print_param() to print public key data from an EVP_PKEY
9981   structure.
9982
9983   *Steve Henson*
9984
9985 * Initial support for pluggable public key ASN1.
9986   De-spaghettify the public key ASN1 handling. Move public and private
9987   key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
9988   algorithm specific handling to a single module within the relevant
9989   algorithm directory. Add functions to allow (near) opaque processing
9990   of public and private key structures.
9991
9992   *Steve Henson*
9993
9994 * Implement the Supported Point Formats Extension for
9995   ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
9996
9997   *Douglas Stebila*
9998
9999 * Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
10000   for the psk identity [hint] and the psk callback functions to the
10001   SSL_SESSION, SSL and SSL_CTX structure.
10002
10003   New ciphersuites:
10004           PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
10005           PSK-AES256-CBC-SHA
10006
10007   New functions:
10008           SSL_CTX_use_psk_identity_hint
10009           SSL_get_psk_identity_hint
10010           SSL_get_psk_identity
10011           SSL_use_psk_identity_hint
10012
10013   *Mika Kousa and Pasi Eronen of Nokia Corporation*
10014
10015 * Add RFC 3161 compliant time stamp request creation, response generation
10016   and response verification functionality.
10017
10018   *Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project*
10019
10020 * Add initial support for TLS extensions, specifically for the server_name
10021   extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
10022   have new members for a host name.  The SSL data structure has an
10023   additional member `SSL_CTX *initial_ctx` so that new sessions can be
10024   stored in that context to allow for session resumption, even after the
10025   SSL has been switched to a new SSL_CTX in reaction to a client's
10026   server_name extension.
10027
10028   New functions (subject to change):
10029
10030           SSL_get_servername()
10031           SSL_get_servername_type()
10032           SSL_set_SSL_CTX()
10033
10034   New CTRL codes and macros (subject to change):
10035
10036           SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
10037                               - SSL_CTX_set_tlsext_servername_callback()
10038           SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
10039                                    - SSL_CTX_set_tlsext_servername_arg()
10040           SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
10041
10042   openssl s_client has a new '-servername ...' option.
10043
10044   openssl s_server has new options '-servername_host ...', '-cert2 ...',
10045   '-key2 ...', '-servername_fatal' (subject to change).  This allows
10046   testing the HostName extension for a specific single host name ('-cert'
10047   and '-key' remain fallbacks for handshakes without HostName
10048   negotiation).  If the unrecognized_name alert has to be sent, this by
10049   default is a warning; it becomes fatal with the '-servername_fatal'
10050   option.
10051
10052   *Peter Sylvester,  Remy Allais, Christophe Renou*
10053
10054 * Whirlpool hash implementation is added.
10055
10056   *Andy Polyakov*
10057
10058 * BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
10059   bn(64,32). Because of instruction set limitations it doesn't have
10060   any negative impact on performance. This was done mostly in order
10061   to make it possible to share assembler modules, such as bn_mul_mont
10062   implementations, between 32- and 64-bit builds without hassle.
10063
10064   *Andy Polyakov*
10065
10066 * Move code previously exiled into file crypto/ec/ec2_smpt.c
10067   to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
10068   macro.
10069
10070   *Bodo Moeller*
10071
10072 * New candidate for BIGNUM assembler implementation, bn_mul_mont,
10073   dedicated Montgomery multiplication procedure, is introduced.
10074   BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
10075   "64-bit" performance on certain 32-bit targets.
10076
10077   *Andy Polyakov*
10078
10079 * New option SSL_OP_NO_COMP to disable use of compression selectively
10080   in SSL structures. New SSL ctrl to set maximum send fragment size.
10081   Save memory by setting the I/O buffer sizes dynamically instead of
10082   using the maximum available value.
10083
10084   *Steve Henson*
10085
10086 * New option -V for 'openssl ciphers'. This prints the ciphersuite code
10087   in addition to the text details.
10088
10089   *Bodo Moeller*
10090
10091 * Very, very preliminary EXPERIMENTAL support for printing of general
10092   ASN1 structures. This currently produces rather ugly output and doesn't
10093   handle several customised structures at all.
10094
10095   *Steve Henson*
10096
10097 * Integrated support for PVK file format and some related formats such
10098   as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
10099   these in the 'rsa' and 'dsa' utilities.
10100
10101   *Steve Henson*
10102
10103 * Support for PKCS#1 RSAPublicKey format on rsa utility command line.
10104
10105   *Steve Henson*
10106
10107 * Remove the ancient ASN1_METHOD code. This was only ever used in one
10108   place for the (very old) "NETSCAPE" format certificates which are now
10109   handled using new ASN1 code equivalents.
10110
10111   *Steve Henson*
10112
10113 * Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
10114   pointer and make the SSL_METHOD parameter in SSL_CTX_new,
10115   SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
10116
10117   *Nils Larsch*
10118
10119 * Modify CRL distribution points extension code to print out previously
10120   unsupported fields. Enhance extension setting code to allow setting of
10121   all fields.
10122
10123   *Steve Henson*
10124
10125 * Add print and set support for Issuing Distribution Point CRL extension.
10126
10127   *Steve Henson*
10128
10129 * Change 'Configure' script to enable Camellia by default.
10130
10131   *NTT*
10132
10133OpenSSL 0.9.x
10134-------------
10135
10136### Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
10137
10138 * When rejecting SSL/TLS records due to an incorrect version number, never
10139   update s->server with a new major version number.  As of
10140   - OpenSSL 0.9.8m if 'short' is a 16-bit type,
10141   - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
10142   the previous behavior could result in a read attempt at NULL when
10143   receiving specific incorrect SSL/TLS records once record payload
10144   protection is active.  ([CVE-2010-0740])
10145
10146   *Bodo Moeller, Adam Langley <agl@chromium.org>*
10147
10148 * Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL
10149   could be crashed if the relevant tables were not present (e.g. chrooted).
10150
10151   *Tomas Hoger <thoger@redhat.com>*
10152
10153### Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
10154
10155 * Always check bn_wexpand() return values for failure.  ([CVE-2009-3245])
10156
10157   *Martin Olsson, Neel Mehta*
10158
10159 * Fix X509_STORE locking: Every 'objs' access requires a lock (to
10160   accommodate for stack sorting, always a write lock!).
10161
10162   *Bodo Moeller*
10163
10164 * On some versions of WIN32 Heap32Next is very slow. This can cause
10165   excessive delays in the RAND_poll(): over a minute. As a workaround
10166   include a time check in the inner Heap32Next loop too.
10167
10168   *Steve Henson*
10169
10170 * The code that handled flushing of data in SSL/TLS originally used the
10171   BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
10172   the problem outlined in PR#1949. The fix suggested there however can
10173   trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
10174   of Apache). So instead simplify the code to flush unconditionally.
10175   This should be fine since flushing with no data to flush is a no op.
10176
10177   *Steve Henson*
10178
10179 * Handle TLS versions 2.0 and later properly and correctly use the
10180   highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
10181   off ancient servers have a habit of sticking around for a while...
10182
10183   *Steve Henson*
10184
10185 * Modify compression code so it frees up structures without using the
10186   ex_data callbacks. This works around a problem where some applications
10187   call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
10188   restarting) then use compression (e.g. SSL with compression) later.
10189   This results in significant per-connection memory leaks and
10190   has caused some security issues including CVE-2008-1678 and
10191   CVE-2009-4355.
10192
10193   *Steve Henson*
10194
10195 * Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
10196   change when encrypting or decrypting.
10197
10198   *Bodo Moeller*
10199
10200 * Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
10201   connect and renegotiate with servers which do not support RI.
10202   Until RI is more widely deployed this option is enabled by default.
10203
10204   *Steve Henson*
10205
10206 * Add "missing" ssl ctrls to clear options and mode.
10207
10208   *Steve Henson*
10209
10210 * If client attempts to renegotiate and doesn't support RI respond with
10211   a no_renegotiation alert as required by RFC5746.  Some renegotiating
10212   TLS clients will continue a connection gracefully when they receive
10213   the alert. Unfortunately OpenSSL mishandled this alert and would hang
10214   waiting for a server hello which it will never receive. Now we treat a
10215   received no_renegotiation alert as a fatal error. This is because
10216   applications requesting a renegotiation might well expect it to succeed
10217   and would have no code in place to handle the server denying it so the
10218   only safe thing to do is to terminate the connection.
10219
10220   *Steve Henson*
10221
10222 * Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
10223   peer supports secure renegotiation and 0 otherwise. Print out peer
10224   renegotiation support in s_client/s_server.
10225
10226   *Steve Henson*
10227
10228 * Replace the highly broken and deprecated SPKAC certification method with
10229   the updated NID creation version. This should correctly handle UTF8.
10230
10231   *Steve Henson*
10232
10233 * Implement RFC5746. Re-enable renegotiation but require the extension
10234   as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
10235   turns out to be a bad idea. It has been replaced by
10236   SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
10237   SSL_CTX_set_options(). This is really not recommended unless you
10238   know what you are doing.
10239
10240   *Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson*
10241
10242 * Fixes to stateless session resumption handling. Use initial_ctx when
10243   issuing and attempting to decrypt tickets in case it has changed during
10244   servername handling. Use a non-zero length session ID when attempting
10245   stateless session resumption: this makes it possible to determine if
10246   a resumption has occurred immediately after receiving server hello
10247   (several places in OpenSSL subtly assume this) instead of later in
10248   the handshake.
10249
10250   *Steve Henson*
10251
10252 * The functions ENGINE_ctrl(), OPENSSL_isservice(),
10253   CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
10254   fixes for a few places where the return code is not checked
10255   correctly.
10256
10257   *Julia Lawall <julia@diku.dk>*
10258
10259 * Add --strict-warnings option to Configure script to include devteam
10260   warnings in other configurations.
10261
10262   *Steve Henson*
10263
10264 * Add support for --libdir option and LIBDIR variable in makefiles. This
10265   makes it possible to install openssl libraries in locations which
10266   have names other than "lib", for example "/usr/lib64" which some
10267   systems need.
10268
10269   *Steve Henson, based on patch from Jeremy Utley*
10270
10271 * Don't allow the use of leading 0x80 in OIDs. This is a violation of
10272   X690 8.9.12 and can produce some misleading textual output of OIDs.
10273
10274   *Steve Henson, reported by Dan Kaminsky*
10275
10276 * Delete MD2 from algorithm tables. This follows the recommendation in
10277   several standards that it is not used in new applications due to
10278   several cryptographic weaknesses. For binary compatibility reasons
10279   the MD2 API is still compiled in by default.
10280
10281   *Steve Henson*
10282
10283 * Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
10284   and restored.
10285
10286   *Steve Henson*
10287
10288 * Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
10289   OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
10290   clash.
10291
10292   *Guenter <lists@gknw.net>*
10293
10294 * Fix the server certificate chain building code to use X509_verify_cert(),
10295   it used to have an ad-hoc builder which was unable to cope with anything
10296   other than a simple chain.
10297
10298   *David Woodhouse <dwmw2@infradead.org>, Steve Henson*
10299
10300 * Don't check self signed certificate signatures in X509_verify_cert()
10301   by default (a flag can override this): it just wastes time without
10302   adding any security. As a useful side effect self signed root CAs
10303   with non-FIPS digests are now usable in FIPS mode.
10304
10305   *Steve Henson*
10306
10307 * In dtls1_process_out_of_seq_message() the check if the current message
10308   is already buffered was missing. For every new message was memory
10309   allocated, allowing an attacker to perform an denial of service attack
10310   with sending out of seq handshake messages until there is no memory
10311   left. Additionally every future message was buffered, even if the
10312   sequence number made no sense and would be part of another handshake.
10313   So only messages with sequence numbers less than 10 in advance will be
10314   buffered.  ([CVE-2009-1378])
10315
10316   *Robin Seggelmann, discovered by Daniel Mentz*
10317
10318 * Records are buffered if they arrive with a future epoch to be
10319   processed after finishing the corresponding handshake. There is
10320   currently no limitation to this buffer allowing an attacker to perform
10321   a DOS attack with sending records with future epochs until there is no
10322   memory left. This patch adds the pqueue_size() function to determine
10323   the size of a buffer and limits the record buffer to 100 entries.
10324   ([CVE-2009-1377])
10325
10326   *Robin Seggelmann, discovered by Daniel Mentz*
10327
10328 * Keep a copy of frag->msg_header.frag_len so it can be used after the
10329   parent structure is freed.  ([CVE-2009-1379])
10330
10331   *Daniel Mentz*
10332
10333 * Handle non-blocking I/O properly in SSL_shutdown() call.
10334
10335   *Darryl Miles <darryl-mailinglists@netbauds.net>*
10336
10337 * Add `2.5.4.*` OIDs
10338
10339   *Ilya O. <vrghost@gmail.com>*
10340
10341### Changes between 0.9.8k and 0.9.8l  [5 Nov 2009]
10342
10343 * Disable renegotiation completely - this fixes a severe security
10344   problem ([CVE-2009-3555]) at the cost of breaking all
10345   renegotiation. Renegotiation can be re-enabled by setting
10346   SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
10347   run-time. This is really not recommended unless you know what
10348   you're doing.
10349
10350   *Ben Laurie*
10351
10352### Changes between 0.9.8j and 0.9.8k  [25 Mar 2009]
10353
10354 * Don't set val to NULL when freeing up structures, it is freed up by
10355   underlying code. If `sizeof(void *) > sizeof(long)` this can result in
10356   zeroing past the valid field. ([CVE-2009-0789])
10357
10358   *Paolo Ganci <Paolo.Ganci@AdNovum.CH>*
10359
10360 * Fix bug where return value of CMS_SignerInfo_verify_content() was not
10361   checked correctly. This would allow some invalid signed attributes to
10362   appear to verify correctly. ([CVE-2009-0591])
10363
10364   *Ivan Nestlerode <inestlerode@us.ibm.com>*
10365
10366 * Reject UniversalString and BMPString types with invalid lengths. This
10367   prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
10368   a legal length. ([CVE-2009-0590])
10369
10370   *Steve Henson*
10371
10372 * Set S/MIME signing as the default purpose rather than setting it
10373   unconditionally. This allows applications to override it at the store
10374   level.
10375
10376   *Steve Henson*
10377
10378 * Permit restricted recursion of ASN1 strings. This is needed in practice
10379   to handle some structures.
10380
10381   *Steve Henson*
10382
10383 * Improve efficiency of mem_gets: don't search whole buffer each time
10384   for a '\n'
10385
10386   *Jeremy Shapiro <jnshapir@us.ibm.com>*
10387
10388 * New -hex option for openssl rand.
10389
10390   *Matthieu Herrb*
10391
10392 * Print out UTF8String and NumericString when parsing ASN1.
10393
10394   *Steve Henson*
10395
10396 * Support NumericString type for name components.
10397
10398   *Steve Henson*
10399
10400 * Allow CC in the environment to override the automatically chosen
10401   compiler. Note that nothing is done to ensure flags work with the
10402   chosen compiler.
10403
10404   *Ben Laurie*
10405
10406### Changes between 0.9.8i and 0.9.8j  [07 Jan 2009]
10407
10408 * Properly check EVP_VerifyFinal() and similar return values
10409   ([CVE-2008-5077]).
10410
10411   *Ben Laurie, Bodo Moeller, Google Security Team*
10412
10413 * Enable TLS extensions by default.
10414
10415   *Ben Laurie*
10416
10417 * Allow the CHIL engine to be loaded, whether the application is
10418   multithreaded or not. (This does not release the developer from the
10419   obligation to set up the dynamic locking callbacks.)
10420
10421   *Sander Temme <sander@temme.net>*
10422
10423 * Use correct exit code if there is an error in dgst command.
10424
10425   *Steve Henson; problem pointed out by Roland Dirlewanger*
10426
10427 * Tweak Configure so that you need to say "experimental-jpake" to enable
10428   JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
10429
10430   *Bodo Moeller*
10431
10432 * Add experimental JPAKE support, including demo authentication in
10433   s_client and s_server.
10434
10435   *Ben Laurie*
10436
10437 * Set the comparison function in v3_addr_canonize().
10438
10439   *Rob Austein <sra@hactrn.net>*
10440
10441 * Add support for XMPP STARTTLS in s_client.
10442
10443   *Philip Paeps <philip@freebsd.org>*
10444
10445 * Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
10446   to ensure that even with this option, only ciphersuites in the
10447   server's preference list will be accepted.  (Note that the option
10448   applies only when resuming a session, so the earlier behavior was
10449   just about the algorithm choice for symmetric cryptography.)
10450
10451   *Bodo Moeller*
10452
10453### Changes between 0.9.8h and 0.9.8i  [15 Sep 2008]
10454
10455 * Fix NULL pointer dereference if a DTLS server received
10456   ChangeCipherSpec as first record ([CVE-2009-1386]).
10457
10458   *PR #1679*
10459
10460 * Fix a state transition in s3_srvr.c and d1_srvr.c
10461   (was using SSL3_ST_CW_CLNT_HELLO_B, should be `..._ST_SW_SRVR_...`).
10462
10463   *Nagendra Modadugu*
10464
10465 * The fix in 0.9.8c that supposedly got rid of unsafe
10466   double-checked locking was incomplete for RSA blinding,
10467   addressing just one layer of what turns out to have been
10468   doubly unsafe triple-checked locking.
10469
10470   So now fix this for real by retiring the MONT_HELPER macro
10471   in crypto/rsa/rsa_eay.c.
10472
10473   *Bodo Moeller; problem pointed out by Marius Schilder*
10474
10475 * Various precautionary measures:
10476
10477   - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
10478
10479   - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
10480     (NB: This would require knowledge of the secret session ticket key
10481     to exploit, in which case you'd be SOL either way.)
10482
10483   - Change bn_nist.c so that it will properly handle input BIGNUMs
10484     outside the expected range.
10485
10486   - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
10487     builds.
10488
10489   *Neel Mehta, Bodo Moeller*
10490
10491 * Allow engines to be "soft loaded" - i.e. optionally don't die if
10492   the load fails. Useful for distros.
10493
10494   *Ben Laurie and the FreeBSD team*
10495
10496 * Add support for Local Machine Keyset attribute in PKCS#12 files.
10497
10498   *Steve Henson*
10499
10500 * Fix BN_GF2m_mod_arr() top-bit cleanup code.
10501
10502   *Huang Ying*
10503
10504 * Expand ENGINE to support engine supplied SSL client certificate functions.
10505
10506   This work was sponsored by Logica.
10507
10508   *Steve Henson*
10509
10510 * Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
10511   keystores. Support for SSL/TLS client authentication too.
10512   Not compiled unless enable-capieng specified to Configure.
10513
10514   This work was sponsored by Logica.
10515
10516   *Steve Henson*
10517
10518 * Fix bug in X509_ATTRIBUTE creation: don't set attribute using
10519   ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
10520   attribute creation routines such as certificate requests and PKCS#12
10521   files.
10522
10523   *Steve Henson*
10524
10525### Changes between 0.9.8g and 0.9.8h  [28 May 2008]
10526
10527 * Fix flaw if 'Server Key exchange message' is omitted from a TLS
10528   handshake which could lead to a client crash as found using the
10529   Codenomicon TLS test suite ([CVE-2008-1672])
10530
10531   *Steve Henson, Mark Cox*
10532
10533 * Fix double free in TLS server name extensions which could lead to
10534   a remote crash found by Codenomicon TLS test suite ([CVE-2008-0891])
10535
10536   *Joe Orton*
10537
10538 * Clear error queue in SSL_CTX_use_certificate_chain_file()
10539
10540   Clear the error queue to ensure that error entries left from
10541   older function calls do not interfere with the correct operation.
10542
10543   *Lutz Jaenicke, Erik de Castro Lopo*
10544
10545 * Remove root CA certificates of commercial CAs:
10546
10547   The OpenSSL project does not recommend any specific CA and does not
10548   have any policy with respect to including or excluding any CA.
10549   Therefore it does not make any sense to ship an arbitrary selection
10550   of root CA certificates with the OpenSSL software.
10551
10552   *Lutz Jaenicke*
10553
10554 * RSA OAEP patches to fix two separate invalid memory reads.
10555   The first one involves inputs when 'lzero' is greater than
10556   'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
10557   before the beginning of from). The second one involves inputs where
10558   the 'db' section contains nothing but zeroes (there is a one-byte
10559   invalid read after the end of 'db').
10560
10561   *Ivan Nestlerode <inestlerode@us.ibm.com>*
10562
10563 * Partial backport from 0.9.9-dev:
10564
10565   Introduce bn_mul_mont (dedicated Montgomery multiplication
10566   procedure) as a candidate for BIGNUM assembler implementation.
10567   While 0.9.9-dev uses assembler for various architectures, only
10568   x86_64 is available by default here in the 0.9.8 branch, and
10569   32-bit x86 is available through a compile-time setting.
10570
10571   To try the 32-bit x86 assembler implementation, use Configure
10572   option "enable-montasm" (which exists only for this backport).
10573
10574   As "enable-montasm" for 32-bit x86 disclaims code stability
10575   anyway, in this constellation we activate additional code
10576   backported from 0.9.9-dev for further performance improvements,
10577   namely BN_from_montgomery_word.  (To enable this otherwise,
10578   e.g. x86_64, try `-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD`.)
10579
10580   *Andy Polyakov (backport partially by Bodo Moeller)*
10581
10582 * Add TLS session ticket callback. This allows an application to set
10583   TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
10584   values. This is useful for key rollover for example where several key
10585   sets may exist with different names.
10586
10587   *Steve Henson*
10588
10589 * Reverse ENGINE-internal logic for caching default ENGINE handles.
10590   This was broken until now in 0.9.8 releases, such that the only way
10591   a registered ENGINE could be used (assuming it initialises
10592   successfully on the host) was to explicitly set it as the default
10593   for the relevant algorithms. This is in contradiction with 0.9.7
10594   behaviour and the documentation. With this fix, when an ENGINE is
10595   registered into a given algorithm's table of implementations, the
10596   'uptodate' flag is reset so that auto-discovery will be used next
10597   time a new context for that algorithm attempts to select an
10598   implementation.
10599
10600   *Ian Lister (tweaked by Geoff Thorpe)*
10601
10602 * Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
10603   implementation in the following ways:
10604
10605   Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
10606   hard coded.
10607
10608   Lack of BER streaming support means one pass streaming processing is
10609   only supported if data is detached: setting the streaming flag is
10610   ignored for embedded content.
10611
10612   CMS support is disabled by default and must be explicitly enabled
10613   with the enable-cms configuration option.
10614
10615   *Steve Henson*
10616
10617 * Update the GMP engine glue to do direct copies between BIGNUM and
10618   mpz_t when openssl and GMP use the same limb size. Otherwise the
10619   existing "conversion via a text string export" trick is still used.
10620
10621   *Paul Sheer <paulsheer@gmail.com>*
10622
10623 * Zlib compression BIO. This is a filter BIO which compressed and
10624   uncompresses any data passed through it.
10625
10626   *Steve Henson*
10627
10628 * Add AES_wrap_key() and AES_unwrap_key() functions to implement
10629   RFC3394 compatible AES key wrapping.
10630
10631   *Steve Henson*
10632
10633 * Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
10634   sets string data without copying. X509_ALGOR_set0() and
10635   X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
10636   data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
10637   from an X509_ATTRIBUTE structure optionally checking it occurs only
10638   once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
10639   data.
10640
10641   *Steve Henson*
10642
10643 * Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
10644   to get the expected BN_FLG_CONSTTIME behavior.
10645
10646   *Bodo Moeller (Google)*
10647
10648 * Netware support:
10649
10650   - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
10651   - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
10652   - added some more tests to do_tests.pl
10653   - fixed RunningProcess usage so that it works with newer LIBC NDKs too
10654   - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
10655   - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
10656     netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
10657   - various changes to netware.pl to enable gcc-cross builds on Win32
10658     platform
10659   - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
10660   - various changes to fix missing prototype warnings
10661   - fixed x86nasm.pl to create correct asm files for NASM COFF output
10662   - added AES, WHIRLPOOL and CPUID assembler code to build files
10663   - added missing AES assembler make rules to mk1mf.pl
10664   - fixed order of includes in `apps/ocsp.c` so that `e_os.h` settings apply
10665
10666   *Guenter Knauf <eflash@gmx.net>*
10667
10668 * Implement certificate status request TLS extension defined in RFC3546.
10669   A client can set the appropriate parameters and receive the encoded
10670   OCSP response via a callback. A server can query the supplied parameters
10671   and set the encoded OCSP response in the callback. Add simplified examples
10672   to s_client and s_server.
10673
10674   *Steve Henson*
10675
10676### Changes between 0.9.8f and 0.9.8g  [19 Oct 2007]
10677
10678 * Fix various bugs:
10679   + Binary incompatibility of ssl_ctx_st structure
10680   + DTLS interoperation with non-compliant servers
10681   + Don't call get_session_cb() without proposed session
10682   + Fix ia64 assembler code
10683
10684   *Andy Polyakov, Steve Henson*
10685
10686### Changes between 0.9.8e and 0.9.8f  [11 Oct 2007]
10687
10688 * DTLS Handshake overhaul. There were longstanding issues with
10689   OpenSSL DTLS implementation, which were making it impossible for
10690   RFC 4347 compliant client to communicate with OpenSSL server.
10691   Unfortunately just fixing these incompatibilities would "cut off"
10692   pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
10693   server keeps tolerating non RFC compliant syntax. The opposite is
10694   not true, 0.9.8f client can not communicate with earlier server.
10695   This update even addresses CVE-2007-4995.
10696
10697   *Andy Polyakov*
10698
10699 * Changes to avoid need for function casts in OpenSSL: some compilers
10700   (gcc 4.2 and later) reject their use.
10701   *Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
10702    Steve Henson*
10703
10704 * Add RFC4507 support to OpenSSL. This includes the corrections in
10705   RFC4507bis. The encrypted ticket format is an encrypted encoded
10706   SSL_SESSION structure, that way new session features are automatically
10707   supported.
10708
10709   If a client application caches session in an SSL_SESSION structure
10710   support is transparent because tickets are now stored in the encoded
10711   SSL_SESSION.
10712
10713   The SSL_CTX structure automatically generates keys for ticket
10714   protection in servers so again support should be possible
10715   with no application modification.
10716
10717   If a client or server wishes to disable RFC4507 support then the option
10718   SSL_OP_NO_TICKET can be set.
10719
10720   Add a TLS extension debugging callback to allow the contents of any client
10721   or server extensions to be examined.
10722
10723   This work was sponsored by Google.
10724
10725   *Steve Henson*
10726
10727 * Add initial support for TLS extensions, specifically for the server_name
10728   extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
10729   have new members for a host name.  The SSL data structure has an
10730   additional member `SSL_CTX *initial_ctx` so that new sessions can be
10731   stored in that context to allow for session resumption, even after the
10732   SSL has been switched to a new SSL_CTX in reaction to a client's
10733   server_name extension.
10734
10735   New functions (subject to change):
10736
10737           SSL_get_servername()
10738           SSL_get_servername_type()
10739           SSL_set_SSL_CTX()
10740
10741   New CTRL codes and macros (subject to change):
10742
10743           SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
10744                               - SSL_CTX_set_tlsext_servername_callback()
10745           SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
10746                                    - SSL_CTX_set_tlsext_servername_arg()
10747           SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
10748
10749   openssl s_client has a new '-servername ...' option.
10750
10751   openssl s_server has new options '-servername_host ...', '-cert2 ...',
10752   '-key2 ...', '-servername_fatal' (subject to change).  This allows
10753   testing the HostName extension for a specific single host name ('-cert'
10754   and '-key' remain fallbacks for handshakes without HostName
10755   negotiation).  If the unrecognized_name alert has to be sent, this by
10756   default is a warning; it becomes fatal with the '-servername_fatal'
10757   option.
10758
10759   *Peter Sylvester,  Remy Allais, Christophe Renou, Steve Henson*
10760
10761 * Add AES and SSE2 assembly language support to VC++ build.
10762
10763   *Steve Henson*
10764
10765 * Mitigate attack on final subtraction in Montgomery reduction.
10766
10767   *Andy Polyakov*
10768
10769 * Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
10770   (which previously caused an internal error).
10771
10772   *Bodo Moeller*
10773
10774 * Squeeze another 10% out of IGE mode when in != out.
10775
10776   *Ben Laurie*
10777
10778 * AES IGE mode speedup.
10779
10780   *Dean Gaudet (Google)*
10781
10782 * Add the Korean symmetric 128-bit cipher SEED (see
10783   <http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp>) and
10784   add SEED ciphersuites from RFC 4162:
10785
10786           TLS_RSA_WITH_SEED_CBC_SHA      =  "SEED-SHA"
10787           TLS_DHE_DSS_WITH_SEED_CBC_SHA  =  "DHE-DSS-SEED-SHA"
10788           TLS_DHE_RSA_WITH_SEED_CBC_SHA  =  "DHE-RSA-SEED-SHA"
10789           TLS_DH_anon_WITH_SEED_CBC_SHA  =  "ADH-SEED-SHA"
10790
10791   To minimize changes between patchlevels in the OpenSSL 0.9.8
10792   series, SEED remains excluded from compilation unless OpenSSL
10793   is configured with 'enable-seed'.
10794
10795   *KISA, Bodo Moeller*
10796
10797 * Mitigate branch prediction attacks, which can be practical if a
10798   single processor is shared, allowing a spy process to extract
10799   information.  For detailed background information, see
10800   <http://eprint.iacr.org/2007/039> (O. Aciicmez, S. Gueron,
10801   J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
10802   and Necessary Software Countermeasures").  The core of the change
10803   are new versions BN_div_no_branch() and
10804   BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
10805   respectively, which are slower, but avoid the security-relevant
10806   conditional branches.  These are automatically called by BN_div()
10807   and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
10808   of the input BIGNUMs.  Also, BN_is_bit_set() has been changed to
10809   remove a conditional branch.
10810
10811   BN_FLG_CONSTTIME is the new name for the previous
10812   BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
10813   modular exponentiation.  (Since OpenSSL 0.9.7h, setting this flag
10814   in the exponent causes BN_mod_exp_mont() to use the alternative
10815   implementation in BN_mod_exp_mont_consttime().)  The old name
10816   remains as a deprecated alias.
10817
10818   Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
10819   RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
10820   constant-time implementations for more than just exponentiation.
10821   Here too the old name is kept as a deprecated alias.
10822
10823   BN_BLINDING_new() will now use BN_dup() for the modulus so that
10824   the BN_BLINDING structure gets an independent copy of the
10825   modulus.  This means that the previous `BIGNUM *m` argument to
10826   BN_BLINDING_new() and to BN_BLINDING_create_param() now
10827   essentially becomes `const BIGNUM *m`, although we can't actually
10828   change this in the header file before 0.9.9.  It allows
10829   RSA_setup_blinding() to use BN_with_flags() on the modulus to
10830   enable BN_FLG_CONSTTIME.
10831
10832   *Matthew D Wood (Intel Corp)*
10833
10834 * In the SSL/TLS server implementation, be strict about session ID
10835   context matching (which matters if an application uses a single
10836   external cache for different purposes).  Previously,
10837   out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
10838   set.  This did ensure strict client verification, but meant that,
10839   with applications using a single external cache for quite
10840   different requirements, clients could circumvent ciphersuite
10841   restrictions for a given session ID context by starting a session
10842   in a different context.
10843
10844   *Bodo Moeller*
10845
10846 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
10847   a ciphersuite string such as "DEFAULT:RSA" cannot enable
10848   authentication-only ciphersuites.
10849
10850   *Bodo Moeller*
10851
10852 * Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
10853   not complete and could lead to a possible single byte overflow
10854   ([CVE-2007-5135]) [Ben Laurie]
10855
10856### Changes between 0.9.8d and 0.9.8e  [23 Feb 2007]
10857
10858 * Since AES128 and AES256 (and similarly Camellia128 and
10859   Camellia256) share a single mask bit in the logic of
10860   ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
10861   kludge to work properly if AES128 is available and AES256 isn't
10862   (or if Camellia128 is available and Camellia256 isn't).
10863
10864   *Victor Duchovni*
10865
10866 * Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
10867   (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
10868   When a point or a seed is encoded in a BIT STRING, we need to
10869   prevent the removal of trailing zero bits to get the proper DER
10870   encoding.  (By default, crypto/asn1/a_bitstr.c assumes the case
10871   of a NamedBitList, for which trailing 0 bits need to be removed.)
10872
10873   *Bodo Moeller*
10874
10875 * Have SSL/TLS server implementation tolerate "mismatched" record
10876   protocol version while receiving ClientHello even if the
10877   ClientHello is fragmented.  (The server can't insist on the
10878   particular protocol version it has chosen before the ServerHello
10879   message has informed the client about his choice.)
10880
10881   *Bodo Moeller*
10882
10883 * Add RFC 3779 support.
10884
10885   *Rob Austein for ARIN, Ben Laurie*
10886
10887 * Load error codes if they are not already present instead of using a
10888   static variable. This allows them to be cleanly unloaded and reloaded.
10889   Improve header file function name parsing.
10890
10891   *Steve Henson*
10892
10893 * extend SMTP and IMAP protocol emulation in s_client to use EHLO
10894   or CAPABILITY handshake as required by RFCs.
10895
10896   *Goetz Babin-Ebell*
10897
10898### Changes between 0.9.8c and 0.9.8d  [28 Sep 2006]
10899
10900 * Introduce limits to prevent malicious keys being able to
10901   cause a denial of service.  ([CVE-2006-2940])
10902
10903   *Steve Henson, Bodo Moeller*
10904
10905 * Fix ASN.1 parsing of certain invalid structures that can result
10906   in a denial of service.  ([CVE-2006-2937])  [Steve Henson]
10907
10908 * Fix buffer overflow in SSL_get_shared_ciphers() function.
10909   ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team]
10910
10911 * Fix SSL client code which could crash if connecting to a
10912   malicious SSLv2 server.  ([CVE-2006-4343])
10913
10914   *Tavis Ormandy and Will Drewry, Google Security Team*
10915
10916 * Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
10917   match only those.  Before that, "AES256-SHA" would be interpreted
10918   as a pattern and match "AES128-SHA" too (since AES128-SHA got
10919   the same strength classification in 0.9.7h) as we currently only
10920   have a single AES bit in the ciphersuite description bitmap.
10921   That change, however, also applied to ciphersuite strings such as
10922   "RC4-MD5" that intentionally matched multiple ciphersuites --
10923   namely, SSL 2.0 ciphersuites in addition to the more common ones
10924   from SSL 3.0/TLS 1.0.
10925
10926   So we change the selection algorithm again: Naming an explicit
10927   ciphersuite selects this one ciphersuite, and any other similar
10928   ciphersuite (same bitmap) from *other* protocol versions.
10929   Thus, "RC4-MD5" again will properly select both the SSL 2.0
10930   ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
10931
10932   Since SSL 2.0 does not have any ciphersuites for which the
10933   128/256 bit distinction would be relevant, this works for now.
10934   The proper fix will be to use different bits for AES128 and
10935   AES256, which would have avoided the problems from the beginning;
10936   however, bits are scarce, so we can only do this in a new release
10937   (not just a patchlevel) when we can change the SSL_CIPHER
10938   definition to split the single 'unsigned long mask' bitmap into
10939   multiple values to extend the available space.
10940
10941   *Bodo Moeller*
10942
10943### Changes between 0.9.8b and 0.9.8c  [05 Sep 2006]
10944
10945 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
10946   ([CVE-2006-4339])  [Ben Laurie and Google Security Team]
10947
10948 * Add AES IGE and biIGE modes.
10949
10950   *Ben Laurie*
10951
10952 * Change the Unix randomness entropy gathering to use poll() when
10953   possible instead of select(), since the latter has some
10954   undesirable limitations.
10955
10956   *Darryl Miles via Richard Levitte and Bodo Moeller*
10957
10958 * Disable "ECCdraft" ciphersuites more thoroughly.  Now special
10959   treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
10960   cannot be implicitly activated as part of, e.g., the "AES" alias.
10961   However, please upgrade to OpenSSL 0.9.9[-dev] for
10962   non-experimental use of the ECC ciphersuites to get TLS extension
10963   support, which is required for curve and point format negotiation
10964   to avoid potential handshake problems.
10965
10966   *Bodo Moeller*
10967
10968 * Disable rogue ciphersuites:
10969
10970   - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
10971   - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
10972   - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
10973
10974   The latter two were purportedly from
10975   draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
10976   appear there.
10977
10978   Also deactivate the remaining ciphersuites from
10979   draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
10980   unofficial, and the ID has long expired.
10981
10982   *Bodo Moeller*
10983
10984 * Fix RSA blinding Heisenbug (problems sometimes occurred on
10985   dual-core machines) and other potential thread-safety issues.
10986
10987   *Bodo Moeller*
10988
10989 * Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
10990   versions), which is now available for royalty-free use
10991   (see <http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html>).
10992   Also, add Camellia TLS ciphersuites from RFC 4132.
10993
10994   To minimize changes between patchlevels in the OpenSSL 0.9.8
10995   series, Camellia remains excluded from compilation unless OpenSSL
10996   is configured with 'enable-camellia'.
10997
10998   *NTT*
10999
11000 * Disable the padding bug check when compression is in use. The padding
11001   bug check assumes the first packet is of even length, this is not
11002   necessarily true if compression is enabled and can result in false
11003   positives causing handshake failure. The actual bug test is ancient
11004   code so it is hoped that implementations will either have fixed it by
11005   now or any which still have the bug do not support compression.
11006
11007   *Steve Henson*
11008
11009### Changes between 0.9.8a and 0.9.8b  [04 May 2006]
11010
11011 * When applying a cipher rule check to see if string match is an explicit
11012   cipher suite and only match that one cipher suite if it is.
11013
11014   *Steve Henson*
11015
11016 * Link in manifests for VC++ if needed.
11017
11018   *Austin Ziegler <halostatue@gmail.com>*
11019
11020 * Update support for ECC-based TLS ciphersuites according to
11021   draft-ietf-tls-ecc-12.txt with proposed changes (but without
11022   TLS extensions, which are supported starting with the 0.9.9
11023   branch, not in the OpenSSL 0.9.8 branch).
11024
11025   *Douglas Stebila*
11026
11027 * New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
11028   opaque EVP_CIPHER_CTX handling.
11029
11030   *Steve Henson*
11031
11032 * Fixes and enhancements to zlib compression code. We now only use
11033   "zlib1.dll" and use the default `__cdecl` calling convention on Win32
11034   to conform with the standards mentioned here:
11035   <http://www.zlib.net/DLL_FAQ.txt>
11036   Static zlib linking now works on Windows and the new --with-zlib-include
11037   --with-zlib-lib options to Configure can be used to supply the location
11038   of the headers and library. Gracefully handle case where zlib library
11039   can't be loaded.
11040
11041   *Steve Henson*
11042
11043 * Several fixes and enhancements to the OID generation code. The old code
11044   sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
11045   handle numbers larger than ULONG_MAX, truncated printing and had a
11046   non standard OBJ_obj2txt() behaviour.
11047
11048   *Steve Henson*
11049
11050 * Add support for building of engines under engine/ as shared libraries
11051   under VC++ build system.
11052
11053   *Steve Henson*
11054
11055 * Corrected the numerous bugs in the Win32 path splitter in DSO.
11056   Hopefully, we will not see any false combination of paths any more.
11057
11058   *Richard Levitte*
11059
11060### Changes between 0.9.8 and 0.9.8a  [11 Oct 2005]
11061
11062 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
11063   (part of SSL_OP_ALL).  This option used to disable the
11064   countermeasure against man-in-the-middle protocol-version
11065   rollback in the SSL 2.0 server implementation, which is a bad
11066   idea.  ([CVE-2005-2969])
11067
11068   *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
11069   for Information Security, National Institute of Advanced Industrial
11070   Science and Technology [AIST], Japan)*
11071
11072 * Add two function to clear and return the verify parameter flags.
11073
11074   *Steve Henson*
11075
11076 * Keep cipherlists sorted in the source instead of sorting them at
11077   runtime, thus removing the need for a lock.
11078
11079   *Nils Larsch*
11080
11081 * Avoid some small subgroup attacks in Diffie-Hellman.
11082
11083   *Nick Mathewson and Ben Laurie*
11084
11085 * Add functions for well-known primes.
11086
11087   *Nick Mathewson*
11088
11089 * Extended Windows CE support.
11090
11091   *Satoshi Nakamura and Andy Polyakov*
11092
11093 * Initialize SSL_METHOD structures at compile time instead of during
11094   runtime, thus removing the need for a lock.
11095
11096   *Steve Henson*
11097
11098 * Make PKCS7_decrypt() work even if no certificate is supplied by
11099   attempting to decrypt each encrypted key in turn. Add support to
11100   smime utility.
11101
11102   *Steve Henson*
11103
11104### Changes between 0.9.7h and 0.9.8  [05 Jul 2005]
11105
11106[NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
11107OpenSSL 0.9.8.]
11108
11109 * Add libcrypto.pc and libssl.pc for those who feel they need them.
11110
11111   *Richard Levitte*
11112
11113 * Change CA.sh and CA.pl so they don't bundle the CSR and the private
11114   key into the same file any more.
11115
11116   *Richard Levitte*
11117
11118 * Add initial support for Win64, both IA64 and AMD64/x64 flavors.
11119
11120   *Andy Polyakov*
11121
11122 * Add -utf8 command line and config file option to 'ca'.
11123
11124   *Stefan <stf@udoma.org*
11125
11126 * Removed the macro des_crypt(), as it seems to conflict with some
11127   libraries.  Use DES_crypt().
11128
11129   *Richard Levitte*
11130
11131 * Correct naming of the 'chil' and '4758cca' ENGINEs. This
11132   involves renaming the source and generated shared-libs for
11133   both. The engines will accept the corrected or legacy ids
11134   ('ncipher' and '4758_cca' respectively) when binding. NB,
11135   this only applies when building 'shared'.
11136
11137   *Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe*
11138
11139 * Add attribute functions to EVP_PKEY structure. Modify
11140   PKCS12_create() to recognize a CSP name attribute and
11141   use it. Make -CSP option work again in pkcs12 utility.
11142
11143   *Steve Henson*
11144
11145 * Add new functionality to the bn blinding code:
11146   - automatic re-creation of the BN_BLINDING parameters after
11147     a fixed number of uses (currently 32)
11148   - add new function for parameter creation
11149   - introduce flags to control the update behaviour of the
11150     BN_BLINDING parameters
11151   - hide BN_BLINDING structure
11152   Add a second BN_BLINDING slot to the RSA structure to improve
11153   performance when a single RSA object is shared among several
11154   threads.
11155
11156   *Nils Larsch*
11157
11158 * Add support for DTLS.
11159
11160   *Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie*
11161
11162 * Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
11163   to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
11164
11165   *Walter Goulet*
11166
11167 * Remove buggy and incomplete DH cert support from
11168   ssl/ssl_rsa.c and ssl/s3_both.c
11169
11170   *Nils Larsch*
11171
11172 * Use SHA-1 instead of MD5 as the default digest algorithm for
11173   the `apps/openssl` commands.
11174
11175   *Nils Larsch*
11176
11177 * Compile clean with "-Wall -Wmissing-prototypes
11178   -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
11179   DEBUG_SAFESTACK must also be set.
11180
11181   *Ben Laurie*
11182
11183 * Change ./Configure so that certain algorithms can be disabled by default.
11184   The new counterpiece to "no-xxx" is "enable-xxx".
11185
11186   The patented RC5 and MDC2 algorithms will now be disabled unless
11187   "enable-rc5" and "enable-mdc2", respectively, are specified.
11188
11189   (IDEA remains enabled despite being patented.  This is because IDEA
11190   is frequently required for interoperability, and there is no license
11191   fee for non-commercial use.  As before, "no-idea" can be used to
11192   avoid this algorithm.)
11193
11194   *Bodo Moeller*
11195
11196 * Add processing of proxy certificates (see RFC 3820).  This work was
11197   sponsored by KTH (The Royal Institute of Technology in Stockholm) and
11198   EGEE (Enabling Grids for E-science in Europe).
11199
11200   *Richard Levitte*
11201
11202 * RC4 performance overhaul on modern architectures/implementations, such
11203   as Intel P4, IA-64 and AMD64.
11204
11205   *Andy Polyakov*
11206
11207 * New utility extract-section.pl. This can be used specify an alternative
11208   section number in a pod file instead of having to treat each file as
11209   a separate case in Makefile. This can be done by adding two lines to the
11210   pod file:
11211
11212   =for comment openssl_section:XXX
11213
11214   The blank line is mandatory.
11215
11216   *Steve Henson*
11217
11218 * New arguments -certform, -keyform and -pass for s_client and s_server
11219   to allow alternative format key and certificate files and passphrase
11220   sources.
11221
11222   *Steve Henson*
11223
11224 * New structure X509_VERIFY_PARAM which combines current verify parameters,
11225   update associated structures and add various utility functions.
11226
11227   Add new policy related verify parameters, include policy checking in
11228   standard verify code. Enhance 'smime' application with extra parameters
11229   to support policy checking and print out.
11230
11231   *Steve Henson*
11232
11233 * Add a new engine to support VIA PadLock ACE extensions in the VIA C3
11234   Nehemiah processors. These extensions support AES encryption in hardware
11235   as well as RNG (though RNG support is currently disabled).
11236
11237   *Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov*
11238
11239 * Deprecate `BN_[get|set]_params()` functions (they were ignored internally).
11240
11241   *Geoff Thorpe*
11242
11243 * New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
11244
11245   *Andy Polyakov and a number of other people*
11246
11247 * Improved PowerPC platform support. Most notably BIGNUM assembler
11248   implementation contributed by IBM.
11249
11250   *Suresh Chari, Peter Waltenberg, Andy Polyakov*
11251
11252 * The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
11253   exponent rather than 'unsigned long'. There is a corresponding change to
11254   the new 'rsa_keygen' element of the RSA_METHOD structure.
11255
11256   *Jelte Jansen, Geoff Thorpe*
11257
11258 * Functionality for creating the initial serial number file is now
11259   moved from CA.pl to the 'ca' utility with a new option -create_serial.
11260
11261   (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
11262   number file to 1, which is bound to cause problems.  To avoid
11263   the problems while respecting compatibility between different 0.9.7
11264   patchlevels, 0.9.7e  employed 'openssl x509 -next_serial' in
11265   CA.pl for serial number initialization.  With the new release 0.9.8,
11266   we can fix the problem directly in the 'ca' utility.)
11267
11268   *Steve Henson*
11269
11270 * Reduced header interdependencies by declaring more opaque objects in
11271   ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
11272   give fewer recursive includes, which could break lazy source code - so
11273   this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
11274   developers should define this symbol when building and using openssl to
11275   ensure they track the recommended behaviour, interfaces, [etc], but
11276   backwards-compatible behaviour prevails when this isn't defined.
11277
11278   *Geoff Thorpe*
11279
11280 * New function X509_POLICY_NODE_print() which prints out policy nodes.
11281
11282   *Steve Henson*
11283
11284 * Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
11285   This will generate a random key of the appropriate length based on the
11286   cipher context. The EVP_CIPHER can provide its own random key generation
11287   routine to support keys of a specific form. This is used in the des and
11288   3des routines to generate a key of the correct parity. Update S/MIME
11289   code to use new functions and hence generate correct parity DES keys.
11290   Add EVP_CHECK_DES_KEY #define to return an error if the key is not
11291   valid (weak or incorrect parity).
11292
11293   *Steve Henson*
11294
11295 * Add a local set of CRLs that can be used by X509_verify_cert() as well
11296   as looking them up. This is useful when the verified structure may contain
11297   CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
11298   present unless the new PKCS7_NO_CRL flag is asserted.
11299
11300   *Steve Henson*
11301
11302 * Extend ASN1 oid configuration module. It now additionally accepts the
11303   syntax:
11304
11305   shortName = some long name, 1.2.3.4
11306
11307   *Steve Henson*
11308
11309 * Reimplemented the BN_CTX implementation. There is now no more static
11310   limitation on the number of variables it can handle nor the depth of the
11311   "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
11312   information can now expand as required, and rather than having a single
11313   static array of bignums, BN_CTX now uses a linked-list of such arrays
11314   allowing it to expand on demand whilst maintaining the usefulness of
11315   BN_CTX's "bundling".
11316
11317   *Geoff Thorpe*
11318
11319 * Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
11320   to allow all RSA operations to function using a single BN_CTX.
11321
11322   *Geoff Thorpe*
11323
11324 * Preliminary support for certificate policy evaluation and checking. This
11325   is initially intended to pass the tests outlined in "Conformance Testing
11326   of Relying Party Client Certificate Path Processing Logic" v1.07.
11327
11328   *Steve Henson*
11329
11330 * bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
11331   remained unused and not that useful. A variety of other little bignum
11332   tweaks and fixes have also been made continuing on from the audit (see
11333   below).
11334
11335   *Geoff Thorpe*
11336
11337 * Constify all or almost all d2i, c2i, s2i and r2i functions, along with
11338   associated ASN1, EVP and SSL functions and old ASN1 macros.
11339
11340   *Richard Levitte*
11341
11342 * BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
11343   and this should never fail. So the return value from the use of
11344   BN_set_word() (which can fail due to needless expansion) is now deprecated;
11345   if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
11346
11347   *Geoff Thorpe*
11348
11349 * BN_CTX_get() should return zero-valued bignums, providing the same
11350   initialised value as BN_new().
11351
11352   *Geoff Thorpe, suggested by Ulf Möller*
11353
11354 * Support for inhibitAnyPolicy certificate extension.
11355
11356   *Steve Henson*
11357
11358 * An audit of the BIGNUM code is underway, for which debugging code is
11359   enabled when BN_DEBUG is defined. This makes stricter enforcements on what
11360   is considered valid when processing BIGNUMs, and causes execution to
11361   assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
11362   further steps are taken to deliberately pollute unused data in BIGNUM
11363   structures to try and expose faulty code further on. For now, openssl will
11364   (in its default mode of operation) continue to tolerate the inconsistent
11365   forms that it has tolerated in the past, but authors and packagers should
11366   consider trying openssl and their own applications when compiled with
11367   these debugging symbols defined. It will help highlight potential bugs in
11368   their own code, and will improve the test coverage for OpenSSL itself. At
11369   some point, these tighter rules will become openssl's default to improve
11370   maintainability, though the assert()s and other overheads will remain only
11371   in debugging configurations. See bn.h for more details.
11372
11373   *Geoff Thorpe, Nils Larsch, Ulf Möller*
11374
11375 * BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
11376   that can only be obtained through BN_CTX_new() (which implicitly
11377   initialises it). The presence of this function only made it possible
11378   to overwrite an existing structure (and cause memory leaks).
11379
11380   *Geoff Thorpe*
11381
11382 * Because of the callback-based approach for implementing LHASH as a
11383   template type, lh_insert() adds opaque objects to hash-tables and
11384   lh_doall() or lh_doall_arg() are typically used with a destructor callback
11385   to clean up those corresponding objects before destroying the hash table
11386   (and losing the object pointers). So some over-zealous constifications in
11387   LHASH have been relaxed so that lh_insert() does not take (nor store) the
11388   objects as "const" and the `lh_doall[_arg]` callback wrappers are not
11389   prototyped to have "const" restrictions on the object pointers they are
11390   given (and so aren't required to cast them away any more).
11391
11392   *Geoff Thorpe*
11393
11394 * The tmdiff.h API was so ugly and minimal that our own timing utility
11395   (speed) prefers to use its own implementation. The two implementations
11396   haven't been consolidated as yet (volunteers?) but the tmdiff API has had
11397   its object type properly exposed (MS_TM) instead of casting to/from
11398   `char *`. This may still change yet if someone realises MS_TM and
11399   `ms_time_***`
11400   aren't necessarily the greatest nomenclatures - but this is what was used
11401   internally to the implementation so I've used that for now.
11402
11403   *Geoff Thorpe*
11404
11405 * Ensure that deprecated functions do not get compiled when
11406   OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
11407   the self-tests were still using deprecated key-generation functions so
11408   these have been updated also.
11409
11410   *Geoff Thorpe*
11411
11412 * Reorganise PKCS#7 code to separate the digest location functionality
11413   into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest().
11414   New function PKCS7_set_digest() to set the digest type for PKCS#7
11415   digestedData type. Add additional code to correctly generate the
11416   digestedData type and add support for this type in PKCS7 initialization
11417   functions.
11418
11419   *Steve Henson*
11420
11421 * New function PKCS7_set0_type_other() this initializes a PKCS7
11422   structure of type "other".
11423
11424   *Steve Henson*
11425
11426 * Fix prime generation loop in crypto/bn/bn_prime.pl by making
11427   sure the loop does correctly stop and breaking ("division by zero")
11428   modulus operations are not performed. The (pre-generated) prime
11429   table crypto/bn/bn_prime.h was already correct, but it could not be
11430   re-generated on some platforms because of the "division by zero"
11431   situation in the script.
11432
11433   *Ralf S. Engelschall*
11434
11435 * Update support for ECC-based TLS ciphersuites according to
11436   draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
11437   SHA-1 now is only used for "small" curves (where the
11438   representation of a field element takes up to 24 bytes); for
11439   larger curves, the field element resulting from ECDH is directly
11440   used as premaster secret.
11441
11442   *Douglas Stebila (Sun Microsystems Laboratories)*
11443
11444 * Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
11445   curve secp160r1 to the tests.
11446
11447   *Douglas Stebila (Sun Microsystems Laboratories)*
11448
11449 * Add the possibility to load symbols globally with DSO.
11450
11451   *Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte*
11452
11453 * Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
11454   control of the error stack.
11455
11456   *Richard Levitte*
11457
11458 * Add support for STORE in ENGINE.
11459
11460   *Richard Levitte*
11461
11462 * Add the STORE type.  The intention is to provide a common interface
11463   to certificate and key stores, be they simple file-based stores, or
11464   HSM-type store, or LDAP stores, or...
11465   NOTE: The code is currently UNTESTED and isn't really used anywhere.
11466
11467   *Richard Levitte*
11468
11469 * Add a generic structure called OPENSSL_ITEM.  This can be used to
11470   pass a list of arguments to any function as well as provide a way
11471   for a function to pass data back to the caller.
11472
11473   *Richard Levitte*
11474
11475 * Add the functions BUF_strndup() and BUF_memdup().  BUF_strndup()
11476   works like BUF_strdup() but can be used to duplicate a portion of
11477   a string.  The copy gets NUL-terminated.  BUF_memdup() duplicates
11478   a memory area.
11479
11480   *Richard Levitte*
11481
11482 * Add the function sk_find_ex() which works like sk_find(), but will
11483   return an index to an element even if an exact match couldn't be
11484   found.  The index is guaranteed to point at the element where the
11485   searched-for key would be inserted to preserve sorting order.
11486
11487   *Richard Levitte*
11488
11489 * Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
11490   takes an extra flags argument for optional functionality.  Currently,
11491   the following flags are defined:
11492
11493      OBJ_BSEARCH_VALUE_ON_NOMATCH
11494      This one gets OBJ_bsearch_ex() to return a pointer to the first
11495      element where the comparing function returns a negative or zero
11496      number.
11497
11498      OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
11499      This one gets OBJ_bsearch_ex() to return a pointer to the first
11500      element where the comparing function returns zero.  This is useful
11501      if there are more than one element where the comparing function
11502      returns zero.
11503
11504   *Richard Levitte*
11505
11506 * Make it possible to create self-signed certificates with 'openssl ca'
11507   in such a way that the self-signed certificate becomes part of the
11508   CA database and uses the same mechanisms for serial number generation
11509   as all other certificate signing.  The new flag '-selfsign' enables
11510   this functionality.  Adapt CA.sh and CA.pl.in.
11511
11512   *Richard Levitte*
11513
11514 * Add functionality to check the public key of a certificate request
11515   against a given private.  This is useful to check that a certificate
11516   request can be signed by that key (self-signing).
11517
11518   *Richard Levitte*
11519
11520 * Make it possible to have multiple active certificates with the same
11521   subject in the CA index file.  This is done only if the keyword
11522   'unique_subject' is set to 'no' in the main CA section (default
11523   if 'CA_default') of the configuration file.  The value is saved
11524   with the database itself in a separate index attribute file,
11525   named like the index file with '.attr' appended to the name.
11526
11527   *Richard Levitte*
11528
11529 * Generate multi-valued AVAs using '+' notation in config files for
11530   req and dirName.
11531
11532   *Steve Henson*
11533
11534 * Support for nameConstraints certificate extension.
11535
11536   *Steve Henson*
11537
11538 * Support for policyConstraints certificate extension.
11539
11540   *Steve Henson*
11541
11542 * Support for policyMappings certificate extension.
11543
11544   *Steve Henson*
11545
11546 * Make sure the default DSA_METHOD implementation only uses its
11547   dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
11548   and change its own handlers to be NULL so as to remove unnecessary
11549   indirection. This lets alternative implementations fallback to the
11550   default implementation more easily.
11551
11552   *Geoff Thorpe*
11553
11554 * Support for directoryName in GeneralName related extensions
11555   in config files.
11556
11557   *Steve Henson*
11558
11559 * Make it possible to link applications using Makefile.shared.
11560   Make that possible even when linking against static libraries!
11561
11562   *Richard Levitte*
11563
11564 * Support for single pass processing for S/MIME signing. This now
11565   means that S/MIME signing can be done from a pipe, in addition
11566   cleartext signing (multipart/signed type) is effectively streaming
11567   and the signed data does not need to be all held in memory.
11568
11569   This is done with a new flag PKCS7_STREAM. When this flag is set
11570   PKCS7_sign() only initializes the PKCS7 structure and the actual signing
11571   is done after the data is output (and digests calculated) in
11572   SMIME_write_PKCS7().
11573
11574   *Steve Henson*
11575
11576 * Add full support for -rpath/-R, both in shared libraries and
11577   applications, at least on the platforms where it's known how
11578   to do it.
11579
11580   *Richard Levitte*
11581
11582 * In crypto/ec/ec_mult.c, implement fast point multiplication with
11583   precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
11584   will now compute a table of multiples of the generator that
11585   makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
11586   faster (notably in the case of a single point multiplication,
11587   scalar * generator).
11588
11589   *Nils Larsch, Bodo Moeller*
11590
11591 * IPv6 support for certificate extensions. The various extensions
11592   which use the IP:a.b.c.d can now take IPv6 addresses using the
11593   formats of RFC1884 2.2 . IPv6 addresses are now also displayed
11594   correctly.
11595
11596   *Steve Henson*
11597
11598 * Added an ENGINE that implements RSA by performing private key
11599   exponentiations with the GMP library. The conversions to and from
11600   GMP's mpz_t format aren't optimised nor are any montgomery forms
11601   cached, and on x86 it appears OpenSSL's own performance has caught up.
11602   However there are likely to be other architectures where GMP could
11603   provide a boost. This ENGINE is not built in by default, but it can be
11604   specified at Configure time and should be accompanied by the necessary
11605   linker additions, eg;
11606           ./config -DOPENSSL_USE_GMP -lgmp
11607
11608   *Geoff Thorpe*
11609
11610 * "openssl engine" will not display ENGINE/DSO load failure errors when
11611   testing availability of engines with "-t" - the old behaviour is
11612   produced by increasing the feature's verbosity with "-tt".
11613
11614   *Geoff Thorpe*
11615
11616 * ECDSA routines: under certain error conditions uninitialized BN objects
11617   could be freed. Solution: make sure initialization is performed early
11618   enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
11619   via PR#459)
11620
11621   *Lutz Jaenicke*
11622
11623 * Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
11624   and DH_METHOD (eg. by ENGINE implementations) to override the normal
11625   software implementations. For DSA and DH, parameter generation can
11626   also be overridden by providing the appropriate method callbacks.
11627
11628   *Geoff Thorpe*
11629
11630 * Change the "progress" mechanism used in key-generation and
11631   primality testing to functions that take a new BN_GENCB pointer in
11632   place of callback/argument pairs. The new API functions have `_ex`
11633   postfixes and the older functions are reimplemented as wrappers for
11634   the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
11635   declarations of the old functions to help (graceful) attempts to
11636   migrate to the new functions. Also, the new key-generation API
11637   functions operate on a caller-supplied key-structure and return
11638   success/failure rather than returning a key or NULL - this is to
11639   help make "keygen" another member function of RSA_METHOD etc.
11640
11641   Example for using the new callback interface:
11642
11643           int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
11644           void *my_arg = ...;
11645           BN_GENCB my_cb;
11646
11647           BN_GENCB_set(&my_cb, my_callback, my_arg);
11648
11649           return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
11650           /* For the meaning of a, b in calls to my_callback(), see the
11651            * documentation of the function that calls the callback.
11652            * cb will point to my_cb; my_arg can be retrieved as cb->arg.
11653            * my_callback should return 1 if it wants BN_is_prime_ex()
11654            * to continue, or 0 to stop.
11655            */
11656
11657   *Geoff Thorpe*
11658
11659 * Change the ZLIB compression method to be stateful, and make it
11660   available to TLS with the number defined in
11661   draft-ietf-tls-compression-04.txt.
11662
11663   *Richard Levitte*
11664
11665 * Add the ASN.1 structures and functions for CertificatePair, which
11666   is defined as follows (according to X.509_4thEditionDraftV6.pdf):
11667
11668           CertificatePair ::= SEQUENCE {
11669              forward         [0]     Certificate OPTIONAL,
11670              reverse         [1]     Certificate OPTIONAL,
11671              -- at least one of the pair shall be present -- }
11672
11673   Also implement the PEM functions to read and write certificate
11674   pairs, and defined the PEM tag as "CERTIFICATE PAIR".
11675
11676   This needed to be defined, mostly for the sake of the LDAP
11677   attribute crossCertificatePair, but may prove useful elsewhere as
11678   well.
11679
11680   *Richard Levitte*
11681
11682 * Make it possible to inhibit symlinking of shared libraries in
11683   Makefile.shared, for Cygwin's sake.
11684
11685   *Richard Levitte*
11686
11687 * Extend the BIGNUM API by creating a function
11688           void BN_set_negative(BIGNUM *a, int neg);
11689   and a macro that behave like
11690           int  BN_is_negative(const BIGNUM *a);
11691
11692   to avoid the need to access 'a->neg' directly in applications.
11693
11694   *Nils Larsch*
11695
11696 * Implement fast modular reduction for pseudo-Mersenne primes
11697   used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
11698   EC_GROUP_new_curve_GFp() will now automatically use this
11699   if applicable.
11700
11701   *Nils Larsch <nla@trustcenter.de>*
11702
11703 * Add new lock type (CRYPTO_LOCK_BN).
11704
11705   *Bodo Moeller*
11706
11707 * Change the ENGINE framework to automatically load engines
11708   dynamically from specific directories unless they could be
11709   found to already be built in or loaded.  Move all the
11710   current engines except for the cryptodev one to a new
11711   directory engines/.
11712   The engines in engines/ are built as shared libraries if
11713   the "shared" options was given to ./Configure or ./config.
11714   Otherwise, they are inserted in libcrypto.a.
11715   /usr/local/ssl/engines is the default directory for dynamic
11716   engines, but that can be overridden at configure time through
11717   the usual use of --prefix and/or --openssldir, and at run
11718   time with the environment variable OPENSSL_ENGINES.
11719
11720   *Geoff Thorpe and Richard Levitte*
11721
11722 * Add Makefile.shared, a helper makefile to build shared
11723   libraries.  Adapt Makefile.org.
11724
11725   *Richard Levitte*
11726
11727 * Add version info to Win32 DLLs.
11728
11729   *Peter 'Luna' Runestig" <peter@runestig.com>*
11730
11731 * Add new 'medium level' PKCS#12 API. Certificates and keys
11732   can be added using this API to created arbitrary PKCS#12
11733   files while avoiding the low-level API.
11734
11735   New options to PKCS12_create(), key or cert can be NULL and
11736   will then be omitted from the output file. The encryption
11737   algorithm NIDs can be set to -1 for no encryption, the mac
11738   iteration count can be set to 0 to omit the mac.
11739
11740   Enhance pkcs12 utility by making the -nokeys and -nocerts
11741   options work when creating a PKCS#12 file. New option -nomac
11742   to omit the mac, NONE can be set for an encryption algorithm.
11743   New code is modified to use the enhanced PKCS12_create()
11744   instead of the low-level API.
11745
11746   *Steve Henson*
11747
11748 * Extend ASN1 encoder to support indefinite length constructed
11749   encoding. This can output sequences tags and octet strings in
11750   this form. Modify pk7_asn1.c to support indefinite length
11751   encoding. This is experimental and needs additional code to
11752   be useful, such as an ASN1 bio and some enhanced streaming
11753   PKCS#7 code.
11754
11755   Extend template encode functionality so that tagging is passed
11756   down to the template encoder.
11757
11758   *Steve Henson*
11759
11760 * Let 'openssl req' fail if an argument to '-newkey' is not
11761   recognized instead of using RSA as a default.
11762
11763   *Bodo Moeller*
11764
11765 * Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
11766   As these are not official, they are not included in "ALL";
11767   the "ECCdraft" ciphersuite group alias can be used to select them.
11768
11769   *Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)*
11770
11771 * Add ECDH engine support.
11772
11773   *Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)*
11774
11775 * Add ECDH in new directory crypto/ecdh/.
11776
11777   *Douglas Stebila (Sun Microsystems Laboratories)*
11778
11779 * Let BN_rand_range() abort with an error after 100 iterations
11780   without success (which indicates a broken PRNG).
11781
11782   *Bodo Moeller*
11783
11784 * Change BN_mod_sqrt() so that it verifies that the input value
11785   is really the square of the return value.  (Previously,
11786   BN_mod_sqrt would show GIGO behaviour.)
11787
11788   *Bodo Moeller*
11789
11790 * Add named elliptic curves over binary fields from X9.62, SECG,
11791   and WAP/WTLS; add OIDs that were still missing.
11792
11793   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11794
11795 * Extend the EC library for elliptic curves over binary fields
11796   (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
11797   New EC_METHOD:
11798
11799           EC_GF2m_simple_method
11800
11801   New API functions:
11802
11803           EC_GROUP_new_curve_GF2m
11804           EC_GROUP_set_curve_GF2m
11805           EC_GROUP_get_curve_GF2m
11806           EC_POINT_set_affine_coordinates_GF2m
11807           EC_POINT_get_affine_coordinates_GF2m
11808           EC_POINT_set_compressed_coordinates_GF2m
11809
11810   Point compression for binary fields is disabled by default for
11811   patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
11812   enable it).
11813
11814   As binary polynomials are represented as BIGNUMs, various members
11815   of the EC_GROUP and EC_POINT data structures can be shared
11816   between the implementations for prime fields and binary fields;
11817   the above `..._GF2m functions` (except for EX_GROUP_new_curve_GF2m)
11818   are essentially identical to their `..._GFp` counterparts.
11819   (For simplicity, the `..._GFp` prefix has been dropped from
11820   various internal method names.)
11821
11822   An internal 'field_div' method (similar to 'field_mul' and
11823   'field_sqr') has been added; this is used only for binary fields.
11824
11825   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11826
11827 * Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
11828   through methods ('mul', 'precompute_mult').
11829
11830   The generic implementations (now internally called 'ec_wNAF_mul'
11831   and 'ec_wNAF_precomputed_mult') remain the default if these
11832   methods are undefined.
11833
11834   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11835
11836 * New function EC_GROUP_get_degree, which is defined through
11837   EC_METHOD.  For curves over prime fields, this returns the bit
11838   length of the modulus.
11839
11840   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11841
11842 * New functions EC_GROUP_dup, EC_POINT_dup.
11843   (These simply call ..._new  and ..._copy).
11844
11845   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11846
11847 * Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
11848   Polynomials are represented as BIGNUMs (where the sign bit is not
11849   used) in the following functions [macros]:
11850
11851           BN_GF2m_add
11852           BN_GF2m_sub             [= BN_GF2m_add]
11853           BN_GF2m_mod             [wrapper for BN_GF2m_mod_arr]
11854           BN_GF2m_mod_mul         [wrapper for BN_GF2m_mod_mul_arr]
11855           BN_GF2m_mod_sqr         [wrapper for BN_GF2m_mod_sqr_arr]
11856           BN_GF2m_mod_inv
11857           BN_GF2m_mod_exp         [wrapper for BN_GF2m_mod_exp_arr]
11858           BN_GF2m_mod_sqrt        [wrapper for BN_GF2m_mod_sqrt_arr]
11859           BN_GF2m_mod_solve_quad  [wrapper for BN_GF2m_mod_solve_quad_arr]
11860           BN_GF2m_cmp             [= BN_ucmp]
11861
11862   (Note that only the 'mod' functions are actually for fields GF(2^m).
11863   BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
11864
11865   For some functions, an the irreducible polynomial defining a
11866   field can be given as an 'unsigned int[]' with strictly
11867   decreasing elements giving the indices of those bits that are set;
11868   i.e., p[] represents the polynomial
11869           f(t) = t^p[0] + t^p[1] + ... + t^p[k]
11870   where
11871           p[0] > p[1] > ... > p[k] = 0.
11872   This applies to the following functions:
11873
11874           BN_GF2m_mod_arr
11875           BN_GF2m_mod_mul_arr
11876           BN_GF2m_mod_sqr_arr
11877           BN_GF2m_mod_inv_arr        [wrapper for BN_GF2m_mod_inv]
11878           BN_GF2m_mod_div_arr        [wrapper for BN_GF2m_mod_div]
11879           BN_GF2m_mod_exp_arr
11880           BN_GF2m_mod_sqrt_arr
11881           BN_GF2m_mod_solve_quad_arr
11882           BN_GF2m_poly2arr
11883           BN_GF2m_arr2poly
11884
11885   Conversion can be performed by the following functions:
11886
11887           BN_GF2m_poly2arr
11888           BN_GF2m_arr2poly
11889
11890   bntest.c has additional tests for binary polynomial arithmetic.
11891
11892   Two implementations for BN_GF2m_mod_div() are available.
11893   The default algorithm simply uses BN_GF2m_mod_inv() and
11894   BN_GF2m_mod_mul().  The alternative algorithm is compiled in only
11895   if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
11896   copyright notice in crypto/bn/bn_gf2m.c before enabling it).
11897
11898   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11899
11900 * Add new error code 'ERR_R_DISABLED' that can be used when some
11901   functionality is disabled at compile-time.
11902
11903   *Douglas Stebila <douglas.stebila@sun.com>*
11904
11905 * Change default behaviour of 'openssl asn1parse' so that more
11906   information is visible when viewing, e.g., a certificate:
11907
11908   Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
11909   mode the content of non-printable OCTET STRINGs is output in a
11910   style similar to INTEGERs, but with '[HEX DUMP]' prepended to
11911   avoid the appearance of a printable string.
11912
11913   *Nils Larsch <nla@trustcenter.de>*
11914
11915 * Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
11916   functions
11917           EC_GROUP_set_asn1_flag()
11918           EC_GROUP_get_asn1_flag()
11919           EC_GROUP_set_point_conversion_form()
11920           EC_GROUP_get_point_conversion_form()
11921   These control ASN1 encoding details:
11922   - Curves (i.e., groups) are encoded explicitly unless asn1_flag
11923     has been set to OPENSSL_EC_NAMED_CURVE.
11924   - Points are encoded in uncompressed form by default; options for
11925     asn1_for are as for point2oct, namely
11926           POINT_CONVERSION_COMPRESSED
11927           POINT_CONVERSION_UNCOMPRESSED
11928           POINT_CONVERSION_HYBRID
11929
11930   Also add 'seed' and 'seed_len' members to EC_GROUP with access
11931   functions
11932           EC_GROUP_set_seed()
11933           EC_GROUP_get0_seed()
11934           EC_GROUP_get_seed_len()
11935   This is used only for ASN1 purposes (so far).
11936
11937   *Nils Larsch <nla@trustcenter.de>*
11938
11939 * Add 'field_type' member to EC_METHOD, which holds the NID
11940   of the appropriate field type OID.  The new function
11941   EC_METHOD_get_field_type() returns this value.
11942
11943   *Nils Larsch <nla@trustcenter.de>*
11944
11945 * Add functions
11946           EC_POINT_point2bn()
11947           EC_POINT_bn2point()
11948           EC_POINT_point2hex()
11949           EC_POINT_hex2point()
11950   providing useful interfaces to EC_POINT_point2oct() and
11951   EC_POINT_oct2point().
11952
11953   *Nils Larsch <nla@trustcenter.de>*
11954
11955 * Change internals of the EC library so that the functions
11956           EC_GROUP_set_generator()
11957           EC_GROUP_get_generator()
11958           EC_GROUP_get_order()
11959           EC_GROUP_get_cofactor()
11960   are implemented directly in crypto/ec/ec_lib.c and not dispatched
11961   to methods, which would lead to unnecessary code duplication when
11962   adding different types of curves.
11963
11964   *Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller*
11965
11966 * Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
11967   arithmetic, and such that modified wNAFs are generated
11968   (which avoid length expansion in many cases).
11969
11970   *Bodo Moeller*
11971
11972 * Add a function EC_GROUP_check_discriminant() (defined via
11973   EC_METHOD) that verifies that the curve discriminant is non-zero.
11974
11975   Add a function EC_GROUP_check() that makes some sanity tests
11976   on a EC_GROUP, its generator and order.  This includes
11977   EC_GROUP_check_discriminant().
11978
11979   *Nils Larsch <nla@trustcenter.de>*
11980
11981 * Add ECDSA in new directory crypto/ecdsa/.
11982
11983   Add applications 'openssl ecparam' and 'openssl ecdsa'
11984   (these are based on 'openssl dsaparam' and 'openssl dsa').
11985
11986   ECDSA support is also included in various other files across the
11987   library.  Most notably,
11988   - 'openssl req' now has a '-newkey ecdsa:file' option;
11989   - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
11990   - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
11991     d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
11992     them suitable for ECDSA where domain parameters must be
11993     extracted before the specific public key;
11994   - ECDSA engine support has been added.
11995
11996   *Nils Larsch <nla@trustcenter.de>*
11997
11998 * Include some named elliptic curves, and add OIDs from X9.62,
11999   SECG, and WAP/WTLS.  Each curve can be obtained from the new
12000   function
12001           EC_GROUP_new_by_curve_name(),
12002   and the list of available named curves can be obtained with
12003           EC_get_builtin_curves().
12004   Also add a 'curve_name' member to EC_GROUP objects, which can be
12005   accessed via
12006           EC_GROUP_set_curve_name()
12007           EC_GROUP_get_curve_name()
12008
12009   *Nils Larsch <larsch@trustcenter.de, Bodo Moeller*
12010
12011 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
12012   was actually never needed) and in BN_mul().  The removal in BN_mul()
12013   required a small change in bn_mul_part_recursive() and the addition
12014   of the functions bn_cmp_part_words(), bn_sub_part_words() and
12015   bn_add_part_words(), which do the same thing as bn_cmp_words(),
12016   bn_sub_words() and bn_add_words() except they take arrays with
12017   differing sizes.
12018
12019   *Richard Levitte*
12020
12021### Changes between 0.9.7l and 0.9.7m  [23 Feb 2007]
12022
12023 * Cleanse PEM buffers before freeing them since they may contain
12024   sensitive data.
12025
12026   *Benjamin Bennett <ben@psc.edu>*
12027
12028 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
12029   a ciphersuite string such as "DEFAULT:RSA" cannot enable
12030   authentication-only ciphersuites.
12031
12032   *Bodo Moeller*
12033
12034 * Since AES128 and AES256 share a single mask bit in the logic of
12035   ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
12036   kludge to work properly if AES128 is available and AES256 isn't.
12037
12038   *Victor Duchovni*
12039
12040 * Expand security boundary to match 1.1.1 module.
12041
12042   *Steve Henson*
12043
12044 * Remove redundant features: hash file source, editing of test vectors
12045   modify fipsld to use external fips_premain.c signature.
12046
12047   *Steve Henson*
12048
12049 * New perl script mkfipsscr.pl to create shell scripts or batch files to
12050   run algorithm test programs.
12051
12052   *Steve Henson*
12053
12054 * Make algorithm test programs more tolerant of whitespace.
12055
12056   *Steve Henson*
12057
12058 * Have SSL/TLS server implementation tolerate "mismatched" record
12059   protocol version while receiving ClientHello even if the
12060   ClientHello is fragmented.  (The server can't insist on the
12061   particular protocol version it has chosen before the ServerHello
12062   message has informed the client about his choice.)
12063
12064   *Bodo Moeller*
12065
12066 * Load error codes if they are not already present instead of using a
12067   static variable. This allows them to be cleanly unloaded and reloaded.
12068
12069   *Steve Henson*
12070
12071### Changes between 0.9.7k and 0.9.7l  [28 Sep 2006]
12072
12073 * Introduce limits to prevent malicious keys being able to
12074   cause a denial of service.  ([CVE-2006-2940])
12075
12076   *Steve Henson, Bodo Moeller*
12077
12078 * Fix ASN.1 parsing of certain invalid structures that can result
12079   in a denial of service.  ([CVE-2006-2937])  [Steve Henson]
12080
12081 * Fix buffer overflow in SSL_get_shared_ciphers() function.
12082   ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team]
12083
12084 * Fix SSL client code which could crash if connecting to a
12085   malicious SSLv2 server.  ([CVE-2006-4343])
12086
12087   *Tavis Ormandy and Will Drewry, Google Security Team*
12088
12089 * Change ciphersuite string processing so that an explicit
12090   ciphersuite selects this one ciphersuite (so that "AES256-SHA"
12091   will no longer include "AES128-SHA"), and any other similar
12092   ciphersuite (same bitmap) from *other* protocol versions (so that
12093   "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
12094   SSL 3.0/TLS 1.0 ciphersuite).  This is a backport combining
12095   changes from 0.9.8b and 0.9.8d.
12096
12097   *Bodo Moeller*
12098
12099### Changes between 0.9.7j and 0.9.7k  [05 Sep 2006]
12100
12101 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
12102   ([CVE-2006-4339])  [Ben Laurie and Google Security Team]
12103
12104 * Change the Unix randomness entropy gathering to use poll() when
12105   possible instead of select(), since the latter has some
12106   undesirable limitations.
12107
12108   *Darryl Miles via Richard Levitte and Bodo Moeller*
12109
12110 * Disable rogue ciphersuites:
12111
12112   - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
12113   - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
12114   - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
12115
12116   The latter two were purportedly from
12117   draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
12118   appear there.
12119
12120   Also deactivate the remaining ciphersuites from
12121   draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
12122   unofficial, and the ID has long expired.
12123
12124   *Bodo Moeller*
12125
12126 * Fix RSA blinding Heisenbug (problems sometimes occurred on
12127   dual-core machines) and other potential thread-safety issues.
12128
12129   *Bodo Moeller*
12130
12131### Changes between 0.9.7i and 0.9.7j  [04 May 2006]
12132
12133 * Adapt fipsld and the build system to link against the validated FIPS
12134   module in FIPS mode.
12135
12136   *Steve Henson*
12137
12138 * Fixes for VC++ 2005 build under Windows.
12139
12140   *Steve Henson*
12141
12142 * Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make
12143   from a Windows bash shell such as MSYS. It is autodetected from the
12144   "config" script when run from a VC++ environment. Modify standard VC++
12145   build to use fipscanister.o from the GNU make build.
12146
12147   *Steve Henson*
12148
12149### Changes between 0.9.7h and 0.9.7i  [14 Oct 2005]
12150
12151 * Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
12152   The value now differs depending on if you build for FIPS or not.
12153   BEWARE!  A program linked with a shared FIPSed libcrypto can't be
12154   safely run with a non-FIPSed libcrypto, as it may crash because of
12155   the difference induced by this change.
12156
12157   *Andy Polyakov*
12158
12159### Changes between 0.9.7g and 0.9.7h  [11 Oct 2005]
12160
12161 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
12162   (part of SSL_OP_ALL).  This option used to disable the
12163   countermeasure against man-in-the-middle protocol-version
12164   rollback in the SSL 2.0 server implementation, which is a bad
12165   idea.  ([CVE-2005-2969])
12166
12167   *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
12168   for Information Security, National Institute of Advanced Industrial
12169   Science and Technology [AIST, Japan)]*
12170
12171 * Minimal support for X9.31 signatures and PSS padding modes. This is
12172   mainly for FIPS compliance and not fully integrated at this stage.
12173
12174   *Steve Henson*
12175
12176 * For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
12177   the exponentiation using a fixed-length exponent.  (Otherwise,
12178   the information leaked through timing could expose the secret key
12179   after many signatures; cf. Bleichenbacher's attack on DSA with
12180   biased k.)
12181
12182   *Bodo Moeller*
12183
12184 * Make a new fixed-window mod_exp implementation the default for
12185   RSA, DSA, and DH private-key operations so that the sequence of
12186   squares and multiplies and the memory access pattern are
12187   independent of the particular secret key.  This will mitigate
12188   cache-timing and potential related attacks.
12189
12190   BN_mod_exp_mont_consttime() is the new exponentiation implementation,
12191   and this is automatically used by BN_mod_exp_mont() if the new flag
12192   BN_FLG_EXP_CONSTTIME is set for the exponent.  RSA, DSA, and DH
12193   will use this BN flag for private exponents unless the flag
12194   RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
12195   DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
12196
12197   *Matthew D Wood (Intel Corp), with some changes by Bodo Moeller*
12198
12199 * Change the client implementation for SSLv23_method() and
12200   SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
12201   Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
12202   (Previously, the SSL 2.0 backwards compatible Client Hello
12203   message format would be used even with SSL_OP_NO_SSLv2.)
12204
12205   *Bodo Moeller*
12206
12207 * Add support for smime-type MIME parameter in S/MIME messages which some
12208   clients need.
12209
12210   *Steve Henson*
12211
12212 * New function BN_MONT_CTX_set_locked() to set montgomery parameters in
12213   a threadsafe manner. Modify rsa code to use new function and add calls
12214   to dsa and dh code (which had race conditions before).
12215
12216   *Steve Henson*
12217
12218 * Include the fixed error library code in the C error file definitions
12219   instead of fixing them up at runtime. This keeps the error code
12220   structures constant.
12221
12222   *Steve Henson*
12223
12224### Changes between 0.9.7f and 0.9.7g  [11 Apr 2005]
12225
12226[NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
12227OpenSSL 0.9.8.]
12228
12229 * Fixes for newer kerberos headers. NB: the casts are needed because
12230   the 'length' field is signed on one version and unsigned on another
12231   with no (?) obvious way to tell the difference, without these VC++
12232   complains. Also the "definition" of FAR (blank) is no longer included
12233   nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
12234   some needed definitions.
12235
12236   *Steve Henson*
12237
12238 * Undo Cygwin change.
12239
12240   *Ulf Möller*
12241
12242 * Added support for proxy certificates according to RFC 3820.
12243   Because they may be a security thread to unaware applications,
12244   they must be explicitly allowed in run-time.  See
12245   docs/HOWTO/proxy_certificates.txt for further information.
12246
12247   *Richard Levitte*
12248
12249### Changes between 0.9.7e and 0.9.7f  [22 Mar 2005]
12250
12251 * Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
12252   server and client random values. Previously
12253   (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
12254   less random data when sizeof(time_t) > 4 (some 64 bit platforms).
12255
12256   This change has negligible security impact because:
12257
12258   1. Server and client random values still have 24 bytes of pseudo random
12259      data.
12260
12261   2. Server and client random values are sent in the clear in the initial
12262      handshake.
12263
12264   3. The master secret is derived using the premaster secret (48 bytes in
12265      size for static RSA ciphersuites) as well as client server and random
12266      values.
12267
12268   The OpenSSL team would like to thank the UK NISCC for bringing this issue
12269   to our attention.
12270
12271   *Stephen Henson, reported by UK NISCC*
12272
12273 * Use Windows randomness collection on Cygwin.
12274
12275   *Ulf Möller*
12276
12277 * Fix hang in EGD/PRNGD query when communication socket is closed
12278   prematurely by EGD/PRNGD.
12279
12280   *Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014*
12281
12282 * Prompt for pass phrases when appropriate for PKCS12 input format.
12283
12284   *Steve Henson*
12285
12286 * Back-port of selected performance improvements from development
12287   branch, as well as improved support for PowerPC platforms.
12288
12289   *Andy Polyakov*
12290
12291 * Add lots of checks for memory allocation failure, error codes to indicate
12292   failure and freeing up memory if a failure occurs.
12293
12294   *Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson*
12295
12296 * Add new -passin argument to dgst.
12297
12298   *Steve Henson*
12299
12300 * Perform some character comparisons of different types in X509_NAME_cmp:
12301   this is needed for some certificates that re-encode DNs into UTF8Strings
12302   (in violation of RFC3280) and can't or won't issue name rollover
12303   certificates.
12304
12305   *Steve Henson*
12306
12307 * Make an explicit check during certificate validation to see that
12308   the CA setting in each certificate on the chain is correct.  As a
12309   side effect always do the following basic checks on extensions,
12310   not just when there's an associated purpose to the check:
12311
12312   - if there is an unhandled critical extension (unless the user
12313     has chosen to ignore this fault)
12314   - if the path length has been exceeded (if one is set at all)
12315   - that certain extensions fit the associated purpose (if one has
12316     been given)
12317
12318   *Richard Levitte*
12319
12320### Changes between 0.9.7d and 0.9.7e  [25 Oct 2004]
12321
12322 * Avoid a race condition when CRLs are checked in a multi threaded
12323   environment. This would happen due to the reordering of the revoked
12324   entries during signature checking and serial number lookup. Now the
12325   encoding is cached and the serial number sort performed under a lock.
12326   Add new STACK function sk_is_sorted().
12327
12328   *Steve Henson*
12329
12330 * Add Delta CRL to the extension code.
12331
12332   *Steve Henson*
12333
12334 * Various fixes to s3_pkt.c so alerts are sent properly.
12335
12336   *David Holmes <d.holmes@f5.com>*
12337
12338 * Reduce the chances of duplicate issuer name and serial numbers (in
12339   violation of RFC3280) using the OpenSSL certificate creation utilities.
12340   This is done by creating a random 64 bit value for the initial serial
12341   number when a serial number file is created or when a self signed
12342   certificate is created using 'openssl req -x509'. The initial serial
12343   number file is created using 'openssl x509 -next_serial' in CA.pl
12344   rather than being initialized to 1.
12345
12346   *Steve Henson*
12347
12348### Changes between 0.9.7c and 0.9.7d  [17 Mar 2004]
12349
12350 * Fix null-pointer assignment in do_change_cipher_spec() revealed
12351   by using the Codenomicon TLS Test Tool ([CVE-2004-0079])
12352
12353   *Joe Orton, Steve Henson*
12354
12355 * Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
12356   ([CVE-2004-0112])
12357
12358   *Joe Orton, Steve Henson*
12359
12360 * Make it possible to have multiple active certificates with the same
12361   subject in the CA index file.  This is done only if the keyword
12362   'unique_subject' is set to 'no' in the main CA section (default
12363   if 'CA_default') of the configuration file.  The value is saved
12364   with the database itself in a separate index attribute file,
12365   named like the index file with '.attr' appended to the name.
12366
12367   *Richard Levitte*
12368
12369 * X509 verify fixes. Disable broken certificate workarounds when
12370   X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
12371   keyUsage extension present. Don't accept CRLs with unhandled critical
12372   extensions: since verify currently doesn't process CRL extensions this
12373   rejects a CRL with *any* critical extensions. Add new verify error codes
12374   for these cases.
12375
12376   *Steve Henson*
12377
12378 * When creating an OCSP nonce use an OCTET STRING inside the extnValue.
12379   A clarification of RFC2560 will require the use of OCTET STRINGs and
12380   some implementations cannot handle the current raw format. Since OpenSSL
12381   copies and compares OCSP nonces as opaque blobs without any attempt at
12382   parsing them this should not create any compatibility issues.
12383
12384   *Steve Henson*
12385
12386 * New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
12387   calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
12388   this HMAC (and other) operations are several times slower than OpenSSL
12389   < 0.9.7.
12390
12391   *Steve Henson*
12392
12393 * Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
12394
12395   *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>*
12396
12397 * Use the correct content when signing type "other".
12398
12399   *Steve Henson*
12400
12401### Changes between 0.9.7b and 0.9.7c  [30 Sep 2003]
12402
12403 * Fix various bugs revealed by running the NISCC test suite:
12404
12405   Stop out of bounds reads in the ASN1 code when presented with
12406   invalid tags (CVE-2003-0543 and CVE-2003-0544).
12407
12408   Free up ASN1_TYPE correctly if ANY type is invalid ([CVE-2003-0545]).
12409
12410   If verify callback ignores invalid public key errors don't try to check
12411   certificate signature with the NULL public key.
12412
12413   *Steve Henson*
12414
12415 * New -ignore_err option in ocsp application to stop the server
12416   exiting on the first error in a request.
12417
12418   *Steve Henson*
12419
12420 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
12421   if the server requested one: as stated in TLS 1.0 and SSL 3.0
12422   specifications.
12423
12424   *Steve Henson*
12425
12426 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
12427   extra data after the compression methods not only for TLS 1.0
12428   but also for SSL 3.0 (as required by the specification).
12429
12430   *Bodo Moeller; problem pointed out by Matthias Loepfe*
12431
12432 * Change X509_certificate_type() to mark the key as exported/exportable
12433   when it's 512 *bits* long, not 512 bytes.
12434
12435   *Richard Levitte*
12436
12437 * Change AES_cbc_encrypt() so it outputs exact multiple of
12438   blocks during encryption.
12439
12440   *Richard Levitte*
12441
12442 * Various fixes to base64 BIO and non blocking I/O. On write
12443   flushes were not handled properly if the BIO retried. On read
12444   data was not being buffered properly and had various logic bugs.
12445   This also affects blocking I/O when the data being decoded is a
12446   certain size.
12447
12448   *Steve Henson*
12449
12450 * Various S/MIME bugfixes and compatibility changes:
12451   output correct application/pkcs7 MIME type if
12452   PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
12453   Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
12454   of files as .eml work). Correctly handle very long lines in MIME
12455   parser.
12456
12457   *Steve Henson*
12458
12459### Changes between 0.9.7a and 0.9.7b  [10 Apr 2003]
12460
12461 * Countermeasure against the Klima-Pokorny-Rosa extension of
12462   Bleichbacher's attack on PKCS #1 v1.5 padding: treat
12463   a protocol version number mismatch like a decryption error
12464   in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
12465
12466   *Bodo Moeller*
12467
12468 * Turn on RSA blinding by default in the default implementation
12469   to avoid a timing attack. Applications that don't want it can call
12470   RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
12471   They would be ill-advised to do so in most cases.
12472
12473   *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller*
12474
12475 * Change RSA blinding code so that it works when the PRNG is not
12476   seeded (in this case, the secret RSA exponent is abused as
12477   an unpredictable seed -- if it is not unpredictable, there
12478   is no point in blinding anyway).  Make RSA blinding thread-safe
12479   by remembering the creator's thread ID in rsa->blinding and
12480   having all other threads use local one-time blinding factors
12481   (this requires more computation than sharing rsa->blinding, but
12482   avoids excessive locking; and if an RSA object is not shared
12483   between threads, blinding will still be very fast).
12484
12485   *Bodo Moeller*
12486
12487 * Fixed a typo bug that would cause ENGINE_set_default() to set an
12488   ENGINE as defaults for all supported algorithms irrespective of
12489   the 'flags' parameter. 'flags' is now honoured, so applications
12490   should make sure they are passing it correctly.
12491
12492   *Geoff Thorpe*
12493
12494 * Target "mingw" now allows native Windows code to be generated in
12495   the Cygwin environment as well as with the MinGW compiler.
12496
12497   *Ulf Moeller*
12498
12499### Changes between 0.9.7 and 0.9.7a  [19 Feb 2003]
12500
12501 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
12502   via timing by performing a MAC computation even if incorrect
12503   block cipher padding has been found.  This is a countermeasure
12504   against active attacks where the attacker has to distinguish
12505   between bad padding and a MAC verification error. ([CVE-2003-0078])
12506
12507   *Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
12508   Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
12509   Martin Vuagnoux (EPFL, Ilion)*
12510
12511 * Make the no-err option work as intended.  The intention with no-err
12512   is not to have the whole error stack handling routines removed from
12513   libcrypto, it's only intended to remove all the function name and
12514   reason texts, thereby removing some of the footprint that may not
12515   be interesting if those errors aren't displayed anyway.
12516
12517   NOTE: it's still possible for any application or module to have its
12518   own set of error texts inserted.  The routines are there, just not
12519   used by default when no-err is given.
12520
12521   *Richard Levitte*
12522
12523 * Add support for FreeBSD on IA64.
12524
12525   *dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454*
12526
12527 * Adjust DES_cbc_cksum() so it returns the same value as the MIT
12528   Kerberos function mit_des_cbc_cksum().  Before this change,
12529   the value returned by DES_cbc_cksum() was like the one from
12530   mit_des_cbc_cksum(), except the bytes were swapped.
12531
12532   *Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte*
12533
12534 * Allow an application to disable the automatic SSL chain building.
12535   Before this a rather primitive chain build was always performed in
12536   ssl3_output_cert_chain(): an application had no way to send the
12537   correct chain if the automatic operation produced an incorrect result.
12538
12539   Now the chain builder is disabled if either:
12540
12541   1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
12542
12543   2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
12544
12545   The reasoning behind this is that an application would not want the
12546   auto chain building to take place if extra chain certificates are
12547   present and it might also want a means of sending no additional
12548   certificates (for example the chain has two certificates and the
12549   root is omitted).
12550
12551   *Steve Henson*
12552
12553 * Add the possibility to build without the ENGINE framework.
12554
12555   *Steven Reddie <smr@essemer.com.au> via Richard Levitte*
12556
12557 * Under Win32 gmtime() can return NULL: check return value in
12558   OPENSSL_gmtime(). Add error code for case where gmtime() fails.
12559
12560   *Steve Henson*
12561
12562 * DSA routines: under certain error conditions uninitialized BN objects
12563   could be freed. Solution: make sure initialization is performed early
12564   enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
12565   Nils Larsch <nla@trustcenter.de> via PR#459)
12566
12567   *Lutz Jaenicke*
12568
12569 * Another fix for SSLv2 session ID handling: the session ID was incorrectly
12570   checked on reconnect on the client side, therefore session resumption
12571   could still fail with a "ssl session id is different" error. This
12572   behaviour is masked when SSL_OP_ALL is used due to
12573   SSL_OP_MICROSOFT_SESS_ID_BUG being set.
12574   Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
12575   followup to PR #377.
12576
12577   *Lutz Jaenicke*
12578
12579 * IA-32 assembler support enhancements: unified ELF targets, support
12580   for SCO/Caldera platforms, fix for Cygwin shared build.
12581
12582   *Andy Polyakov*
12583
12584 * Add support for FreeBSD on sparc64.  As a consequence, support for
12585   FreeBSD on non-x86 processors is separate from x86 processors on
12586   the config script, much like the NetBSD support.
12587
12588   *Richard Levitte & Kris Kennaway <kris@obsecurity.org>*
12589
12590### Changes between 0.9.6h and 0.9.7  [31 Dec 2002]
12591
12592[NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
12593OpenSSL 0.9.7.]
12594
12595 * Fix session ID handling in SSLv2 client code: the SERVER FINISHED
12596   code (06) was taken as the first octet of the session ID and the last
12597   octet was ignored consequently. As a result SSLv2 client side session
12598   caching could not have worked due to the session ID mismatch between
12599   client and server.
12600   Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
12601   PR #377.
12602
12603   *Lutz Jaenicke*
12604
12605 * Change the declaration of needed Kerberos libraries to use EX_LIBS
12606   instead of the special (and badly supported) LIBKRB5.  LIBKRB5 is
12607   removed entirely.
12608
12609   *Richard Levitte*
12610
12611 * The hw_ncipher.c engine requires dynamic locks.  Unfortunately, it
12612   seems that in spite of existing for more than a year, many application
12613   author have done nothing to provide the necessary callbacks, which
12614   means that this particular engine will not work properly anywhere.
12615   This is a very unfortunate situation which forces us, in the name
12616   of usability, to give the hw_ncipher.c a static lock, which is part
12617   of libcrypto.
12618   NOTE: This is for the 0.9.7 series ONLY.  This hack will never
12619   appear in 0.9.8 or later.  We EXPECT application authors to have
12620   dealt properly with this when 0.9.8 is released (unless we actually
12621   make such changes in the libcrypto locking code that changes will
12622   have to be made anyway).
12623
12624   *Richard Levitte*
12625
12626 * In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
12627   octets have been read, EOF or an error occurs. Without this change
12628   some truncated ASN1 structures will not produce an error.
12629
12630   *Steve Henson*
12631
12632 * Disable Heimdal support, since it hasn't been fully implemented.
12633   Still give the possibility to force the use of Heimdal, but with
12634   warnings and a request that patches get sent to openssl-dev.
12635
12636   *Richard Levitte*
12637
12638 * Add the VC-CE target, introduce the WINCE sysname, and add
12639   INSTALL.WCE and appropriate conditionals to make it build.
12640
12641   *Steven Reddie <smr@essemer.com.au> via Richard Levitte*
12642
12643 * Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
12644   cygssl-x.y.z.dll, where x, y and z are the major, minor and
12645   edit numbers of the version.
12646
12647   *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte*
12648
12649 * Introduce safe string copy and catenation functions
12650   (BUF_strlcpy() and BUF_strlcat()).
12651
12652   *Ben Laurie (CHATS) and Richard Levitte*
12653
12654 * Avoid using fixed-size buffers for one-line DNs.
12655
12656   *Ben Laurie (CHATS)*
12657
12658 * Add BUF_MEM_grow_clean() to avoid information leakage when
12659   resizing buffers containing secrets, and use where appropriate.
12660
12661   *Ben Laurie (CHATS)*
12662
12663 * Avoid using fixed size buffers for configuration file location.
12664
12665   *Ben Laurie (CHATS)*
12666
12667 * Avoid filename truncation for various CA files.
12668
12669   *Ben Laurie (CHATS)*
12670
12671 * Use sizeof in preference to magic numbers.
12672
12673   *Ben Laurie (CHATS)*
12674
12675 * Avoid filename truncation in cert requests.
12676
12677   *Ben Laurie (CHATS)*
12678
12679 * Add assertions to check for (supposedly impossible) buffer
12680   overflows.
12681
12682   *Ben Laurie (CHATS)*
12683
12684 * Don't cache truncated DNS entries in the local cache (this could
12685   potentially lead to a spoofing attack).
12686
12687   *Ben Laurie (CHATS)*
12688
12689 * Fix various buffers to be large enough for hex/decimal
12690   representations in a platform independent manner.
12691
12692   *Ben Laurie (CHATS)*
12693
12694 * Add CRYPTO_realloc_clean() to avoid information leakage when
12695   resizing buffers containing secrets, and use where appropriate.
12696
12697   *Ben Laurie (CHATS)*
12698
12699 * Add BIO_indent() to avoid much slightly worrying code to do
12700   indents.
12701
12702   *Ben Laurie (CHATS)*
12703
12704 * Convert sprintf()/BIO_puts() to BIO_printf().
12705
12706   *Ben Laurie (CHATS)*
12707
12708 * buffer_gets() could terminate with the buffer only half
12709   full. Fixed.
12710
12711   *Ben Laurie (CHATS)*
12712
12713 * Add assertions to prevent user-supplied crypto functions from
12714   overflowing internal buffers by having large block sizes, etc.
12715
12716   *Ben Laurie (CHATS)*
12717
12718 * New OPENSSL_assert() macro (similar to assert(), but enabled
12719   unconditionally).
12720
12721   *Ben Laurie (CHATS)*
12722
12723 * Eliminate unused copy of key in RC4.
12724
12725   *Ben Laurie (CHATS)*
12726
12727 * Eliminate unused and incorrectly sized buffers for IV in pem.h.
12728
12729   *Ben Laurie (CHATS)*
12730
12731 * Fix off-by-one error in EGD path.
12732
12733   *Ben Laurie (CHATS)*
12734
12735 * If RANDFILE path is too long, ignore instead of truncating.
12736
12737   *Ben Laurie (CHATS)*
12738
12739 * Eliminate unused and incorrectly sized X.509 structure
12740   CBCParameter.
12741
12742   *Ben Laurie (CHATS)*
12743
12744 * Eliminate unused and dangerous function knumber().
12745
12746   *Ben Laurie (CHATS)*
12747
12748 * Eliminate unused and dangerous structure, KSSL_ERR.
12749
12750   *Ben Laurie (CHATS)*
12751
12752 * Protect against overlong session ID context length in an encoded
12753   session object. Since these are local, this does not appear to be
12754   exploitable.
12755
12756   *Ben Laurie (CHATS)*
12757
12758 * Change from security patch (see 0.9.6e below) that did not affect
12759   the 0.9.6 release series:
12760
12761   Remote buffer overflow in SSL3 protocol - an attacker could
12762   supply an oversized master key in Kerberos-enabled versions.
12763   ([CVE-2002-0657])
12764
12765   *Ben Laurie (CHATS)*
12766
12767 * Change the SSL kerb5 codes to match RFC 2712.
12768
12769   *Richard Levitte*
12770
12771 * Make -nameopt work fully for req and add -reqopt switch.
12772
12773   *Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson*
12774
12775 * The "block size" for block ciphers in CFB and OFB mode should be 1.
12776
12777   *Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>*
12778
12779 * Make sure tests can be performed even if the corresponding algorithms
12780   have been removed entirely.  This was also the last step to make
12781   OpenSSL compilable with DJGPP under all reasonable conditions.
12782
12783   *Richard Levitte, Doug Kaufman <dkaufman@rahul.net>*
12784
12785 * Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
12786   to allow version independent disabling of normally unselected ciphers,
12787   which may be activated as a side-effect of selecting a single cipher.
12788
12789   (E.g., cipher list string "RSA" enables ciphersuites that are left
12790   out of "ALL" because they do not provide symmetric encryption.
12791   "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
12792
12793   *Lutz Jaenicke, Bodo Moeller*
12794
12795 * Add appropriate support for separate platform-dependent build
12796   directories.  The recommended way to make a platform-dependent
12797   build directory is the following (tested on Linux), maybe with
12798   some local tweaks:
12799
12800           # Place yourself outside of the OpenSSL source tree.  In
12801           # this example, the environment variable OPENSSL_SOURCE
12802           # is assumed to contain the absolute OpenSSL source directory.
12803           mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
12804           cd objtree/"`uname -s`-`uname -r`-`uname -m`"
12805           (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
12806                   mkdir -p `dirname $F`
12807                   ln -s $OPENSSL_SOURCE/$F $F
12808           done
12809
12810   To be absolutely sure not to disturb the source tree, a "make clean"
12811   is a good thing.  If it isn't successful, don't worry about it,
12812   it probably means the source directory is very clean.
12813
12814   *Richard Levitte*
12815
12816 * Make sure any ENGINE control commands make local copies of string
12817   pointers passed to them whenever necessary. Otherwise it is possible
12818   the caller may have overwritten (or deallocated) the original string
12819   data when a later ENGINE operation tries to use the stored values.
12820
12821   *Götz Babin-Ebell <babinebell@trustcenter.de>*
12822
12823 * Improve diagnostics in file reading and command-line digests.
12824
12825   *Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>*
12826
12827 * Add AES modes CFB and OFB to the object database.  Correct an
12828   error in AES-CFB decryption.
12829
12830   *Richard Levitte*
12831
12832 * Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
12833   allows existing EVP_CIPHER_CTX structures to be reused after
12834   calling `EVP_*Final()`. This behaviour is used by encryption
12835   BIOs and some applications. This has the side effect that
12836   applications must explicitly clean up cipher contexts with
12837   EVP_CIPHER_CTX_cleanup() or they will leak memory.
12838
12839   *Steve Henson*
12840
12841 * Check the values of dna and dnb in bn_mul_recursive before calling
12842   bn_mul_comba (a non zero value means the a or b arrays do not contain
12843   n2 elements) and fallback to bn_mul_normal if either is not zero.
12844
12845   *Steve Henson*
12846
12847 * Fix escaping of non-ASCII characters when using the -subj option
12848   of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
12849
12850   *Lutz Jaenicke*
12851
12852 * Make object definitions compliant to LDAP (RFC2256): SN is the short
12853   form for "surname", serialNumber has no short form.
12854   Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
12855   therefore remove "mail" short name for "internet 7".
12856   The OID for unique identifiers in X509 certificates is
12857   x500UniqueIdentifier, not uniqueIdentifier.
12858   Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
12859
12860   *Lutz Jaenicke*
12861
12862 * Add an "init" command to the ENGINE config module and auto initialize
12863   ENGINEs. Without any "init" command the ENGINE will be initialized
12864   after all ctrl commands have been executed on it. If init=1 the
12865   ENGINE is initialized at that point (ctrls before that point are run
12866   on the uninitialized ENGINE and after on the initialized one). If
12867   init=0 then the ENGINE will not be initialized at all.
12868
12869   *Steve Henson*
12870
12871 * Fix the 'app_verify_callback' interface so that the user-defined
12872   argument is actually passed to the callback: In the
12873   SSL_CTX_set_cert_verify_callback() prototype, the callback
12874   declaration has been changed from
12875           int (*cb)()
12876   into
12877           int (*cb)(X509_STORE_CTX *,void *);
12878   in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
12879           i=s->ctx->app_verify_callback(&ctx)
12880   has been changed into
12881           i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
12882
12883   To update applications using SSL_CTX_set_cert_verify_callback(),
12884   a dummy argument can be added to their callback functions.
12885
12886   *D. K. Smetters <smetters@parc.xerox.com>*
12887
12888 * Added the '4758cca' ENGINE to support IBM 4758 cards.
12889
12890   *Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe*
12891
12892 * Add and OPENSSL_LOAD_CONF define which will cause
12893   OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
12894   This allows older applications to transparently support certain
12895   OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
12896   Two new functions OPENSSL_add_all_algorithms_noconf() which will never
12897   load the config file and OPENSSL_add_all_algorithms_conf() which will
12898   always load it have also been added.
12899
12900   *Steve Henson*
12901
12902 * Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
12903   Adjust NIDs and EVP layer.
12904
12905   *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte*
12906
12907 * Config modules support in openssl utility.
12908
12909   Most commands now load modules from the config file,
12910   though in a few (such as version) this isn't done
12911   because it couldn't be used for anything.
12912
12913   In the case of ca and req the config file used is
12914   the same as the utility itself: that is the -config
12915   command line option can be used to specify an
12916   alternative file.
12917
12918   *Steve Henson*
12919
12920 * Move default behaviour from OPENSSL_config(). If appname is NULL
12921   use "openssl_conf" if filename is NULL use default openssl config file.
12922
12923   *Steve Henson*
12924
12925 * Add an argument to OPENSSL_config() to allow the use of an alternative
12926   config section name. Add a new flag to tolerate a missing config file
12927   and move code to CONF_modules_load_file().
12928
12929   *Steve Henson*
12930
12931 * Support for crypto accelerator cards from Accelerated Encryption
12932   Processing, www.aep.ie.  (Use engine 'aep')
12933   The support was copied from 0.9.6c [engine] and adapted/corrected
12934   to work with the new engine framework.
12935
12936   *AEP Inc. and Richard Levitte*
12937
12938 * Support for SureWare crypto accelerator cards from Baltimore
12939   Technologies.  (Use engine 'sureware')
12940   The support was copied from 0.9.6c [engine] and adapted
12941   to work with the new engine framework.
12942
12943   *Richard Levitte*
12944
12945 * Have the CHIL engine fork-safe (as defined by nCipher) and actually
12946   make the newer ENGINE framework commands for the CHIL engine work.
12947
12948   *Toomas Kiisk <vix@cyber.ee> and Richard Levitte*
12949
12950 * Make it possible to produce shared libraries on ReliantUNIX.
12951
12952   *Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte*
12953
12954 * Add the configuration target debug-linux-ppro.
12955   Make 'openssl rsa' use the general key loading routines
12956   implemented in `apps.c`, and make those routines able to
12957   handle the key format FORMAT_NETSCAPE and the variant
12958   FORMAT_IISSGC.
12959
12960   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
12961
12962 * Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
12963
12964   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
12965
12966 * Add -keyform to rsautl, and document -engine.
12967
12968   *Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>*
12969
12970 * Change BIO_new_file (crypto/bio/bss_file.c) to use new
12971   BIO_R_NO_SUCH_FILE error code rather than the generic
12972   ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
12973
12974   *Ben Laurie*
12975
12976 * Add new functions
12977           ERR_peek_last_error
12978           ERR_peek_last_error_line
12979           ERR_peek_last_error_line_data.
12980   These are similar to
12981           ERR_peek_error
12982           ERR_peek_error_line
12983           ERR_peek_error_line_data,
12984   but report on the latest error recorded rather than the first one
12985   still in the error queue.
12986
12987   *Ben Laurie, Bodo Moeller*
12988
12989 * default_algorithms option in ENGINE config module. This allows things
12990   like:
12991   default_algorithms = ALL
12992   default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
12993
12994   *Steve Henson*
12995
12996 * Preliminary ENGINE config module.
12997
12998   *Steve Henson*
12999
13000 * New experimental application configuration code.
13001
13002   *Steve Henson*
13003
13004 * Change the AES code to follow the same name structure as all other
13005   symmetric ciphers, and behave the same way.  Move everything to
13006   the directory crypto/aes, thereby obsoleting crypto/rijndael.
13007
13008   *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte*
13009
13010 * SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
13011
13012   *Ben Laurie and Theo de Raadt*
13013
13014 * Add option to output public keys in req command.
13015
13016   *Massimiliano Pala madwolf@openca.org*
13017
13018 * Use wNAFs in EC_POINTs_mul() for improved efficiency
13019   (up to about 10% better than before for P-192 and P-224).
13020
13021   *Bodo Moeller*
13022
13023 * New functions/macros
13024
13025           SSL_CTX_set_msg_callback(ctx, cb)
13026           SSL_CTX_set_msg_callback_arg(ctx, arg)
13027           SSL_set_msg_callback(ssl, cb)
13028           SSL_set_msg_callback_arg(ssl, arg)
13029
13030   to request calling a callback function
13031
13032           void cb(int write_p, int version, int content_type,
13033                   const void *buf, size_t len, SSL *ssl, void *arg)
13034
13035   whenever a protocol message has been completely received
13036   (write_p == 0) or sent (write_p == 1).  Here 'version' is the
13037   protocol version  according to which the SSL library interprets
13038   the current protocol message (SSL2_VERSION, SSL3_VERSION, or
13039   TLS1_VERSION).  'content_type' is 0 in the case of SSL 2.0, or
13040   the content type as defined in the SSL 3.0/TLS 1.0 protocol
13041   specification (change_cipher_spec(20), alert(21), handshake(22)).
13042   'buf' and 'len' point to the actual message, 'ssl' to the
13043   SSL object, and 'arg' is the application-defined value set by
13044   SSL[_CTX]_set_msg_callback_arg().
13045
13046   'openssl s_client' and 'openssl s_server' have new '-msg' options
13047   to enable a callback that displays all protocol messages.
13048
13049   *Bodo Moeller*
13050
13051 * Change the shared library support so shared libraries are built as
13052   soon as the corresponding static library is finished, and thereby get
13053   openssl and the test programs linked against the shared library.
13054   This still only happens when the keyword "shard" has been given to
13055   the configuration scripts.
13056
13057   NOTE: shared library support is still an experimental thing, and
13058   backward binary compatibility is still not guaranteed.
13059
13060   *"Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte*
13061
13062 * Add support for Subject Information Access extension.
13063
13064   *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>*
13065
13066 * Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
13067   additional bytes when new memory had to be allocated, not just
13068   when reusing an existing buffer.
13069
13070   *Bodo Moeller*
13071
13072 * New command line and configuration option 'utf8' for the req command.
13073   This allows field values to be specified as UTF8 strings.
13074
13075   *Steve Henson*
13076
13077 * Add -multi and -mr options to "openssl speed" - giving multiple parallel
13078   runs for the former and machine-readable output for the latter.
13079
13080   *Ben Laurie*
13081
13082 * Add '-noemailDN' option to 'openssl ca'.  This prevents inclusion
13083   of the e-mail address in the DN (i.e., it will go into a certificate
13084   extension only).  The new configuration file option 'email_in_dn = no'
13085   has the same effect.
13086
13087   *Massimiliano Pala madwolf@openca.org*
13088
13089 * Change all functions with names starting with `des_` to be starting
13090   with `DES_` instead.  Add wrappers that are compatible with libdes,
13091   but are named `_ossl_old_des_*`.  Finally, add macros that map the
13092   `des_*` symbols to the corresponding `_ossl_old_des_*` if libdes
13093   compatibility is desired.  If OpenSSL 0.9.6c compatibility is
13094   desired, the `des_*` symbols will be mapped to `DES_*`, with one
13095   exception.
13096
13097   Since we provide two compatibility mappings, the user needs to
13098   define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
13099   compatibility is desired.  The default (i.e., when that macro
13100   isn't defined) is OpenSSL 0.9.6c compatibility.
13101
13102   There are also macros that enable and disable the support of old
13103   des functions altogether.  Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
13104   and OPENSSL_DISABLE_OLD_DES_SUPPORT.  If none or both of those
13105   are defined, the default will apply: to support the old des routines.
13106
13107   In either case, one must include openssl/des.h to get the correct
13108   definitions.  Do not try to just include openssl/des_old.h, that
13109   won't work.
13110
13111   NOTE: This is a major break of an old API into a new one.  Software
13112   authors are encouraged to switch to the `DES_` style functions.  Some
13113   time in the future, des_old.h and the libdes compatibility functions
13114   will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
13115   default), and then completely removed.
13116
13117   *Richard Levitte*
13118
13119 * Test for certificates which contain unsupported critical extensions.
13120   If such a certificate is found during a verify operation it is
13121   rejected by default: this behaviour can be overridden by either
13122   handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
13123   by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
13124   X509_supported_extension() has also been added which returns 1 if a
13125   particular extension is supported.
13126
13127   *Steve Henson*
13128
13129 * Modify the behaviour of EVP cipher functions in similar way to digests
13130   to retain compatibility with existing code.
13131
13132   *Steve Henson*
13133
13134 * Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
13135   compatibility with existing code. In particular the 'ctx' parameter does
13136   not have to be to be initialized before the call to EVP_DigestInit() and
13137   it is tidied up after a call to EVP_DigestFinal(). New function
13138   EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
13139   EVP_MD_CTX_copy() changed to not require the destination to be
13140   initialized valid and new function EVP_MD_CTX_copy_ex() added which
13141   requires the destination to be valid.
13142
13143   Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
13144   EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
13145
13146   *Steve Henson*
13147
13148 * Change ssl3_get_message (ssl/s3_both.c) and the functions using it
13149   so that complete 'Handshake' protocol structures are kept in memory
13150   instead of overwriting 'msg_type' and 'length' with 'body' data.
13151
13152   *Bodo Moeller*
13153
13154 * Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
13155
13156   *Massimo Santin via Richard Levitte*
13157
13158 * Major restructuring to the underlying ENGINE code. This includes
13159   reduction of linker bloat, separation of pure "ENGINE" manipulation
13160   (initialisation, etc) from functionality dealing with implementations
13161   of specific crypto interfaces. This change also introduces integrated
13162   support for symmetric ciphers and digest implementations - so ENGINEs
13163   can now accelerate these by providing EVP_CIPHER and EVP_MD
13164   implementations of their own. This is detailed in
13165   [crypto/engine/README.md](crypto/engine/README.md)
13166   as it couldn't be adequately described here. However, there are a few
13167   API changes worth noting - some RSA, DSA, DH, and RAND functions that
13168   were changed in the original introduction of ENGINE code have now
13169   reverted back - the hooking from this code to ENGINE is now a good
13170   deal more passive and at run-time, operations deal directly with
13171   RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
13172   dereferencing through an ENGINE pointer any more. Also, the ENGINE
13173   functions dealing with `BN_MOD_EXP[_CRT]` handlers have been removed -
13174   they were not being used by the framework as there is no concept of a
13175   BIGNUM_METHOD and they could not be generalised to the new
13176   'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
13177   ENGINE_cpy() has been removed as it cannot be consistently defined in
13178   the new code.
13179
13180   *Geoff Thorpe*
13181
13182 * Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
13183
13184   *Steve Henson*
13185
13186 * Change mkdef.pl to sort symbols that get the same entry number,
13187   and make sure the automatically generated functions `ERR_load_*`
13188   become part of libeay.num as well.
13189
13190   *Richard Levitte*
13191
13192 * New function SSL_renegotiate_pending().  This returns true once
13193   renegotiation has been requested (either SSL_renegotiate() call
13194   or HelloRequest/ClientHello received from the peer) and becomes
13195   false once a handshake has been completed.
13196   (For servers, SSL_renegotiate() followed by SSL_do_handshake()
13197   sends a HelloRequest, but does not ensure that a handshake takes
13198   place.  SSL_renegotiate_pending() is useful for checking if the
13199   client has followed the request.)
13200
13201   *Bodo Moeller*
13202
13203 * New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
13204   By default, clients may request session resumption even during
13205   renegotiation (if session ID contexts permit); with this option,
13206   session resumption is possible only in the first handshake.
13207
13208   SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL.  This makes
13209   more bits available for options that should not be part of
13210   SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
13211
13212   *Bodo Moeller*
13213
13214 * Add some demos for certificate and certificate request creation.
13215
13216   *Steve Henson*
13217
13218 * Make maximum certificate chain size accepted from the peer application
13219   settable (`SSL*_get/set_max_cert_list()`), as proposed by
13220   "Douglas E. Engert" <deengert@anl.gov>.
13221
13222   *Lutz Jaenicke*
13223
13224 * Add support for shared libraries for Unixware-7
13225   (Boyd Lynn Gerber <gerberb@zenez.com>).
13226
13227   *Lutz Jaenicke*
13228
13229 * Add a "destroy" handler to ENGINEs that allows structural cleanup to
13230   be done prior to destruction. Use this to unload error strings from
13231   ENGINEs that load their own error strings. NB: This adds two new API
13232   functions to "get" and "set" this destroy handler in an ENGINE.
13233
13234   *Geoff Thorpe*
13235
13236 * Alter all existing ENGINE implementations (except "openssl" and
13237   "openbsd") to dynamically instantiate their own error strings. This
13238   makes them more flexible to be built both as statically-linked ENGINEs
13239   and self-contained shared-libraries loadable via the "dynamic" ENGINE.
13240   Also, add stub code to each that makes building them as self-contained
13241   shared-libraries easier (see [README-Engine.md](README-Engine.md)).
13242
13243   *Geoff Thorpe*
13244
13245 * Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
13246   implementations into applications that are completely implemented in
13247   self-contained shared-libraries. The "dynamic" ENGINE exposes control
13248   commands that can be used to configure what shared-library to load and
13249   to control aspects of the way it is handled. Also, made an update to
13250   the [README-Engine.md](README-Engine.md) file
13251   that brings its information up-to-date and
13252   provides some information and instructions on the "dynamic" ENGINE
13253   (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
13254
13255   *Geoff Thorpe*
13256
13257 * Make it possible to unload ranges of ERR strings with a new
13258   "ERR_unload_strings" function.
13259
13260   *Geoff Thorpe*
13261
13262 * Add a copy() function to EVP_MD.
13263
13264   *Ben Laurie*
13265
13266 * Make EVP_MD routines take a context pointer instead of just the
13267   md_data void pointer.
13268
13269   *Ben Laurie*
13270
13271 * Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
13272   that the digest can only process a single chunk of data
13273   (typically because it is provided by a piece of
13274   hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
13275   is only going to provide a single chunk of data, and hence the
13276   framework needn't accumulate the data for oneshot drivers.
13277
13278   *Ben Laurie*
13279
13280 * As with "ERR", make it possible to replace the underlying "ex_data"
13281   functions. This change also alters the storage and management of global
13282   ex_data state - it's now all inside ex_data.c and all "class" code (eg.
13283   RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
13284   index counters. The API functions that use this state have been changed
13285   to take a "class_index" rather than pointers to the class's local STACK
13286   and counter, and there is now an API function to dynamically create new
13287   classes. This centralisation allows us to (a) plug a lot of the
13288   thread-safety problems that existed, and (b) makes it possible to clean
13289   up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
13290   such data would previously have always leaked in application code and
13291   workarounds were in place to make the memory debugging turn a blind eye
13292   to it. Application code that doesn't use this new function will still
13293   leak as before, but their memory debugging output will announce it now
13294   rather than letting it slide.
13295
13296   Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
13297   induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
13298   has a return value to indicate success or failure.
13299
13300   *Geoff Thorpe*
13301
13302 * Make it possible to replace the underlying "ERR" functions such that the
13303   global state (2 LHASH tables and 2 locks) is only used by the "default"
13304   implementation. This change also adds two functions to "get" and "set"
13305   the implementation prior to it being automatically set the first time
13306   any other ERR function takes place. Ie. an application can call "get",
13307   pass the return value to a module it has just loaded, and that module
13308   can call its own "set" function using that value. This means the
13309   module's "ERR" operations will use (and modify) the error state in the
13310   application and not in its own statically linked copy of OpenSSL code.
13311
13312   *Geoff Thorpe*
13313
13314 * Give DH, DSA, and RSA types their own `*_up_ref()` function to increment
13315   reference counts. This performs normal REF_PRINT/REF_CHECK macros on
13316   the operation, and provides a more encapsulated way for external code
13317   (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
13318   to use these functions rather than manually incrementing the counts.
13319
13320   Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
13321
13322   *Geoff Thorpe*
13323
13324 * Add EVP test program.
13325
13326   *Ben Laurie*
13327
13328 * Add symmetric cipher support to ENGINE. Expect the API to change!
13329
13330   *Ben Laurie*
13331
13332 * New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
13333   X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
13334   X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
13335   These allow a CRL to be built without having to access X509_CRL fields
13336   directly. Modify 'ca' application to use new functions.
13337
13338   *Steve Henson*
13339
13340 * Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
13341   bug workarounds. Rollback attack detection is a security feature.
13342   The problem will only arise on OpenSSL servers when TLSv1 is not
13343   available (sslv3_server_method() or SSL_OP_NO_TLSv1).
13344   Software authors not wanting to support TLSv1 will have special reasons
13345   for their choice and can explicitly enable this option.
13346
13347   *Bodo Moeller, Lutz Jaenicke*
13348
13349 * Rationalise EVP so it can be extended: don't include a union of
13350   cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
13351   (similar to those existing for EVP_CIPHER_CTX).
13352   Usage example:
13353
13354           EVP_MD_CTX md;
13355
13356           EVP_MD_CTX_init(&md);             /* new function call */
13357           EVP_DigestInit(&md, EVP_sha1());
13358           EVP_DigestUpdate(&md, in, len);
13359           EVP_DigestFinal(&md, out, NULL);
13360           EVP_MD_CTX_cleanup(&md);          /* new function call */
13361
13362   *Ben Laurie*
13363
13364 * Make DES key schedule conform to the usual scheme, as well as
13365   correcting its structure. This means that calls to DES functions
13366   now have to pass a pointer to a des_key_schedule instead of a
13367   plain des_key_schedule (which was actually always a pointer
13368   anyway): E.g.,
13369
13370           des_key_schedule ks;
13371
13372           des_set_key_checked(..., &ks);
13373           des_ncbc_encrypt(..., &ks, ...);
13374
13375   (Note that a later change renames 'des_...' into 'DES_...'.)
13376
13377   *Ben Laurie*
13378
13379 * Initial reduction of linker bloat: the use of some functions, such as
13380   PEM causes large amounts of unused functions to be linked in due to
13381   poor organisation. For example pem_all.c contains every PEM function
13382   which has a knock on effect of linking in large amounts of (unused)
13383   ASN1 code. Grouping together similar functions and splitting unrelated
13384   functions prevents this.
13385
13386   *Steve Henson*
13387
13388 * Cleanup of EVP macros.
13389
13390   *Ben Laurie*
13391
13392 * Change historical references to `{NID,SN,LN}_des_ede` and ede3 to add the
13393   correct `_ecb suffix`.
13394
13395   *Ben Laurie*
13396
13397 * Add initial OCSP responder support to ocsp application. The
13398   revocation information is handled using the text based index
13399   use by the ca application. The responder can either handle
13400   requests generated internally, supplied in files (for example
13401   via a CGI script) or using an internal minimal server.
13402
13403   *Steve Henson*
13404
13405 * Add configuration choices to get zlib compression for TLS.
13406
13407   *Richard Levitte*
13408
13409 * Changes to Kerberos SSL for RFC 2712 compliance:
13410   1. Implemented real KerberosWrapper, instead of just using
13411      KRB5 AP_REQ message.  [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
13412   2. Implemented optional authenticator field of KerberosWrapper.
13413
13414   Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
13415   and authenticator structs; see crypto/krb5/.
13416
13417   Generalized Kerberos calls to support multiple Kerberos libraries.
13418   *Vern Staats <staatsvr@asc.hpc.mil>, Jeffrey Altman <jaltman@columbia.edu>
13419   via Richard Levitte*
13420
13421 * Cause 'openssl speed' to use fully hard-coded DSA keys as it
13422   already does with RSA. testdsa.h now has 'priv_key/pub_key'
13423   values for each of the key sizes rather than having just
13424   parameters (and 'speed' generating keys each time).
13425
13426   *Geoff Thorpe*
13427
13428 * Speed up EVP routines.
13429   Before:
13430crypt
13431pe              8 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
13432s-cbc           4408.85k     5560.51k     5778.46k     5862.20k     5825.16k
13433s-cbc           4389.55k     5571.17k     5792.23k     5846.91k     5832.11k
13434s-cbc           4394.32k     5575.92k     5807.44k     5848.37k     5841.30k
13435crypt
13436s-cbc           3482.66k     5069.49k     5496.39k     5614.16k     5639.28k
13437s-cbc           3480.74k     5068.76k     5510.34k     5609.87k     5635.52k
13438s-cbc           3483.72k     5067.62k     5504.60k     5708.01k     5724.80k
13439   After:
13440crypt
13441s-cbc           4660.16k     5650.19k     5807.19k     5827.13k     5783.32k
13442crypt
13443s-cbc           3624.96k     5258.21k     5530.91k     5624.30k     5628.26k
13444
13445   *Ben Laurie*
13446
13447 * Added the OS2-EMX target.
13448
13449   *"Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte*
13450
13451 * Rewrite commands to use `NCONF` routines instead of the old `CONF`.
13452   New functions to support `NCONF` routines in extension code.
13453   New function `CONF_set_nconf()`
13454   to allow functions which take an `NCONF` to also handle the old `LHASH`
13455   structure: this means that the old `CONF` compatible routines can be
13456   retained (in particular w.rt. extensions) without having to duplicate the
13457   code. New function `X509V3_add_ext_nconf_sk()` to add extensions to a stack.
13458
13459   *Steve Henson*
13460
13461 * Enhance the general user interface with mechanisms for inner control
13462   and with possibilities to have yes/no kind of prompts.
13463
13464   *Richard Levitte*
13465
13466 * Change all calls to low-level digest routines in the library and
13467   applications to use EVP. Add missing calls to HMAC_cleanup() and
13468   don't assume HMAC_CTX can be copied using memcpy().
13469
13470   *Verdon Walker <VWalker@novell.com>, Steve Henson*
13471
13472 * Add the possibility to control engines through control names but with
13473   arbitrary arguments instead of just a string.
13474   Change the key loaders to take a UI_METHOD instead of a callback
13475   function pointer.  NOTE: this breaks binary compatibility with earlier
13476   versions of OpenSSL [engine].
13477   Adapt the nCipher code for these new conditions and add a card insertion
13478   callback.
13479
13480   *Richard Levitte*
13481
13482 * Enhance the general user interface with mechanisms to better support
13483   dialog box interfaces, application-defined prompts, the possibility
13484   to use defaults (for example default passwords from somewhere else)
13485   and interrupts/cancellations.
13486
13487   *Richard Levitte*
13488
13489 * Tidy up PKCS#12 attribute handling. Add support for the CSP name
13490   attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
13491
13492   *Steve Henson*
13493
13494 * Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
13495   tidy up some unnecessarily weird code in 'sk_new()').
13496
13497   *Geoff, reported by Diego Tartara <dtartara@novamens.com>*
13498
13499 * Change the key loading routines for ENGINEs to use the same kind
13500   callback (pem_password_cb) as all other routines that need this
13501   kind of callback.
13502
13503   *Richard Levitte*
13504
13505 * Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
13506   256 bit (=32 byte) keys. Of course seeding with more entropy bytes
13507   than this minimum value is recommended.
13508
13509   *Lutz Jaenicke*
13510
13511 * New random seeder for OpenVMS, using the system process statistics
13512   that are easily reachable.
13513
13514   *Richard Levitte*
13515
13516 * Windows apparently can't transparently handle global
13517   variables defined in DLLs. Initialisations such as:
13518
13519           const ASN1_ITEM *it = &ASN1_INTEGER_it;
13520
13521   won't compile. This is used by the any applications that need to
13522   declare their own ASN1 modules. This was fixed by adding the option
13523   EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
13524   needed for static libraries under Win32.
13525
13526   *Steve Henson*
13527
13528 * New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
13529   setting of purpose and trust fields. New X509_STORE trust and
13530   purpose functions and tidy up setting in other SSL functions.
13531
13532   *Steve Henson*
13533
13534 * Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
13535   structure. These are inherited by X509_STORE_CTX when it is
13536   initialised. This allows various defaults to be set in the
13537   X509_STORE structure (such as flags for CRL checking and custom
13538   purpose or trust settings) for functions which only use X509_STORE_CTX
13539   internally such as S/MIME.
13540
13541   Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
13542   trust settings if they are not set in X509_STORE. This allows X509_STORE
13543   purposes and trust (in S/MIME for example) to override any set by default.
13544
13545   Add command line options for CRL checking to smime, s_client and s_server
13546   applications.
13547
13548   *Steve Henson*
13549
13550 * Initial CRL based revocation checking. If the CRL checking flag(s)
13551   are set then the CRL is looked up in the X509_STORE structure and
13552   its validity and signature checked, then if the certificate is found
13553   in the CRL the verify fails with a revoked error.
13554
13555   Various new CRL related callbacks added to X509_STORE_CTX structure.
13556
13557   Command line options added to 'verify' application to support this.
13558
13559   This needs some additional work, such as being able to handle multiple
13560   CRLs with different times, extension based lookup (rather than just
13561   by subject name) and ultimately more complete V2 CRL extension
13562   handling.
13563
13564   *Steve Henson*
13565
13566 * Add a general user interface API (crypto/ui/).  This is designed
13567   to replace things like des_read_password and friends (backward
13568   compatibility functions using this new API are provided).
13569   The purpose is to remove prompting functions from the DES code
13570   section as well as provide for prompting through dialog boxes in
13571   a window system and the like.
13572
13573   *Richard Levitte*
13574
13575 * Add "ex_data" support to ENGINE so implementations can add state at a
13576   per-structure level rather than having to store it globally.
13577
13578   *Geoff*
13579
13580 * Make it possible for ENGINE structures to be copied when retrieved by
13581   ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
13582   This causes the "original" ENGINE structure to act like a template,
13583   analogous to the RSA vs. RSA_METHOD type of separation. Because of this
13584   operational state can be localised to each ENGINE structure, despite the
13585   fact they all share the same "methods". New ENGINE structures returned in
13586   this case have no functional references and the return value is the single
13587   structural reference. This matches the single structural reference returned
13588   by ENGINE_by_id() normally, when it is incremented on the pre-existing
13589   ENGINE structure.
13590
13591   *Geoff*
13592
13593 * Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
13594   needs to match any other type at all we need to manually clear the
13595   tag cache.
13596
13597   *Steve Henson*
13598
13599 * Changes to the "openssl engine" utility to include;
13600   - verbosity levels ('-v', '-vv', and '-vvv') that provide information
13601     about an ENGINE's available control commands.
13602   - executing control commands from command line arguments using the
13603     '-pre' and '-post' switches. '-post' is only used if '-t' is
13604     specified and the ENGINE is successfully initialised. The syntax for
13605     the individual commands are colon-separated, for example;
13606           openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
13607
13608   *Geoff*
13609
13610 * New dynamic control command support for ENGINEs. ENGINEs can now
13611   declare their own commands (numbers), names (strings), descriptions,
13612   and input types for run-time discovery by calling applications. A
13613   subset of these commands are implicitly classed as "executable"
13614   depending on their input type, and only these can be invoked through
13615   the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
13616   can be based on user input, config files, etc). The distinction is
13617   that "executable" commands cannot return anything other than a boolean
13618   result and can only support numeric or string input, whereas some
13619   discoverable commands may only be for direct use through
13620   ENGINE_ctrl(), eg. supporting the exchange of binary data, function
13621   pointers, or other custom uses. The "executable" commands are to
13622   support parameterisations of ENGINE behaviour that can be
13623   unambiguously defined by ENGINEs and used consistently across any
13624   OpenSSL-based application. Commands have been added to all the
13625   existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
13626   control over shared-library paths without source code alterations.
13627
13628   *Geoff*
13629
13630 * Changed all ENGINE implementations to dynamically allocate their
13631   ENGINEs rather than declaring them statically. Apart from this being
13632   necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
13633   this also allows the implementations to compile without using the
13634   internal engine_int.h header.
13635
13636   *Geoff*
13637
13638 * Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
13639   'const' value. Any code that should be able to modify a RAND_METHOD
13640   should already have non-const pointers to it (ie. they should only
13641   modify their own ones).
13642
13643   *Geoff*
13644
13645 * Made a variety of little tweaks to the ENGINE code.
13646   - "atalla" and "ubsec" string definitions were moved from header files
13647     to C code. "nuron" string definitions were placed in variables
13648     rather than hard-coded - allowing parameterisation of these values
13649     later on via ctrl() commands.
13650   - Removed unused "#if 0"'d code.
13651   - Fixed engine list iteration code so it uses ENGINE_free() to release
13652     structural references.
13653   - Constified the RAND_METHOD element of ENGINE structures.
13654   - Constified various get/set functions as appropriate and added
13655     missing functions (including a catch-all ENGINE_cpy that duplicates
13656     all ENGINE values onto a new ENGINE except reference counts/state).
13657   - Removed NULL parameter checks in get/set functions. Setting a method
13658     or function to NULL is a way of cancelling out a previously set
13659     value.  Passing a NULL ENGINE parameter is just plain stupid anyway
13660     and doesn't justify the extra error symbols and code.
13661   - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
13662     flags from engine_int.h to engine.h.
13663   - Changed prototypes for ENGINE handler functions (init(), finish(),
13664     ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
13665
13666   *Geoff*
13667
13668 * Implement binary inversion algorithm for BN_mod_inverse in addition
13669   to the algorithm using long division.  The binary algorithm can be
13670   used only if the modulus is odd.  On 32-bit systems, it is faster
13671   only for relatively small moduli (roughly 20-30% for 128-bit moduli,
13672   roughly 5-15% for 256-bit moduli), so we use it only for moduli
13673   up to 450 bits.  In 64-bit environments, the binary algorithm
13674   appears to be advantageous for much longer moduli; here we use it
13675   for moduli up to 2048 bits.
13676
13677   *Bodo Moeller*
13678
13679 * Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
13680   could not support the combine flag in choice fields.
13681
13682   *Steve Henson*
13683
13684 * Add a 'copy_extensions' option to the 'ca' utility. This copies
13685   extensions from a certificate request to the certificate.
13686
13687   *Steve Henson*
13688
13689 * Allow multiple 'certopt' and 'nameopt' options to be separated
13690   by commas. Add 'namopt' and 'certopt' options to the 'ca' config
13691   file: this allows the display of the certificate about to be
13692   signed to be customised, to allow certain fields to be included
13693   or excluded and extension details. The old system didn't display
13694   multicharacter strings properly, omitted fields not in the policy
13695   and couldn't display additional details such as extensions.
13696
13697   *Steve Henson*
13698
13699 * Function EC_POINTs_mul for multiple scalar multiplication
13700   of an arbitrary number of elliptic curve points
13701           \sum scalars[i]*points[i],
13702   optionally including the generator defined for the EC_GROUP:
13703           scalar*generator +  \sum scalars[i]*points[i].
13704
13705   EC_POINT_mul is a simple wrapper function for the typical case
13706   that the point list has just one item (besides the optional
13707   generator).
13708
13709   *Bodo Moeller*
13710
13711 * First EC_METHODs for curves over GF(p):
13712
13713   EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
13714   operations and provides various method functions that can also
13715   operate with faster implementations of modular arithmetic.
13716
13717   EC_GFp_mont_method() reuses most functions that are part of
13718   EC_GFp_simple_method, but uses Montgomery arithmetic.
13719
13720   *Bodo Moeller; point addition and point doubling
13721   implementation directly derived from source code provided by
13722   Lenka Fibikova <fibikova@exp-math.uni-essen.de>*
13723
13724 * Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
13725   crypto/ec/ec_lib.c):
13726
13727   Curves are EC_GROUP objects (with an optional group generator)
13728   based on EC_METHODs that are built into the library.
13729
13730   Points are EC_POINT objects based on EC_GROUP objects.
13731
13732   Most of the framework would be able to handle curves over arbitrary
13733   finite fields, but as there are no obvious types for fields other
13734   than GF(p), some functions are limited to that for now.
13735
13736   *Bodo Moeller*
13737
13738 * Add the -HTTP option to s_server.  It is similar to -WWW, but requires
13739   that the file contains a complete HTTP response.
13740
13741   *Richard Levitte*
13742
13743 * Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
13744   change the def and num file printf format specifier from "%-40sXXX"
13745   to "%-39s XXX". The latter will always guarantee a space after the
13746   field while the former will cause them to run together if the field
13747   is 40 of more characters long.
13748
13749   *Steve Henson*
13750
13751 * Constify the cipher and digest 'method' functions and structures
13752   and modify related functions to take constant EVP_MD and EVP_CIPHER
13753   pointers.
13754
13755   *Steve Henson*
13756
13757 * Hide BN_CTX structure details in bn_lcl.h instead of publishing them
13758   in <openssl/bn.h>.  Also further increase BN_CTX_NUM to 32.
13759
13760   *Bodo Moeller*
13761
13762 * Modify `EVP_Digest*()` routines so they now return values. Although the
13763   internal software routines can never fail additional hardware versions
13764   might.
13765
13766   *Steve Henson*
13767
13768 * Clean up crypto/err/err.h and change some error codes to avoid conflicts:
13769
13770   Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
13771   (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
13772
13773   ASN1 error codes
13774           ERR_R_NESTED_ASN1_ERROR
13775           ...
13776           ERR_R_MISSING_ASN1_EOS
13777   were 4 .. 9, conflicting with
13778           ERR_LIB_RSA (= ERR_R_RSA_LIB)
13779           ...
13780           ERR_LIB_PEM (= ERR_R_PEM_LIB).
13781   They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
13782
13783   Add new error code 'ERR_R_INTERNAL_ERROR'.
13784
13785   *Bodo Moeller*
13786
13787 * Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
13788   suffices.
13789
13790   *Bodo Moeller*
13791
13792 * New option '-subj arg' for 'openssl req' and 'openssl ca'.  This
13793   sets the subject name for a new request or supersedes the
13794   subject name in a given request. Formats that can be parsed are
13795           'CN=Some Name, OU=myOU, C=IT'
13796   and
13797           'CN=Some Name/OU=myOU/C=IT'.
13798
13799   Add options '-batch' and '-verbose' to 'openssl req'.
13800
13801   *Massimiliano Pala <madwolf@hackmasters.net>*
13802
13803 * Introduce the possibility to access global variables through
13804   functions on platform were that's the best way to handle exporting
13805   global variables in shared libraries.  To enable this functionality,
13806   one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
13807   "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
13808   is normally done by Configure or something similar).
13809
13810   To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
13811   in the source file (foo.c) like this:
13812
13813           OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
13814           OPENSSL_IMPLEMENT_GLOBAL(double,bar);
13815
13816   To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
13817   and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
13818
13819           OPENSSL_DECLARE_GLOBAL(int,foo);
13820           #define foo OPENSSL_GLOBAL_REF(foo)
13821           OPENSSL_DECLARE_GLOBAL(double,bar);
13822           #define bar OPENSSL_GLOBAL_REF(bar)
13823
13824   The #defines are very important, and therefore so is including the
13825   header file everywhere where the defined globals are used.
13826
13827   The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
13828   of ASN.1 items, but that structure is a bit different.
13829
13830   The largest change is in util/mkdef.pl which has been enhanced with
13831   better and easier to understand logic to choose which symbols should
13832   go into the Windows .def files as well as a number of fixes and code
13833   cleanup (among others, algorithm keywords are now sorted
13834   lexicographically to avoid constant rewrites).
13835
13836   *Richard Levitte*
13837
13838 * In BN_div() keep a copy of the sign of 'num' before writing the
13839   result to 'rm' because if rm==num the value will be overwritten
13840   and produce the wrong result if 'num' is negative: this caused
13841   problems with BN_mod() and BN_nnmod().
13842
13843   *Steve Henson*
13844
13845 * Function OCSP_request_verify(). This checks the signature on an
13846   OCSP request and verifies the signer certificate. The signer
13847   certificate is just checked for a generic purpose and OCSP request
13848   trust settings.
13849
13850   *Steve Henson*
13851
13852 * Add OCSP_check_validity() function to check the validity of OCSP
13853   responses. OCSP responses are prepared in real time and may only
13854   be a few seconds old. Simply checking that the current time lies
13855   between thisUpdate and nextUpdate max reject otherwise valid responses
13856   caused by either OCSP responder or client clock inaccuracy. Instead
13857   we allow thisUpdate and nextUpdate to fall within a certain period of
13858   the current time. The age of the response can also optionally be
13859   checked. Two new options -validity_period and -status_age added to
13860   ocsp utility.
13861
13862   *Steve Henson*
13863
13864 * If signature or public key algorithm is unrecognized print out its
13865   OID rather that just UNKNOWN.
13866
13867   *Steve Henson*
13868
13869 * Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
13870   OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
13871   ID to be generated from the issuer certificate alone which can then be
13872   passed to OCSP_id_issuer_cmp().
13873
13874   *Steve Henson*
13875
13876 * New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
13877   ASN1 modules to export functions returning ASN1_ITEM pointers
13878   instead of the ASN1_ITEM structures themselves. This adds several
13879   new macros which allow the underlying ASN1 function/structure to
13880   be accessed transparently. As a result code should not use ASN1_ITEM
13881   references directly (such as &X509_it) but instead use the relevant
13882   macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
13883   use of the new ASN1 code on platforms where exporting structures
13884   is problematical (for example in shared libraries) but exporting
13885   functions returning pointers to structures is not.
13886
13887   *Steve Henson*
13888
13889 * Add support for overriding the generation of SSL/TLS session IDs.
13890   These callbacks can be registered either in an SSL_CTX or per SSL.
13891   The purpose of this is to allow applications to control, if they wish,
13892   the arbitrary values chosen for use as session IDs, particularly as it
13893   can be useful for session caching in multiple-server environments. A
13894   command-line switch for testing this (and any client code that wishes
13895   to use such a feature) has been added to "s_server".
13896
13897   *Geoff Thorpe, Lutz Jaenicke*
13898
13899 * Modify mkdef.pl to recognise and parse preprocessor conditionals
13900   of the form `#if defined(...) || defined(...) || ...` and
13901   `#if !defined(...) && !defined(...) && ...`.  This also avoids
13902   the growing number of special cases it was previously handling.
13903
13904   *Richard Levitte*
13905
13906 * Make all configuration macros available for application by making
13907   sure they are available in opensslconf.h, by giving them names starting
13908   with `OPENSSL_` to avoid conflicts with other packages and by making
13909   sure e_os2.h will cover all platform-specific cases together with
13910   opensslconf.h.
13911   Additionally, it is now possible to define configuration/platform-
13912   specific names (called "system identities").  In the C code, these
13913   are prefixed with `OPENSSL_SYSNAME_`.  e_os2.h will create another
13914   macro with the name beginning with `OPENSSL_SYS_`, which is determined
13915   from `OPENSSL_SYSNAME_*` or compiler-specific macros depending on
13916   what is available.
13917
13918   *Richard Levitte*
13919
13920 * New option -set_serial to 'req' and 'x509' this allows the serial
13921   number to use to be specified on the command line. Previously self
13922   signed certificates were hard coded with serial number 0 and the
13923   CA options of 'x509' had to use a serial number in a file which was
13924   auto incremented.
13925
13926   *Steve Henson*
13927
13928 * New options to 'ca' utility to support V2 CRL entry extensions.
13929   Currently CRL reason, invalidity date and hold instruction are
13930   supported. Add new CRL extensions to V3 code and some new objects.
13931
13932   *Steve Henson*
13933
13934 * New function EVP_CIPHER_CTX_set_padding() this is used to
13935   disable standard block padding (aka PKCS#5 padding) in the EVP
13936   API, which was previously mandatory. This means that the data is
13937   not padded in any way and so the total length much be a multiple
13938   of the block size, otherwise an error occurs.
13939
13940   *Steve Henson*
13941
13942 * Initial (incomplete) OCSP SSL support.
13943
13944   *Steve Henson*
13945
13946 * New function OCSP_parse_url(). This splits up a URL into its host,
13947   port and path components: primarily to parse OCSP URLs. New -url
13948   option to ocsp utility.
13949
13950   *Steve Henson*
13951
13952 * New nonce behavior. The return value of OCSP_check_nonce() now
13953   reflects the various checks performed. Applications can decide
13954   whether to tolerate certain situations such as an absent nonce
13955   in a response when one was present in a request: the ocsp application
13956   just prints out a warning. New function OCSP_add1_basic_nonce()
13957   this is to allow responders to include a nonce in a response even if
13958   the request is nonce-less.
13959
13960   *Steve Henson*
13961
13962 * Disable stdin buffering in `load_cert()` (`apps/apps.c`) so that no certs are
13963   skipped when using openssl x509 multiple times on a single input file,
13964   e.g. `(openssl x509 -out cert1; openssl x509 -out cert2) <certs`.
13965
13966   *Bodo Moeller*
13967
13968 * Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
13969   set string type: to handle setting ASN1_TIME structures. Fix ca
13970   utility to correctly initialize revocation date of CRLs.
13971
13972   *Steve Henson*
13973
13974 * New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
13975   the clients preferred ciphersuites and rather use its own preferences.
13976   Should help to work around M$ SGC (Server Gated Cryptography) bug in
13977   Internet Explorer by ensuring unchanged hash method during stepup.
13978   (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
13979
13980   *Lutz Jaenicke*
13981
13982 * Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
13983   to aes and add a new 'exist' option to print out symbols that don't
13984   appear to exist.
13985
13986   *Steve Henson*
13987
13988 * Additional options to ocsp utility to allow flags to be set and
13989   additional certificates supplied.
13990
13991   *Steve Henson*
13992
13993 * Add the option -VAfile to 'openssl ocsp', so the user can give the
13994   OCSP client a number of certificate to only verify the response
13995   signature against.
13996
13997   *Richard Levitte*
13998
13999 * Update Rijndael code to version 3.0 and change EVP AES ciphers to
14000   handle the new API. Currently only ECB, CBC modes supported. Add new
14001   AES OIDs.
14002
14003   Add TLS AES ciphersuites as described in RFC3268, "Advanced
14004   Encryption Standard (AES) Ciphersuites for Transport Layer
14005   Security (TLS)".  (In beta versions of OpenSSL 0.9.7, these were
14006   not enabled by default and were not part of the "ALL" ciphersuite
14007   alias because they were not yet official; they could be
14008   explicitly requested by specifying the "AESdraft" ciphersuite
14009   group alias.  In the final release of OpenSSL 0.9.7, the group
14010   alias is called "AES" and is part of "ALL".)
14011
14012   *Ben Laurie, Steve  Henson, Bodo Moeller*
14013
14014 * New function OCSP_copy_nonce() to copy nonce value (if present) from
14015   request to response.
14016
14017   *Steve Henson*
14018
14019 * Functions for OCSP responders. OCSP_request_onereq_count(),
14020   OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
14021   extract information from a certificate request. OCSP_response_create()
14022   creates a response and optionally adds a basic response structure.
14023   OCSP_basic_add1_status() adds a complete single response to a basic
14024   response and returns the OCSP_SINGLERESP structure just added (to allow
14025   extensions to be included for example). OCSP_basic_add1_cert() adds a
14026   certificate to a basic response and OCSP_basic_sign() signs a basic
14027   response with various flags. New helper functions ASN1_TIME_check()
14028   (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
14029   (converts ASN1_TIME to GeneralizedTime).
14030
14031   *Steve Henson*
14032
14033 * Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
14034   in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
14035   structure from a certificate. X509_pubkey_digest() digests the public_key
14036   contents: this is used in various key identifiers.
14037
14038   *Steve Henson*
14039
14040 * Make sk_sort() tolerate a NULL argument.
14041
14042   *Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>*
14043
14044 * New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
14045   passed by the function are trusted implicitly. If any of them signed the
14046   response then it is assumed to be valid and is not verified.
14047
14048   *Steve Henson*
14049
14050 * In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
14051   to data. This was previously part of the PKCS7 ASN1 code. This
14052   was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
14053   *Steve Henson, reported by Kenneth R. Robinette
14054                              <support@securenetterm.com>*
14055
14056 * Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
14057   routines: without these tracing memory leaks is very painful.
14058   Fix leaks in PKCS12 and PKCS7 routines.
14059
14060   *Steve Henson*
14061
14062 * Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
14063   Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
14064   effectively meant GeneralizedTime would never be used. Now it
14065   is initialised to -1 but X509_time_adj() now has to check the value
14066   and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
14067   V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
14068   *Steve Henson, reported by Kenneth R. Robinette
14069                              <support@securenetterm.com>*
14070
14071 * Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
14072   result in a zero length in the ASN1_INTEGER structure which was
14073   not consistent with the structure when d2i_ASN1_INTEGER() was used
14074   and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
14075   to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
14076   where it did not print out a minus for negative ASN1_INTEGER.
14077
14078   *Steve Henson*
14079
14080 * Add summary printout to ocsp utility. The various functions which
14081   convert status values to strings have been renamed to:
14082   OCSP_response_status_str(), OCSP_cert_status_str() and
14083   OCSP_crl_reason_str() and are no longer static. New options
14084   to verify nonce values and to disable verification. OCSP response
14085   printout format cleaned up.
14086
14087   *Steve Henson*
14088
14089 * Add additional OCSP certificate checks. These are those specified
14090   in RFC2560. This consists of two separate checks: the CA of the
14091   certificate being checked must either be the OCSP signer certificate
14092   or the issuer of the OCSP signer certificate. In the latter case the
14093   OCSP signer certificate must contain the OCSP signing extended key
14094   usage. This check is performed by attempting to match the OCSP
14095   signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
14096   in the OCSP_CERTID structures of the response.
14097
14098   *Steve Henson*
14099
14100 * Initial OCSP certificate verification added to OCSP_basic_verify()
14101   and related routines. This uses the standard OpenSSL certificate
14102   verify routines to perform initial checks (just CA validity) and
14103   to obtain the certificate chain. Then additional checks will be
14104   performed on the chain. Currently the root CA is checked to see
14105   if it is explicitly trusted for OCSP signing. This is used to set
14106   a root CA as a global signing root: that is any certificate that
14107   chains to that CA is an acceptable OCSP signing certificate.
14108
14109   *Steve Henson*
14110
14111 * New '-extfile ...' option to 'openssl ca' for reading X.509v3
14112   extensions from a separate configuration file.
14113   As when reading extensions from the main configuration file,
14114   the '-extensions ...' option may be used for specifying the
14115   section to use.
14116
14117   *Massimiliano Pala <madwolf@comune.modena.it>*
14118
14119 * New OCSP utility. Allows OCSP requests to be generated or
14120   read. The request can be sent to a responder and the output
14121   parsed, outputted or printed in text form. Not complete yet:
14122   still needs to check the OCSP response validity.
14123
14124   *Steve Henson*
14125
14126 * New subcommands for 'openssl ca':
14127   `openssl ca -status <serial>` prints the status of the cert with
14128   the given serial number (according to the index file).
14129   `openssl ca -updatedb` updates the expiry status of certificates
14130   in the index file.
14131
14132   *Massimiliano Pala <madwolf@comune.modena.it>*
14133
14134 * New '-newreq-nodes' command option to CA.pl.  This is like
14135   '-newreq', but calls 'openssl req' with the '-nodes' option
14136   so that the resulting key is not encrypted.
14137
14138   *Damien Miller <djm@mindrot.org>*
14139
14140 * New configuration for the GNU Hurd.
14141
14142   *Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte*
14143
14144 * Initial code to implement OCSP basic response verify. This
14145   is currently incomplete. Currently just finds the signer's
14146   certificate and verifies the signature on the response.
14147
14148   *Steve Henson*
14149
14150 * New SSLeay_version code SSLEAY_DIR to determine the compiled-in
14151   value of OPENSSLDIR.  This is available via the new '-d' option
14152   to 'openssl version', and is also included in 'openssl version -a'.
14153
14154   *Bodo Moeller*
14155
14156 * Allowing defining memory allocation callbacks that will be given
14157   file name and line number information in additional arguments
14158   (a `const char*` and an int).  The basic functionality remains, as
14159   well as the original possibility to just replace malloc(),
14160   realloc() and free() by functions that do not know about these
14161   additional arguments.  To register and find out the current
14162   settings for extended allocation functions, the following
14163   functions are provided:
14164
14165           CRYPTO_set_mem_ex_functions
14166           CRYPTO_set_locked_mem_ex_functions
14167           CRYPTO_get_mem_ex_functions
14168           CRYPTO_get_locked_mem_ex_functions
14169
14170   These work the same way as CRYPTO_set_mem_functions and friends.
14171   `CRYPTO_get_[locked_]mem_functions` now writes 0 where such an
14172   extended allocation function is enabled.
14173   Similarly, `CRYPTO_get_[locked_]mem_ex_functions` writes 0 where
14174   a conventional allocation function is enabled.
14175
14176   *Richard Levitte, Bodo Moeller*
14177
14178 * Finish off removing the remaining LHASH function pointer casts.
14179   There should no longer be any prototype-casting required when using
14180   the LHASH abstraction, and any casts that remain are "bugs". See
14181   the callback types and macros at the head of lhash.h for details
14182   (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
14183
14184   *Geoff Thorpe*
14185
14186 * Add automatic query of EGD sockets in RAND_poll() for the unix variant.
14187   If /dev/[u]random devices are not available or do not return enough
14188   entropy, EGD style sockets (served by EGD or PRNGD) will automatically
14189   be queried.
14190   The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
14191   /etc/entropy will be queried once each in this sequence, querying stops
14192   when enough entropy was collected without querying more sockets.
14193
14194   *Lutz Jaenicke*
14195
14196 * Change the Unix RAND_poll() variant to be able to poll several
14197   random devices, as specified by DEVRANDOM, until a sufficient amount
14198   of data has been collected.   We spend at most 10 ms on each file
14199   (select timeout) and read in non-blocking mode.  DEVRANDOM now
14200   defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
14201   (previously it was just the string "/dev/urandom"), so on typical
14202   platforms the 10 ms delay will never occur.
14203   Also separate out the Unix variant to its own file, rand_unix.c.
14204   For VMS, there's a currently-empty rand_vms.c.
14205
14206   *Richard Levitte*
14207
14208 * Move OCSP client related routines to ocsp_cl.c. These
14209   provide utility functions which an application needing
14210   to issue a request to an OCSP responder and analyse the
14211   response will typically need: as opposed to those which an
14212   OCSP responder itself would need which will be added later.
14213
14214   OCSP_request_sign() signs an OCSP request with an API similar
14215   to PKCS7_sign(). OCSP_response_status() returns status of OCSP
14216   response. OCSP_response_get1_basic() extracts basic response
14217   from response. OCSP_resp_find_status(): finds and extracts status
14218   information from an OCSP_CERTID structure (which will be created
14219   when the request structure is built). These are built from lower
14220   level functions which work on OCSP_SINGLERESP structures but
14221   won't normally be used unless the application wishes to examine
14222   extensions in the OCSP response for example.
14223
14224   Replace nonce routines with a pair of functions.
14225   OCSP_request_add1_nonce() adds a nonce value and optionally
14226   generates a random value. OCSP_check_nonce() checks the
14227   validity of the nonce in an OCSP response.
14228
14229   *Steve Henson*
14230
14231 * Change function OCSP_request_add() to OCSP_request_add0_id().
14232   This doesn't copy the supplied OCSP_CERTID and avoids the
14233   need to free up the newly created id. Change return type
14234   to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
14235   This can then be used to add extensions to the request.
14236   Deleted OCSP_request_new(), since most of its functionality
14237   is now in OCSP_REQUEST_new() (and the case insensitive name
14238   clash) apart from the ability to set the request name which
14239   will be added elsewhere.
14240
14241   *Steve Henson*
14242
14243 * Update OCSP API. Remove obsolete extensions argument from
14244   various functions. Extensions are now handled using the new
14245   OCSP extension code. New simple OCSP HTTP function which
14246   can be used to send requests and parse the response.
14247
14248   *Steve Henson*
14249
14250 * Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
14251   ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
14252   uses the special reorder version of SET OF to sort the attributes
14253   and reorder them to match the encoded order. This resolves a long
14254   standing problem: a verify on a PKCS7 structure just after signing
14255   it used to fail because the attribute order did not match the
14256   encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
14257   it uses the received order. This is necessary to tolerate some broken
14258   software that does not order SET OF. This is handled by encoding
14259   as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
14260   to produce the required SET OF.
14261
14262   *Steve Henson*
14263
14264 * Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
14265   OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
14266   files to get correct declarations of the ASN.1 item variables.
14267
14268   *Richard Levitte*
14269
14270 * Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
14271   PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
14272   asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
14273   NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
14274   New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
14275   ASN1_ITEM and no wrapper functions.
14276
14277   *Steve Henson*
14278
14279 * New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
14280   replace the old function pointer based I/O routines. Change most of
14281   the `*_d2i_bio()` and `*_d2i_fp()` functions to use these.
14282
14283   *Steve Henson*
14284
14285 * Enhance mkdef.pl to be more accepting about spacing in C preprocessor
14286   lines, recognize more "algorithms" that can be deselected, and make
14287   it complain about algorithm deselection that isn't recognised.
14288
14289   *Richard Levitte*
14290
14291 * New ASN1 functions to handle dup, sign, verify, digest, pack and
14292   unpack operations in terms of ASN1_ITEM. Modify existing wrappers
14293   to use new functions. Add NO_ASN1_OLD which can be set to remove
14294   some old style ASN1 functions: this can be used to determine if old
14295   code will still work when these eventually go away.
14296
14297   *Steve Henson*
14298
14299 * New extension functions for OCSP structures, these follow the
14300   same conventions as certificates and CRLs.
14301
14302   *Steve Henson*
14303
14304 * New function X509V3_add1_i2d(). This automatically encodes and
14305   adds an extension. Its behaviour can be customised with various
14306   flags to append, replace or delete. Various wrappers added for
14307   certificates and CRLs.
14308
14309   *Steve Henson*
14310
14311 * Fix to avoid calling the underlying ASN1 print routine when
14312   an extension cannot be parsed. Correct a typo in the
14313   OCSP_SERVICELOC extension. Tidy up print OCSP format.
14314
14315   *Steve Henson*
14316
14317 * Make mkdef.pl parse some of the ASN1 macros and add appropriate
14318   entries for variables.
14319
14320   *Steve Henson*
14321
14322 * Add functionality to `apps/openssl.c` for detecting locking
14323   problems: As the program is single-threaded, all we have
14324   to do is register a locking callback using an array for
14325   storing which locks are currently held by the program.
14326
14327   *Bodo Moeller*
14328
14329 * Use a lock around the call to CRYPTO_get_ex_new_index() in
14330   SSL_get_ex_data_X509_STORE_idx(), which is used in
14331   ssl_verify_cert_chain() and thus can be called at any time
14332   during TLS/SSL handshakes so that thread-safety is essential.
14333   Unfortunately, the ex_data design is not at all suited
14334   for multi-threaded use, so it probably should be abolished.
14335
14336   *Bodo Moeller*
14337
14338 * Added Broadcom "ubsec" ENGINE to OpenSSL.
14339
14340   *Broadcom, tweaked and integrated by Geoff Thorpe*
14341
14342 * Move common extension printing code to new function
14343   X509V3_print_extensions(). Reorganise OCSP print routines and
14344   implement some needed OCSP ASN1 functions. Add OCSP extensions.
14345
14346   *Steve Henson*
14347
14348 * New function X509_signature_print() to remove duplication in some
14349   print routines.
14350
14351   *Steve Henson*
14352
14353 * Add a special meaning when SET OF and SEQUENCE OF flags are both
14354   set (this was treated exactly the same as SET OF previously). This
14355   is used to reorder the STACK representing the structure to match the
14356   encoding. This will be used to get round a problem where a PKCS7
14357   structure which was signed could not be verified because the STACK
14358   order did not reflect the encoded order.
14359
14360   *Steve Henson*
14361
14362 * Reimplement the OCSP ASN1 module using the new code.
14363
14364   *Steve Henson*
14365
14366 * Update the X509V3 code to permit the use of an ASN1_ITEM structure
14367   for its ASN1 operations. The old style function pointers still exist
14368   for now but they will eventually go away.
14369
14370   *Steve Henson*
14371
14372 * Merge in replacement ASN1 code from the ASN1 branch. This almost
14373   completely replaces the old ASN1 functionality with a table driven
14374   encoder and decoder which interprets an ASN1_ITEM structure describing
14375   the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
14376   largely maintained. Almost all of the old asn1_mac.h macro based ASN1
14377   has also been converted to the new form.
14378
14379   *Steve Henson*
14380
14381 * Change BN_mod_exp_recp so that negative moduli are tolerated
14382   (the sign is ignored).  Similarly, ignore the sign in BN_MONT_CTX_set
14383   so that BN_mod_exp_mont and BN_mod_exp_mont_word work
14384   for negative moduli.
14385
14386   *Bodo Moeller*
14387
14388 * Fix BN_uadd and BN_usub: Always return non-negative results instead
14389   of not touching the result's sign bit.
14390
14391   *Bodo Moeller*
14392
14393 * BN_div bugfix: If the result is 0, the sign (res->neg) must not be
14394   set.
14395
14396   *Bodo Moeller*
14397
14398 * Changed the LHASH code to use prototypes for callbacks, and created
14399   macros to declare and implement thin (optionally static) functions
14400   that provide type-safety and avoid function pointer casting for the
14401   type-specific callbacks.
14402
14403   *Geoff Thorpe*
14404
14405 * Added Kerberos Cipher Suites to be used with TLS, as written in
14406   RFC 2712.
14407   *Veers Staats <staatsvr@asc.hpc.mil>,
14408   Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte*
14409
14410 * Reformat the FAQ so the different questions and answers can be divided
14411   in sections depending on the subject.
14412
14413   *Richard Levitte*
14414
14415 * Have the zlib compression code load ZLIB.DLL dynamically under
14416   Windows.
14417
14418   *Richard Levitte*
14419
14420 * New function BN_mod_sqrt for computing square roots modulo a prime
14421   (using the probabilistic Tonelli-Shanks algorithm unless
14422   p == 3 (mod 4)  or  p == 5 (mod 8),  which are cases that can
14423   be handled deterministically).
14424
14425   *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller*
14426
14427 * Make BN_mod_inverse faster by explicitly handling small quotients
14428   in the Euclid loop. (Speed gain about 20% for small moduli [256 or
14429   512 bits], about 30% for larger ones [1024 or 2048 bits].)
14430
14431   *Bodo Moeller*
14432
14433 * New function BN_kronecker.
14434
14435   *Bodo Moeller*
14436
14437 * Fix BN_gcd so that it works on negative inputs; the result is
14438   positive unless both parameters are zero.
14439   Previously something reasonably close to an infinite loop was
14440   possible because numbers could be growing instead of shrinking
14441   in the implementation of Euclid's algorithm.
14442
14443   *Bodo Moeller*
14444
14445 * Fix BN_is_word() and BN_is_one() macros to take into account the
14446   sign of the number in question.
14447
14448   Fix BN_is_word(a,w) to work correctly for w == 0.
14449
14450   The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
14451   because its test if the absolute value of 'a' equals 'w'.
14452   Note that BN_abs_is_word does *not* handle w == 0 reliably;
14453   it exists mostly for use in the implementations of BN_is_zero(),
14454   BN_is_one(), and BN_is_word().
14455
14456   *Bodo Moeller*
14457
14458 * New function BN_swap.
14459
14460   *Bodo Moeller*
14461
14462 * Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
14463   the exponentiation functions are more likely to produce reasonable
14464   results on negative inputs.
14465
14466   *Bodo Moeller*
14467
14468 * Change BN_mod_mul so that the result is always non-negative.
14469   Previously, it could be negative if one of the factors was negative;
14470   I don't think anyone really wanted that behaviour.
14471
14472   *Bodo Moeller*
14473
14474 * Move `BN_mod_...` functions into new file `crypto/bn/bn_mod.c`
14475   (except for exponentiation, which stays in `crypto/bn/bn_exp.c`,
14476   and `BN_mod_mul_reciprocal`, which stays in `crypto/bn/bn_recp.c`)
14477   and add new functions:
14478
14479           BN_nnmod
14480           BN_mod_sqr
14481           BN_mod_add
14482           BN_mod_add_quick
14483           BN_mod_sub
14484           BN_mod_sub_quick
14485           BN_mod_lshift1
14486           BN_mod_lshift1_quick
14487           BN_mod_lshift
14488           BN_mod_lshift_quick
14489
14490   These functions always generate non-negative results.
14491
14492   `BN_nnmod` otherwise is `like BN_mod` (if `BN_mod` computes a remainder `r`
14493   such that `|m| < r < 0`, `BN_nnmod` will output `rem + |m|` instead).
14494
14495   `BN_mod_XXX_quick(r, a, [b,] m)` generates the same result as
14496   `BN_mod_XXX(r, a, [b,] m, ctx)`, but requires that `a` [and  `b`]
14497   be reduced modulo `m`.
14498
14499   *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller*
14500
14501<!--
14502   The following entry accidentally appeared in the CHANGES file
14503   distributed with OpenSSL 0.9.7.  The modifications described in
14504   it do *not* apply to OpenSSL 0.9.7.
14505
14506 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
14507   was actually never needed) and in BN_mul().  The removal in BN_mul()
14508   required a small change in bn_mul_part_recursive() and the addition
14509   of the functions bn_cmp_part_words(), bn_sub_part_words() and
14510   bn_add_part_words(), which do the same thing as bn_cmp_words(),
14511   bn_sub_words() and bn_add_words() except they take arrays with
14512   differing sizes.
14513
14514   *Richard Levitte*
14515-->
14516
14517 * In 'openssl passwd', verify passwords read from the terminal
14518   unless the '-salt' option is used (which usually means that
14519   verification would just waste user's time since the resulting
14520   hash is going to be compared with some given password hash)
14521   or the new '-noverify' option is used.
14522
14523   This is an incompatible change, but it does not affect
14524   non-interactive use of 'openssl passwd' (passwords on the command
14525   line, '-stdin' option, '-in ...' option) and thus should not
14526   cause any problems.
14527
14528   *Bodo Moeller*
14529
14530 * Remove all references to RSAref, since there's no more need for it.
14531
14532   *Richard Levitte*
14533
14534 * Make DSO load along a path given through an environment variable
14535   (SHLIB_PATH) with shl_load().
14536
14537   *Richard Levitte*
14538
14539 * Constify the ENGINE code as a result of BIGNUM constification.
14540   Also constify the RSA code and most things related to it.  In a
14541   few places, most notable in the depth of the ASN.1 code, ugly
14542   casts back to non-const were required (to be solved at a later
14543   time)
14544
14545   *Richard Levitte*
14546
14547 * Make it so the openssl application has all engines loaded by default.
14548
14549   *Richard Levitte*
14550
14551 * Constify the BIGNUM routines a little more.
14552
14553   *Richard Levitte*
14554
14555 * Add the following functions:
14556
14557           ENGINE_load_cswift()
14558           ENGINE_load_chil()
14559           ENGINE_load_atalla()
14560           ENGINE_load_nuron()
14561           ENGINE_load_builtin_engines()
14562
14563   That way, an application can itself choose if external engines that
14564   are built-in in OpenSSL shall ever be used or not.  The benefit is
14565   that applications won't have to be linked with libdl or other dso
14566   libraries unless it's really needed.
14567
14568   Changed 'openssl engine' to load all engines on demand.
14569   Changed the engine header files to avoid the duplication of some
14570   declarations (they differed!).
14571
14572   *Richard Levitte*
14573
14574 * 'openssl engine' can now list capabilities.
14575
14576   *Richard Levitte*
14577
14578 * Better error reporting in 'openssl engine'.
14579
14580   *Richard Levitte*
14581
14582 * Never call load_dh_param(NULL) in s_server.
14583
14584   *Bodo Moeller*
14585
14586 * Add engine application.  It can currently list engines by name and
14587   identity, and test if they are actually available.
14588
14589   *Richard Levitte*
14590
14591 * Improve RPM specification file by forcing symbolic linking and making
14592   sure the installed documentation is also owned by root.root.
14593
14594   *Damien Miller <djm@mindrot.org>*
14595
14596 * Give the OpenSSL applications more possibilities to make use of
14597   keys (public as well as private) handled by engines.
14598
14599   *Richard Levitte*
14600
14601 * Add OCSP code that comes from CertCo.
14602
14603   *Richard Levitte*
14604
14605 * Add VMS support for the Rijndael code.
14606
14607   *Richard Levitte*
14608
14609 * Added untested support for Nuron crypto accelerator.
14610
14611   *Ben Laurie*
14612
14613 * Add support for external cryptographic devices.  This code was
14614   previously distributed separately as the "engine" branch.
14615
14616   *Geoff Thorpe, Richard Levitte*
14617
14618 * Rework the filename-translation in the DSO code. It is now possible to
14619   have far greater control over how a "name" is turned into a filename
14620   depending on the operating environment and any oddities about the
14621   different shared library filenames on each system.
14622
14623   *Geoff Thorpe*
14624
14625 * Support threads on FreeBSD-elf in Configure.
14626
14627   *Richard Levitte*
14628
14629 * Fix for SHA1 assembly problem with MASM: it produces
14630   warnings about corrupt line number information when assembling
14631   with debugging information. This is caused by the overlapping
14632   of two sections.
14633
14634   *Bernd Matthes <mainbug@celocom.de>, Steve Henson*
14635
14636 * NCONF changes.
14637   NCONF_get_number() has no error checking at all.  As a replacement,
14638   NCONF_get_number_e() is defined (`_e` for "error checking") and is
14639   promoted strongly.  The old NCONF_get_number is kept around for
14640   binary backward compatibility.
14641   Make it possible for methods to load from something other than a BIO,
14642   by providing a function pointer that is given a name instead of a BIO.
14643   For example, this could be used to load configuration data from an
14644   LDAP server.
14645
14646   *Richard Levitte*
14647
14648 * Fix for non blocking accept BIOs. Added new I/O special reason
14649   BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
14650   with non blocking I/O was not possible because no retry code was
14651   implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
14652   this case.
14653
14654   *Steve Henson*
14655
14656 * Added the beginnings of Rijndael support.
14657
14658   *Ben Laurie*
14659
14660 * Fix for bug in DirectoryString mask setting. Add support for
14661   X509_NAME_print_ex() in 'req' and X509_print_ex() function
14662   to allow certificate printing to more controllable, additional
14663   'certopt' option to 'x509' to allow new printing options to be
14664   set.
14665
14666   *Steve Henson*
14667
14668 * Clean old EAY MD5 hack from e_os.h.
14669
14670   *Richard Levitte*
14671
14672### Changes between 0.9.6l and 0.9.6m  [17 Mar 2004]
14673
14674 * Fix null-pointer assignment in do_change_cipher_spec() revealed
14675   by using the Codenomicon TLS Test Tool ([CVE-2004-0079])
14676
14677   *Joe Orton, Steve Henson*
14678
14679### Changes between 0.9.6k and 0.9.6l  [04 Nov 2003]
14680
14681 * Fix additional bug revealed by the NISCC test suite:
14682
14683   Stop bug triggering large recursion when presented with
14684   certain ASN.1 tags ([CVE-2003-0851])
14685
14686   *Steve Henson*
14687
14688### Changes between 0.9.6j and 0.9.6k  [30 Sep 2003]
14689
14690 * Fix various bugs revealed by running the NISCC test suite:
14691
14692   Stop out of bounds reads in the ASN1 code when presented with
14693   invalid tags (CVE-2003-0543 and CVE-2003-0544).
14694
14695   If verify callback ignores invalid public key errors don't try to check
14696   certificate signature with the NULL public key.
14697
14698   *Steve Henson*
14699
14700 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
14701   if the server requested one: as stated in TLS 1.0 and SSL 3.0
14702   specifications.
14703
14704   *Steve Henson*
14705
14706 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
14707   extra data after the compression methods not only for TLS 1.0
14708   but also for SSL 3.0 (as required by the specification).
14709
14710   *Bodo Moeller; problem pointed out by Matthias Loepfe*
14711
14712 * Change X509_certificate_type() to mark the key as exported/exportable
14713   when it's 512 *bits* long, not 512 bytes.
14714
14715   *Richard Levitte*
14716
14717### Changes between 0.9.6i and 0.9.6j  [10 Apr 2003]
14718
14719 * Countermeasure against the Klima-Pokorny-Rosa extension of
14720   Bleichbacher's attack on PKCS #1 v1.5 padding: treat
14721   a protocol version number mismatch like a decryption error
14722   in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
14723
14724   *Bodo Moeller*
14725
14726 * Turn on RSA blinding by default in the default implementation
14727   to avoid a timing attack. Applications that don't want it can call
14728   RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
14729   They would be ill-advised to do so in most cases.
14730
14731   *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller*
14732
14733 * Change RSA blinding code so that it works when the PRNG is not
14734   seeded (in this case, the secret RSA exponent is abused as
14735   an unpredictable seed -- if it is not unpredictable, there
14736   is no point in blinding anyway).  Make RSA blinding thread-safe
14737   by remembering the creator's thread ID in rsa->blinding and
14738   having all other threads use local one-time blinding factors
14739   (this requires more computation than sharing rsa->blinding, but
14740   avoids excessive locking; and if an RSA object is not shared
14741   between threads, blinding will still be very fast).
14742
14743   *Bodo Moeller*
14744
14745### Changes between 0.9.6h and 0.9.6i  [19 Feb 2003]
14746
14747 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
14748   via timing by performing a MAC computation even if incorrect
14749   block cipher padding has been found.  This is a countermeasure
14750   against active attacks where the attacker has to distinguish
14751   between bad padding and a MAC verification error. ([CVE-2003-0078])
14752
14753   *Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
14754   Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
14755   Martin Vuagnoux (EPFL, Ilion)*
14756
14757### Changes between 0.9.6g and 0.9.6h  [5 Dec 2002]
14758
14759 * New function OPENSSL_cleanse(), which is used to cleanse a section of
14760   memory from its contents.  This is done with a counter that will
14761   place alternating values in each byte.  This can be used to solve
14762   two issues: 1) the removal of calls to memset() by highly optimizing
14763   compilers, and 2) cleansing with other values than 0, since those can
14764   be read through on certain media, for example a swap space on disk.
14765
14766   *Geoff Thorpe*
14767
14768 * Bugfix: client side session caching did not work with external caching,
14769   because the session->cipher setting was not restored when reloading
14770   from the external cache. This problem was masked, when
14771   SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
14772   (Found by Steve Haslam <steve@araqnid.ddts.net>.)
14773
14774   *Lutz Jaenicke*
14775
14776 * Fix client_certificate (ssl/s2_clnt.c): The permissible total
14777   length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
14778
14779   *Zeev Lieber <zeev-l@yahoo.com>*
14780
14781 * Undo an undocumented change introduced in 0.9.6e which caused
14782   repeated calls to OpenSSL_add_all_ciphers() and
14783   OpenSSL_add_all_digests() to be ignored, even after calling
14784   EVP_cleanup().
14785
14786   *Richard Levitte*
14787
14788 * Change the default configuration reader to deal with last line not
14789   being properly terminated.
14790
14791   *Richard Levitte*
14792
14793 * Change X509_NAME_cmp() so it applies the special rules on handling
14794   DN values that are of type PrintableString, as well as RDNs of type
14795   emailAddress where the value has the type ia5String.
14796
14797   *stefank@valicert.com via Richard Levitte*
14798
14799 * Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
14800   the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
14801   doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
14802   the bitwise-OR of the two for use by the majority of applications
14803   wanting this behaviour, and update the docs. The documented
14804   behaviour and actual behaviour were inconsistent and had been
14805   changing anyway, so this is more a bug-fix than a behavioural
14806   change.
14807
14808   *Geoff Thorpe, diagnosed by Nadav Har'El*
14809
14810 * Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
14811   (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
14812
14813   *Bodo Moeller*
14814
14815 * Fix initialization code race conditions in
14816           SSLv23_method(),  SSLv23_client_method(),   SSLv23_server_method(),
14817           SSLv2_method(),   SSLv2_client_method(),    SSLv2_server_method(),
14818           SSLv3_method(),   SSLv3_client_method(),    SSLv3_server_method(),
14819           TLSv1_method(),   TLSv1_client_method(),    TLSv1_server_method(),
14820           ssl2_get_cipher_by_char(),
14821           ssl3_get_cipher_by_char().
14822
14823   *Patrick McCormick <patrick@tellme.com>, Bodo Moeller*
14824
14825 * Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
14826   the cached sessions are flushed, as the remove_cb() might use ex_data
14827   contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
14828   (see [openssl.org #212]).
14829
14830   *Geoff Thorpe, Lutz Jaenicke*
14831
14832 * Fix typo in OBJ_txt2obj which incorrectly passed the content
14833   length, instead of the encoding length to d2i_ASN1_OBJECT.
14834
14835   *Steve Henson*
14836
14837### Changes between 0.9.6f and 0.9.6g  [9 Aug 2002]
14838
14839 * [In 0.9.6g-engine release:]
14840   Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use `_stdcall`).
14841
14842   *Lynn Gazis <lgazis@rainbow.com>*
14843
14844### Changes between 0.9.6e and 0.9.6f  [8 Aug 2002]
14845
14846 * Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
14847   and get fix the header length calculation.
14848   *Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
14849   Alon Kantor <alonk@checkpoint.com> (and others), Steve Henson*
14850
14851 * Use proper error handling instead of 'assertions' in buffer
14852   overflow checks added in 0.9.6e.  This prevents DoS (the
14853   assertions could call abort()).
14854
14855   *Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller*
14856
14857### Changes between 0.9.6d and 0.9.6e  [30 Jul 2002]
14858
14859 * Add various sanity checks to asn1_get_length() to reject
14860   the ASN1 length bytes if they exceed sizeof(long), will appear
14861   negative or the content length exceeds the length of the
14862   supplied buffer.
14863
14864   *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>*
14865
14866 * Fix cipher selection routines: ciphers without encryption had no flags
14867   for the cipher strength set and where therefore not handled correctly
14868   by the selection routines (PR #130).
14869
14870   *Lutz Jaenicke*
14871
14872 * Fix EVP_dsa_sha macro.
14873
14874   *Nils Larsch*
14875
14876 * New option
14877        SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
14878   for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
14879   that was added in OpenSSL 0.9.6d.
14880
14881   As the countermeasure turned out to be incompatible with some
14882   broken SSL implementations, the new option is part of SSL_OP_ALL.
14883   SSL_OP_ALL is usually employed when compatibility with weird SSL
14884   implementations is desired (e.g. '-bugs' option to 's_client' and
14885   's_server'), so the new option is automatically set in many
14886   applications.
14887
14888   *Bodo Moeller*
14889
14890 * Changes in security patch:
14891
14892   Changes marked "(CHATS)" were sponsored by the Defense Advanced
14893   Research Projects Agency (DARPA) and Air Force Research Laboratory,
14894   Air Force Materiel Command, USAF, under agreement number
14895   F30602-01-2-0537.
14896
14897 * Add various sanity checks to asn1_get_length() to reject
14898   the ASN1 length bytes if they exceed sizeof(long), will appear
14899   negative or the content length exceeds the length of the
14900   supplied buffer. ([CVE-2002-0659])
14901
14902   *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>*
14903
14904 * Assertions for various potential buffer overflows, not known to
14905   happen in practice.
14906
14907   *Ben Laurie (CHATS)*
14908
14909 * Various temporary buffers to hold ASCII versions of integers were
14910   too small for 64 bit platforms. ([CVE-2002-0655])
14911   *Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>*
14912
14913 * Remote buffer overflow in SSL3 protocol - an attacker could
14914   supply an oversized session ID to a client. ([CVE-2002-0656])
14915
14916   *Ben Laurie (CHATS)*
14917
14918 * Remote buffer overflow in SSL2 protocol - an attacker could
14919   supply an oversized client master key. ([CVE-2002-0656])
14920
14921   *Ben Laurie (CHATS)*
14922
14923### Changes between 0.9.6c and 0.9.6d  [9 May 2002]
14924
14925 * Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
14926   encoded as NULL) with id-dsa-with-sha1.
14927
14928   *Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller*
14929
14930 * Check various `X509_...()` return values in `apps/req.c`.
14931
14932   *Nils Larsch <nla@trustcenter.de>*
14933
14934 * Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
14935   an end-of-file condition would erroneously be flagged, when the CRLF
14936   was just at the end of a processed block. The bug was discovered when
14937   processing data through a buffering memory BIO handing the data to a
14938   BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
14939   <ptsekov@syntrex.com> and Nedelcho Stanev.
14940
14941   *Lutz Jaenicke*
14942
14943 * Implement a countermeasure against a vulnerability recently found
14944   in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
14945   before application data chunks to avoid the use of known IVs
14946   with data potentially chosen by the attacker.
14947
14948   *Bodo Moeller*
14949
14950 * Fix length checks in ssl3_get_client_hello().
14951
14952   *Bodo Moeller*
14953
14954 * TLS/SSL library bugfix: use s->s3->in_read_app_data differently
14955   to prevent ssl3_read_internal() from incorrectly assuming that
14956   ssl3_read_bytes() found application data while handshake
14957   processing was enabled when in fact s->s3->in_read_app_data was
14958   merely automatically cleared during the initial handshake.
14959
14960   *Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>*
14961
14962 * Fix object definitions for Private and Enterprise: they were not
14963   recognized in their shortname (=lowercase) representation. Extend
14964   obj_dat.pl to issue an error when using undefined keywords instead
14965   of silently ignoring the problem (Svenning Sorensen
14966   <sss@sss.dnsalias.net>).
14967
14968   *Lutz Jaenicke*
14969
14970 * Fix DH_generate_parameters() so that it works for 'non-standard'
14971   generators, i.e. generators other than 2 and 5.  (Previously, the
14972   code did not properly initialise the 'add' and 'rem' values to
14973   BN_generate_prime().)
14974
14975   In the new general case, we do not insist that 'generator' is
14976   actually a primitive root: This requirement is rather pointless;
14977   a generator of the order-q subgroup is just as good, if not
14978   better.
14979
14980   *Bodo Moeller*
14981
14982 * Map new X509 verification errors to alerts. Discovered and submitted by
14983   Tom Wu <tom@arcot.com>.
14984
14985   *Lutz Jaenicke*
14986
14987 * Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
14988   returning non-zero before the data has been completely received
14989   when using non-blocking I/O.
14990
14991   *Bodo Moeller; problem pointed out by John Hughes*
14992
14993 * Some of the ciphers missed the strength entry (SSL_LOW etc).
14994
14995   *Ben Laurie, Lutz Jaenicke*
14996
14997 * Fix bug in SSL_clear(): bad sessions were not removed (found by
14998   Yoram Zahavi <YoramZ@gilian.com>).
14999
15000   *Lutz Jaenicke*
15001
15002 * Add information about CygWin 1.3 and on, and preserve proper
15003   configuration for the versions before that.
15004
15005   *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte*
15006
15007 * Make removal from session cache (SSL_CTX_remove_session()) more robust:
15008   check whether we deal with a copy of a session and do not delete from
15009   the cache in this case. Problem reported by "Izhar Shoshani Levi"
15010   <izhar@checkpoint.com>.
15011
15012   *Lutz Jaenicke*
15013
15014 * Do not store session data into the internal session cache, if it
15015   is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
15016   flag is set). Proposed by Aslam <aslam@funk.com>.
15017
15018   *Lutz Jaenicke*
15019
15020 * Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
15021   value is 0.
15022
15023   *Richard Levitte*
15024
15025 * [In 0.9.6d-engine release:]
15026   Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
15027
15028   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
15029
15030 * Add the configuration target linux-s390x.
15031
15032   *Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte*
15033
15034 * The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
15035   ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
15036   variable as an indication that a ClientHello message has been
15037   received.  As the flag value will be lost between multiple
15038   invocations of ssl3_accept when using non-blocking I/O, the
15039   function may not be aware that a handshake has actually taken
15040   place, thus preventing a new session from being added to the
15041   session cache.
15042
15043   To avoid this problem, we now set s->new_session to 2 instead of
15044   using a local variable.
15045
15046   *Lutz Jaenicke, Bodo Moeller*
15047
15048 * Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
15049   if the SSL_R_LENGTH_MISMATCH error is detected.
15050
15051   *Geoff Thorpe, Bodo Moeller*
15052
15053 * New 'shared_ldflag' column in Configure platform table.
15054
15055   *Richard Levitte*
15056
15057 * Fix EVP_CIPHER_mode macro.
15058
15059   *"Dan S. Camper" <dan@bti.net>*
15060
15061 * Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
15062   type, we must throw them away by setting rr->length to 0.
15063
15064   *D P Chang <dpc@qualys.com>*
15065
15066### Changes between 0.9.6b and 0.9.6c  [21 dec 2001]
15067
15068 * Fix BN_rand_range bug pointed out by Dominikus Scherkl
15069   <Dominikus.Scherkl@biodata.com>.  (The previous implementation
15070   worked incorrectly for those cases where range = `10..._2`  and
15071   `3*range`  is two bits longer than  range.)
15072
15073   *Bodo Moeller*
15074
15075 * Only add signing time to PKCS7 structures if it is not already
15076   present.
15077
15078   *Steve Henson*
15079
15080 * Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
15081   OBJ_ld_ce should be OBJ_id_ce.
15082   Also some ip-pda OIDs in crypto/objects/objects.txt were
15083   incorrect (cf. RFC 3039).
15084
15085   *Matt Cooper, Frederic Giudicelli, Bodo Moeller*
15086
15087 * Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
15088   returns early because it has nothing to do.
15089
15090   *Andy Schneider <andy.schneider@bjss.co.uk>*
15091
15092 * [In 0.9.6c-engine release:]
15093   Fix mutex callback return values in crypto/engine/hw_ncipher.c.
15094
15095   *Andy Schneider <andy.schneider@bjss.co.uk>*
15096
15097 * [In 0.9.6c-engine release:]
15098   Add support for Cryptographic Appliance's keyserver technology.
15099   (Use engine 'keyclient')
15100
15101   *Cryptographic Appliances and Geoff Thorpe*
15102
15103 * Add a configuration entry for OS/390 Unix.  The C compiler 'c89'
15104   is called via tools/c89.sh because arguments have to be
15105   rearranged (all '-L' options must appear before the first object
15106   modules).
15107
15108   *Richard Shapiro <rshapiro@abinitio.com>*
15109
15110 * [In 0.9.6c-engine release:]
15111   Add support for Broadcom crypto accelerator cards, backported
15112   from 0.9.7.
15113
15114   *Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox*
15115
15116 * [In 0.9.6c-engine release:]
15117   Add support for SureWare crypto accelerator cards from
15118   Baltimore Technologies.  (Use engine 'sureware')
15119
15120   *Baltimore Technologies and Mark Cox*
15121
15122 * [In 0.9.6c-engine release:]
15123   Add support for crypto accelerator cards from Accelerated
15124   Encryption Processing, www.aep.ie.  (Use engine 'aep')
15125
15126   *AEP Inc. and Mark Cox*
15127
15128 * Add a configuration entry for gcc on UnixWare.
15129
15130   *Gary Benson <gbenson@redhat.com>*
15131
15132 * Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
15133   messages are stored in a single piece (fixed-length part and
15134   variable-length part combined) and fix various bugs found on the way.
15135
15136   *Bodo Moeller*
15137
15138 * Disable caching in BIO_gethostbyname(), directly use gethostbyname()
15139   instead.  BIO_gethostbyname() does not know what timeouts are
15140   appropriate, so entries would stay in cache even when they have
15141   become invalid.
15142   *Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>*
15143
15144 * Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
15145   faced with a pathologically small ClientHello fragment that does
15146   not contain client_version: Instead of aborting with an error,
15147   simply choose the highest available protocol version (i.e.,
15148   TLS 1.0 unless it is disabled).  In practice, ClientHello
15149   messages are never sent like this, but this change gives us
15150   strictly correct behaviour at least for TLS.
15151
15152   *Bodo Moeller*
15153
15154 * Fix SSL handshake functions and SSL_clear() such that SSL_clear()
15155   never resets s->method to s->ctx->method when called from within
15156   one of the SSL handshake functions.
15157
15158   *Bodo Moeller; problem pointed out by Niko Baric*
15159
15160 * In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
15161   (sent using the client's version number) if client_version is
15162   smaller than the protocol version in use.  Also change
15163   ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
15164   the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
15165   the client will at least see that alert.
15166
15167   *Bodo Moeller*
15168
15169 * Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
15170   correctly.
15171
15172   *Bodo Moeller*
15173
15174 * Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
15175   client receives HelloRequest while in a handshake.
15176
15177   *Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>*
15178
15179 * Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
15180   should end in 'break', not 'goto end' which circumvents various
15181   cleanups done in state SSL_ST_OK.   But session related stuff
15182   must be disabled for SSL_ST_OK in the case that we just sent a
15183   HelloRequest.
15184
15185   Also avoid some overhead by not calling ssl_init_wbio_buffer()
15186   before just sending a HelloRequest.
15187
15188   *Bodo Moeller, Eric Rescorla <ekr@rtfm.com>*
15189
15190 * Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
15191   reveal whether illegal block cipher padding was found or a MAC
15192   verification error occurred.  (Neither SSLerr() codes nor alerts
15193   are directly visible to potential attackers, but the information
15194   may leak via logfiles.)
15195
15196   Similar changes are not required for the SSL 2.0 implementation
15197   because the number of padding bytes is sent in clear for SSL 2.0,
15198   and the extra bytes are just ignored.  However ssl/s2_pkt.c
15199   failed to verify that the purported number of padding bytes is in
15200   the legal range.
15201
15202   *Bodo Moeller*
15203
15204 * Add OpenUNIX-8 support including shared libraries
15205   (Boyd Lynn Gerber <gerberb@zenez.com>).
15206
15207   *Lutz Jaenicke*
15208
15209 * Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
15210   'wristwatch attack' using huge encoding parameters (cf.
15211   James H. Manger's CRYPTO 2001 paper).  Note that the
15212   RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
15213   encoding parameters and hence was not vulnerable.
15214
15215   *Bodo Moeller*
15216
15217 * BN_sqr() bug fix.
15218
15219   *Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>*
15220
15221 * Rabin-Miller test analyses assume uniformly distributed witnesses,
15222   so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
15223   followed by modular reduction.
15224
15225   *Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>*
15226
15227 * Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
15228   equivalent based on BN_pseudo_rand() instead of BN_rand().
15229
15230   *Bodo Moeller*
15231
15232 * s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
15233   This function was broken, as the check for a new client hello message
15234   to handle SGC did not allow these large messages.
15235   (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
15236
15237   *Lutz Jaenicke*
15238
15239 * Add alert descriptions for TLSv1 to `SSL_alert_desc_string[_long]()`.
15240
15241   *Lutz Jaenicke*
15242
15243 * Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
15244   for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
15245
15246   *Lutz Jaenicke*
15247
15248 * Rework the configuration and shared library support for Tru64 Unix.
15249   The configuration part makes use of modern compiler features and
15250   still retains old compiler behavior for those that run older versions
15251   of the OS.  The shared library support part includes a variant that
15252   uses the RPATH feature, and is available through the special
15253   configuration target "alpha-cc-rpath", which will never be selected
15254   automatically.
15255
15256   *Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte*
15257
15258 * In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
15259   with the same message size as in ssl3_get_certificate_request().
15260   Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
15261   messages might inadvertently be reject as too long.
15262
15263   *Petr Lampa <lampa@fee.vutbr.cz>*
15264
15265 * Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
15266
15267   *Andy Polyakov*
15268
15269 * Modified SSL library such that the verify_callback that has been set
15270   specifically for an SSL object with SSL_set_verify() is actually being
15271   used. Before the change, a verify_callback set with this function was
15272   ignored and the verify_callback() set in the SSL_CTX at the time of
15273   the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
15274   to allow the necessary settings.
15275
15276   *Lutz Jaenicke*
15277
15278 * Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
15279   explicitly to NULL, as at least on Solaris 8 this seems not always to be
15280   done automatically (in contradiction to the requirements of the C
15281   standard). This made problems when used from OpenSSH.
15282
15283   *Lutz Jaenicke*
15284
15285 * In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
15286   dh->length and always used
15287
15288           BN_rand_range(priv_key, dh->p).
15289
15290   BN_rand_range() is not necessary for Diffie-Hellman, and this
15291   specific range makes Diffie-Hellman unnecessarily inefficient if
15292   dh->length (recommended exponent length) is much smaller than the
15293   length of dh->p.  We could use BN_rand_range() if the order of
15294   the subgroup was stored in the DH structure, but we only have
15295   dh->length.
15296
15297   So switch back to
15298
15299           BN_rand(priv_key, l, ...)
15300
15301   where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
15302   otherwise.
15303
15304   *Bodo Moeller*
15305
15306 * In
15307
15308           RSA_eay_public_encrypt
15309           RSA_eay_private_decrypt
15310           RSA_eay_private_encrypt (signing)
15311           RSA_eay_public_decrypt (signature verification)
15312
15313   (default implementations for RSA_public_encrypt,
15314   RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
15315   always reject numbers >= n.
15316
15317   *Bodo Moeller*
15318
15319 * In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
15320   to synchronize access to 'locking_thread'.  This is necessary on
15321   systems where access to 'locking_thread' (an 'unsigned long'
15322   variable) is not atomic.
15323
15324   *Bodo Moeller*
15325
15326 * In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
15327   *before* setting the 'crypto_lock_rand' flag.  The previous code had
15328   a race condition if 0 is a valid thread ID.
15329
15330   *Travis Vitek <vitek@roguewave.com>*
15331
15332 * Add support for shared libraries under Irix.
15333
15334   *Albert Chin-A-Young <china@thewrittenword.com>*
15335
15336 * Add configuration option to build on Linux on both big-endian and
15337   little-endian MIPS.
15338
15339   *Ralf Baechle <ralf@uni-koblenz.de>*
15340
15341 * Add the possibility to create shared libraries on HP-UX.
15342
15343   *Richard Levitte*
15344
15345### Changes between 0.9.6a and 0.9.6b  [9 Jul 2001]
15346
15347 * Change ssleay_rand_bytes (crypto/rand/md_rand.c)
15348   to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
15349   Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
15350   PRNG state recovery was possible based on the output of
15351   one PRNG request appropriately sized to gain knowledge on
15352   'md' followed by enough consecutive 1-byte PRNG requests
15353   to traverse all of 'state'.
15354
15355   1. When updating 'md_local' (the current thread's copy of 'md')
15356      during PRNG output generation, hash all of the previous
15357      'md_local' value, not just the half used for PRNG output.
15358
15359   2. Make the number of bytes from 'state' included into the hash
15360      independent from the number of PRNG bytes requested.
15361
15362   The first measure alone would be sufficient to avoid
15363   Markku-Juhani's attack.  (Actually it had never occurred
15364   to me that the half of 'md_local' used for chaining was the
15365   half from which PRNG output bytes were taken -- I had always
15366   assumed that the secret half would be used.)  The second
15367   measure makes sure that additional data from 'state' is never
15368   mixed into 'md_local' in small portions; this heuristically
15369   further strengthens the PRNG.
15370
15371   *Bodo Moeller*
15372
15373 * Fix crypto/bn/asm/mips3.s.
15374
15375   *Andy Polyakov*
15376
15377 * When only the key is given to "enc", the IV is undefined. Print out
15378   an error message in this case.
15379
15380   *Lutz Jaenicke*
15381
15382 * Handle special case when X509_NAME is empty in X509 printing routines.
15383
15384   *Steve Henson*
15385
15386 * In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
15387   positive and less than q.
15388
15389   *Bodo Moeller*
15390
15391 * Don't change `*pointer` in CRYPTO_add_lock() is add_lock_callback is
15392   used: it isn't thread safe and the add_lock_callback should handle
15393   that itself.
15394
15395   *Paul Rose <Paul.Rose@bridge.com>*
15396
15397 * Verify that incoming data obeys the block size in
15398   ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
15399
15400   *Bodo Moeller*
15401
15402 * Fix OAEP check.
15403
15404   *Ulf Möller, Bodo Möller*
15405
15406 * The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
15407   RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
15408   when fixing the server behaviour for backwards-compatible 'client
15409   hello' messages.  (Note that the attack is impractical against
15410   SSL 3.0 and TLS 1.0 anyway because length and version checking
15411   means that the probability of guessing a valid ciphertext is
15412   around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
15413   paper.)
15414
15415   Before 0.9.5, the countermeasure (hide the error by generating a
15416   random 'decryption result') did not work properly because
15417   ERR_clear_error() was missing, meaning that SSL_get_error() would
15418   detect the supposedly ignored error.
15419
15420   Both problems are now fixed.
15421
15422   *Bodo Moeller*
15423
15424 * In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
15425   (previously it was 1024).
15426
15427   *Bodo Moeller*
15428
15429 * Fix for compatibility mode trust settings: ignore trust settings
15430   unless some valid trust or reject settings are present.
15431
15432   *Steve Henson*
15433
15434 * Fix for blowfish EVP: its a variable length cipher.
15435
15436   *Steve Henson*
15437
15438 * Fix various bugs related to DSA S/MIME verification. Handle missing
15439   parameters in DSA public key structures and return an error in the
15440   DSA routines if parameters are absent.
15441
15442   *Steve Henson*
15443
15444 * In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
15445   in the current directory if neither $RANDFILE nor $HOME was set.
15446   RAND_file_name() in 0.9.6a returned NULL in this case.  This has
15447   caused some confusion to Windows users who haven't defined $HOME.
15448   Thus RAND_file_name() is changed again: e_os.h can define a
15449   DEFAULT_HOME, which will be used if $HOME is not set.
15450   For Windows, we use "C:"; on other platforms, we still require
15451   environment variables.
15452
15453 * Move 'if (!initialized) RAND_poll()' into regions protected by
15454   CRYPTO_LOCK_RAND.  This is not strictly necessary, but avoids
15455   having multiple threads call RAND_poll() concurrently.
15456
15457   *Bodo Moeller*
15458
15459 * In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
15460   combination of a flag and a thread ID variable.
15461   Otherwise while one thread is in ssleay_rand_bytes (which sets the
15462   flag), *other* threads can enter ssleay_add_bytes without obeying
15463   the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
15464   that they do not hold after the first thread unsets add_do_not_lock).
15465
15466   *Bodo Moeller*
15467
15468 * Change bctest again: '-x' expressions are not available in all
15469   versions of 'test'.
15470
15471   *Bodo Moeller*
15472
15473### Changes between 0.9.6 and 0.9.6a  [5 Apr 2001]
15474
15475 * Fix a couple of memory leaks in PKCS7_dataDecode()
15476
15477   *Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>*
15478
15479 * Change Configure and Makefiles to provide EXE_EXT, which will contain
15480   the default extension for executables, if any.  Also, make the perl
15481   scripts that use symlink() to test if it really exists and use "cp"
15482   if it doesn't.  All this made OpenSSL compilable and installable in
15483   CygWin.
15484
15485   *Richard Levitte*
15486
15487 * Fix for asn1_GetSequence() for indefinite length constructed data.
15488   If SEQUENCE is length is indefinite just set c->slen to the total
15489   amount of data available.
15490
15491   *Steve Henson, reported by shige@FreeBSD.org*
15492
15493   *This change does not apply to 0.9.7.*
15494
15495 * Change bctest to avoid here-documents inside command substitution
15496   (workaround for FreeBSD /bin/sh bug).
15497   For compatibility with Ultrix, avoid shell functions (introduced
15498   in the bctest version that searches along $PATH).
15499
15500   *Bodo Moeller*
15501
15502 * Rename 'des_encrypt' to 'des_encrypt1'.  This avoids the clashes
15503   with des_encrypt() defined on some operating systems, like Solaris
15504   and UnixWare.
15505
15506   *Richard Levitte*
15507
15508 * Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
15509   On the Importance of Eliminating Errors in Cryptographic
15510   Computations, J. Cryptology 14 (2001) 2, 101-119,
15511   <http://theory.stanford.edu/~dabo/papers/faults.ps.gz>).
15512
15513   *Ulf Moeller*
15514
15515 * MIPS assembler BIGNUM division bug fix.
15516
15517   *Andy Polyakov*
15518
15519 * Disabled incorrect Alpha assembler code.
15520
15521   *Richard Levitte*
15522
15523 * Fix PKCS#7 decode routines so they correctly update the length
15524   after reading an EOC for the EXPLICIT tag.
15525
15526   *Steve Henson*
15527
15528   *This change does not apply to 0.9.7.*
15529
15530 * Fix bug in PKCS#12 key generation routines. This was triggered
15531   if a 3DES key was generated with a 0 initial byte. Include
15532   PKCS12_BROKEN_KEYGEN compilation option to retain the old
15533   (but broken) behaviour.
15534
15535   *Steve Henson*
15536
15537 * Enhance bctest to search for a working bc along $PATH and print
15538   it when found.
15539
15540   *Tim Rice <tim@multitalents.net> via Richard Levitte*
15541
15542 * Fix memory leaks in err.c: free err_data string if necessary;
15543   don't write to the wrong index in ERR_set_error_data.
15544
15545   *Bodo Moeller*
15546
15547 * Implement ssl23_peek (analogous to ssl23_read), which previously
15548   did not exist.
15549
15550   *Bodo Moeller*
15551
15552 * Replace rdtsc with `_emit` statements for VC++ version 5.
15553
15554   *Jeremy Cooper <jeremy@baymoo.org>*
15555
15556 * Make it possible to reuse SSLv2 sessions.
15557
15558   *Richard Levitte*
15559
15560 * In copy_email() check for >= 0 as a return value for
15561   X509_NAME_get_index_by_NID() since 0 is a valid index.
15562
15563   *Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>*
15564
15565 * Avoid coredump with unsupported or invalid public keys by checking if
15566   X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
15567   PKCS7_verify() fails with non detached data.
15568
15569   *Steve Henson*
15570
15571 * Don't use getenv in library functions when run as setuid/setgid.
15572   New function OPENSSL_issetugid().
15573
15574   *Ulf Moeller*
15575
15576 * Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
15577   due to incorrect handling of multi-threading:
15578
15579   1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
15580
15581   2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
15582
15583   3. Count how many times MemCheck_off() has been called so that
15584      nested use can be treated correctly.  This also avoids
15585      inband-signalling in the previous code (which relied on the
15586      assumption that thread ID 0 is impossible).
15587
15588   *Bodo Moeller*
15589
15590 * Add "-rand" option also to s_client and s_server.
15591
15592   *Lutz Jaenicke*
15593
15594 * Fix CPU detection on Irix 6.x.
15595   *Kurt Hockenbury <khockenb@stevens-tech.edu> and
15596   "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>*
15597
15598 * Fix X509_NAME bug which produced incorrect encoding if X509_NAME
15599   was empty.
15600
15601   *Steve Henson*
15602
15603   *This change does not apply to 0.9.7.*
15604
15605 * Use the cached encoding of an X509_NAME structure rather than
15606   copying it. This is apparently the reason for the libsafe "errors"
15607   but the code is actually correct.
15608
15609   *Steve Henson*
15610
15611 * Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
15612   Bleichenbacher's DSA attack.
15613   Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
15614   to be set and top=0 forces the highest bit to be set; top=-1 is new
15615   and leaves the highest bit random.
15616
15617   *Ulf Moeller, Bodo Moeller*
15618
15619 * In the `NCONF_...`-based implementations for `CONF_...` queries
15620   (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
15621   a temporary CONF structure with the data component set to NULL
15622   (which gives segmentation faults in lh_retrieve).
15623   Instead, use NULL for the CONF pointer in CONF_get_string and
15624   CONF_get_number (which may use environment variables) and directly
15625   return NULL from CONF_get_section.
15626
15627   *Bodo Moeller*
15628
15629 * Fix potential buffer overrun for EBCDIC.
15630
15631   *Ulf Moeller*
15632
15633 * Tolerate nonRepudiation as being valid for S/MIME signing and certSign
15634   keyUsage if basicConstraints absent for a CA.
15635
15636   *Steve Henson*
15637
15638 * Make SMIME_write_PKCS7() write mail header values with a format that
15639   is more generally accepted (no spaces before the semicolon), since
15640   some programs can't parse those values properly otherwise.  Also make
15641   sure BIO's that break lines after each write do not create invalid
15642   headers.
15643
15644   *Richard Levitte*
15645
15646 * Make the CRL encoding routines work with empty SEQUENCE OF. The
15647   macros previously used would not encode an empty SEQUENCE OF
15648   and break the signature.
15649
15650   *Steve Henson*
15651
15652   *This change does not apply to 0.9.7.*
15653
15654 * Zero the premaster secret after deriving the master secret in
15655   DH ciphersuites.
15656
15657   *Steve Henson*
15658
15659 * Add some EVP_add_digest_alias registrations (as found in
15660   OpenSSL_add_all_digests()) to SSL_library_init()
15661   aka OpenSSL_add_ssl_algorithms().  This provides improved
15662   compatibility with peers using X.509 certificates
15663   with unconventional AlgorithmIdentifier OIDs.
15664
15665   *Bodo Moeller*
15666
15667 * Fix for Irix with NO_ASM.
15668
15669   *"Bruce W. Forsberg" <bruce.forsberg@baesystems.com>*
15670
15671 * ./config script fixes.
15672
15673   *Ulf Moeller, Richard Levitte*
15674
15675 * Fix 'openssl passwd -1'.
15676
15677   *Bodo Moeller*
15678
15679 * Change PKCS12_key_gen_asc() so it can cope with non null
15680   terminated strings whose length is passed in the passlen
15681   parameter, for example from PEM callbacks. This was done
15682   by adding an extra length parameter to asc2uni().
15683
15684   *Steve Henson, reported by <oddissey@samsung.co.kr>*
15685
15686 * Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
15687   call failed, free the DSA structure.
15688
15689   *Bodo Moeller*
15690
15691 * Fix to uni2asc() to cope with zero length Unicode strings.
15692   These are present in some PKCS#12 files.
15693
15694   *Steve Henson*
15695
15696 * Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
15697   Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
15698   when writing a 32767 byte record.
15699
15700   *Bodo Moeller; problem reported by Eric Day <eday@concentric.net>*
15701
15702 * In `RSA_eay_public_{en,ed}crypt` and RSA_eay_mod_exp (rsa_eay.c),
15703   obtain lock CRYPTO_LOCK_RSA before setting `rsa->_method_mod_{n,p,q}`.
15704
15705   (RSA objects have a reference count access to which is protected
15706   by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
15707   so they are meant to be shared between threads.)
15708   *Bodo Moeller, Geoff Thorpe; original patch submitted by
15709   "Reddie, Steven" <Steven.Reddie@ca.com>*
15710
15711 * Fix a deadlock in CRYPTO_mem_leaks().
15712
15713   *Bodo Moeller*
15714
15715 * Use better test patterns in bntest.
15716
15717   *Ulf Möller*
15718
15719 * rand_win.c fix for Borland C.
15720
15721   *Ulf Möller*
15722
15723 * BN_rshift bugfix for n == 0.
15724
15725   *Bodo Moeller*
15726
15727 * Add a 'bctest' script that checks for some known 'bc' bugs
15728   so that 'make test' does not abort just because 'bc' is broken.
15729
15730   *Bodo Moeller*
15731
15732 * Store verify_result within SSL_SESSION also for client side to
15733   avoid potential security hole. (Re-used sessions on the client side
15734   always resulted in verify_result==X509_V_OK, not using the original
15735   result of the server certificate verification.)
15736
15737   *Lutz Jaenicke*
15738
15739 * Fix ssl3_pending: If the record in s->s3->rrec is not of type
15740   SSL3_RT_APPLICATION_DATA, return 0.
15741   Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
15742
15743   *Bodo Moeller*
15744
15745 * Fix SSL_peek:
15746   Both ssl2_peek and ssl3_peek, which were totally broken in earlier
15747   releases, have been re-implemented by renaming the previous
15748   implementations of ssl2_read and ssl3_read to ssl2_read_internal
15749   and ssl3_read_internal, respectively, and adding 'peek' parameters
15750   to them.  The new ssl[23]_{read,peek} functions are calls to
15751   ssl[23]_read_internal with the 'peek' flag set appropriately.
15752   A 'peek' parameter has also been added to ssl3_read_bytes, which
15753   does the actual work for ssl3_read_internal.
15754
15755   *Bodo Moeller*
15756
15757 * Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
15758   the method-specific "init()" handler. Also clean up ex_data after
15759   calling the method-specific "finish()" handler. Previously, this was
15760   happening the other way round.
15761
15762   *Geoff Thorpe*
15763
15764 * Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
15765   The previous value, 12, was not always sufficient for BN_mod_exp().
15766
15767   *Bodo Moeller*
15768
15769 * Make sure that shared libraries get the internal name engine with
15770   the full version number and not just 0.  This should mark the
15771   shared libraries as not backward compatible.  Of course, this should
15772   be changed again when we can guarantee backward binary compatibility.
15773
15774   *Richard Levitte*
15775
15776 * Fix typo in get_cert_by_subject() in by_dir.c
15777
15778   *Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>*
15779
15780 * Rework the system to generate shared libraries:
15781
15782   - Make note of the expected extension for the shared libraries and
15783     if there is a need for symbolic links from for example libcrypto.so.0
15784     to libcrypto.so.0.9.7.  There is extended info in Configure for
15785     that.
15786
15787   - Make as few rebuilds of the shared libraries as possible.
15788
15789   - Still avoid linking the OpenSSL programs with the shared libraries.
15790
15791   - When installing, install the shared libraries separately from the
15792     static ones.
15793
15794   *Richard Levitte*
15795
15796 * Fix SSL_CTX_set_read_ahead macro to actually use its argument.
15797
15798   Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
15799   and not in SSL_clear because the latter is also used by the
15800   accept/connect functions; previously, the settings made by
15801   SSL_set_read_ahead would be lost during the handshake.
15802
15803   *Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>*
15804
15805 * Correct util/mkdef.pl to be selective about disabled algorithms.
15806   Previously, it would create entries for disabled algorithms no
15807   matter what.
15808
15809   *Richard Levitte*
15810
15811 * Added several new manual pages for SSL_* function.
15812
15813   *Lutz Jaenicke*
15814
15815### Changes between 0.9.5a and 0.9.6  [24 Sep 2000]
15816
15817 * In ssl23_get_client_hello, generate an error message when faced
15818   with an initial SSL 3.0/TLS record that is too small to contain the
15819   first two bytes of the ClientHello message, i.e. client_version.
15820   (Note that this is a pathologic case that probably has never happened
15821   in real life.)  The previous approach was to use the version number
15822   from the record header as a substitute; but our protocol choice
15823   should not depend on that one because it is not authenticated
15824   by the Finished messages.
15825
15826   *Bodo Moeller*
15827
15828 * More robust randomness gathering functions for Windows.
15829
15830   *Jeffrey Altman <jaltman@columbia.edu>*
15831
15832 * For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
15833   not set then we don't setup the error code for issuer check errors
15834   to avoid possibly overwriting other errors which the callback does
15835   handle. If an application does set the flag then we assume it knows
15836   what it is doing and can handle the new informational codes
15837   appropriately.
15838
15839   *Steve Henson*
15840
15841 * Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
15842   a general "ANY" type, as such it should be able to decode anything
15843   including tagged types. However it didn't check the class so it would
15844   wrongly interpret tagged types in the same way as their universal
15845   counterpart and unknown types were just rejected. Changed so that the
15846   tagged and unknown types are handled in the same way as a SEQUENCE:
15847   that is the encoding is stored intact. There is also a new type
15848   "V_ASN1_OTHER" which is used when the class is not universal, in this
15849   case we have no idea what the actual type is so we just lump them all
15850   together.
15851
15852   *Steve Henson*
15853
15854 * On VMS, stdout may very well lead to a file that is written to
15855   in a record-oriented fashion.  That means that every write() will
15856   write a separate record, which will be read separately by the
15857   programs trying to read from it.  This can be very confusing.
15858
15859   The solution is to put a BIO filter in the way that will buffer
15860   text until a linefeed is reached, and then write everything a
15861   line at a time, so every record written will be an actual line,
15862   not chunks of lines and not (usually doesn't happen, but I've
15863   seen it once) several lines in one record.  BIO_f_linebuffer() is
15864   the answer.
15865
15866   Currently, it's a VMS-only method, because that's where it has
15867   been tested well enough.
15868
15869   *Richard Levitte*
15870
15871 * Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
15872   it can return incorrect results.
15873   (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
15874   but it was in 0.9.6-beta[12].)
15875
15876   *Bodo Moeller*
15877
15878 * Disable the check for content being present when verifying detached
15879   signatures in pk7_smime.c. Some versions of Netscape (wrongly)
15880   include zero length content when signing messages.
15881
15882   *Steve Henson*
15883
15884 * New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
15885   BIO_ctrl (for BIO pairs).
15886
15887   *Bodo Möller*
15888
15889 * Add DSO method for VMS.
15890
15891   *Richard Levitte*
15892
15893 * Bug fix: Montgomery multiplication could produce results with the
15894   wrong sign.
15895
15896   *Ulf Möller*
15897
15898 * Add RPM specification openssl.spec and modify it to build three
15899   packages.  The default package contains applications, application
15900   documentation and run-time libraries.  The devel package contains
15901   include files, static libraries and function documentation.  The
15902   doc package contains the contents of the doc directory.  The original
15903   openssl.spec was provided by Damien Miller <djm@mindrot.org>.
15904
15905   *Richard Levitte*
15906
15907 * Add a large number of documentation files for many SSL routines.
15908
15909   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>*
15910
15911 * Add a configuration entry for Sony News 4.
15912
15913   *NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>*
15914
15915 * Don't set the two most significant bits to one when generating a
15916   random number < q in the DSA library.
15917
15918   *Ulf Möller*
15919
15920 * New SSL API mode 'SSL_MODE_AUTO_RETRY'.  This disables the default
15921   behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
15922   the underlying transport is blocking) if a handshake took place.
15923   (The default behaviour is needed by applications such as s_client
15924   and s_server that use select() to determine when to use SSL_read;
15925   but for applications that know in advance when to expect data, it
15926   just makes things more complicated.)
15927
15928   *Bodo Moeller*
15929
15930 * Add RAND_egd_bytes(), which gives control over the number of bytes read
15931   from EGD.
15932
15933   *Ben Laurie*
15934
15935 * Add a few more EBCDIC conditionals that make `req` and `x509`
15936   work better on such systems.
15937
15938   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
15939
15940 * Add two demo programs for PKCS12_parse() and PKCS12_create().
15941   Update PKCS12_parse() so it copies the friendlyName and the
15942   keyid to the certificates aux info.
15943
15944   *Steve Henson*
15945
15946 * Fix bug in PKCS7_verify() which caused an infinite loop
15947   if there was more than one signature.
15948
15949   *Sven Uszpelkat <su@celocom.de>*
15950
15951 * Major change in util/mkdef.pl to include extra information
15952   about each symbol, as well as presenting variables as well
15953   as functions.  This change means that there's n more need
15954   to rebuild the .num files when some algorithms are excluded.
15955
15956   *Richard Levitte*
15957
15958 * Allow the verify time to be set by an application,
15959   rather than always using the current time.
15960
15961   *Steve Henson*
15962
15963 * Phase 2 verify code reorganisation. The certificate
15964   verify code now looks up an issuer certificate by a
15965   number of criteria: subject name, authority key id
15966   and key usage. It also verifies self signed certificates
15967   by the same criteria. The main comparison function is
15968   X509_check_issued() which performs these checks.
15969
15970   Lot of changes were necessary in order to support this
15971   without completely rewriting the lookup code.
15972
15973   Authority and subject key identifier are now cached.
15974
15975   The LHASH 'certs' is X509_STORE has now been replaced
15976   by a STACK_OF(X509_OBJECT). This is mainly because an
15977   LHASH can't store or retrieve multiple objects with
15978   the same hash value.
15979
15980   As a result various functions (which were all internal
15981   use only) have changed to handle the new X509_STORE
15982   structure. This will break anything that messed round
15983   with X509_STORE internally.
15984
15985   The functions X509_STORE_add_cert() now checks for an
15986   exact match, rather than just subject name.
15987
15988   The X509_STORE API doesn't directly support the retrieval
15989   of multiple certificates matching a given criteria, however
15990   this can be worked round by performing a lookup first
15991   (which will fill the cache with candidate certificates)
15992   and then examining the cache for matches. This is probably
15993   the best we can do without throwing out X509_LOOKUP
15994   entirely (maybe later...).
15995
15996   The X509_VERIFY_CTX structure has been enhanced considerably.
15997
15998   All certificate lookup operations now go via a get_issuer()
15999   callback. Although this currently uses an X509_STORE it
16000   can be replaced by custom lookups. This is a simple way
16001   to bypass the X509_STORE hackery necessary to make this
16002   work and makes it possible to use more efficient techniques
16003   in future. A very simple version which uses a simple
16004   STACK for its trusted certificate store is also provided
16005   using X509_STORE_CTX_trusted_stack().
16006
16007   The verify_cb() and verify() callbacks now have equivalents
16008   in the X509_STORE_CTX structure.
16009
16010   X509_STORE_CTX also has a 'flags' field which can be used
16011   to customise the verify behaviour.
16012
16013   *Steve Henson*
16014
16015 * Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
16016   excludes S/MIME capabilities.
16017
16018   *Steve Henson*
16019
16020 * When a certificate request is read in keep a copy of the
16021   original encoding of the signed data and use it when outputting
16022   again. Signatures then use the original encoding rather than
16023   a decoded, encoded version which may cause problems if the
16024   request is improperly encoded.
16025
16026   *Steve Henson*
16027
16028 * For consistency with other BIO_puts implementations, call
16029   buffer_write(b, ...) directly in buffer_puts instead of calling
16030   BIO_write(b, ...).
16031
16032   In BIO_puts, increment b->num_write as in BIO_write.
16033
16034   *Peter.Sylvester@EdelWeb.fr*
16035
16036 * Fix BN_mul_word for the case where the word is 0. (We have to use
16037   BN_zero, we may not return a BIGNUM with an array consisting of
16038   words set to zero.)
16039
16040   *Bodo Moeller*
16041
16042 * Avoid calling abort() from within the library when problems are
16043   detected, except if preprocessor symbols have been defined
16044   (such as REF_CHECK, BN_DEBUG etc.).
16045
16046   *Bodo Moeller*
16047
16048 * New openssl application 'rsautl'. This utility can be
16049   used for low-level RSA operations. DER public key
16050   BIO/fp routines also added.
16051
16052   *Steve Henson*
16053
16054 * New Configure entry and patches for compiling on QNX 4.
16055
16056   *Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>*
16057
16058 * A demo state-machine implementation was sponsored by
16059   Nuron (<http://www.nuron.com/>) and is now available in
16060   demos/state_machine.
16061
16062   *Ben Laurie*
16063
16064 * New options added to the 'dgst' utility for signature
16065   generation and verification.
16066
16067   *Steve Henson*
16068
16069 * Unrecognized PKCS#7 content types are now handled via a
16070   catch all ASN1_TYPE structure. This allows unsupported
16071   types to be stored as a "blob" and an application can
16072   encode and decode it manually.
16073
16074   *Steve Henson*
16075
16076 * Fix various signed/unsigned issues to make a_strex.c
16077   compile under VC++.
16078
16079   *Oscar Jacobsson <oscar.jacobsson@celocom.com>*
16080
16081 * ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
16082   length if passed a buffer. ASN1_INTEGER_to_BN failed
16083   if passed a NULL BN and its argument was negative.
16084
16085   *Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>*
16086
16087 * Modification to PKCS#7 encoding routines to output definite
16088   length encoding. Since currently the whole structures are in
16089   memory there's not real point in using indefinite length
16090   constructed encoding. However if OpenSSL is compiled with
16091   the flag PKCS7_INDEFINITE_ENCODING the old form is used.
16092
16093   *Steve Henson*
16094
16095 * Added BIO_vprintf() and BIO_vsnprintf().
16096
16097   *Richard Levitte*
16098
16099 * Added more prefixes to parse for in the strings written
16100   through a logging bio, to cover all the levels that are available
16101   through syslog.  The prefixes are now:
16102
16103           PANIC, EMERG, EMR       =>      LOG_EMERG
16104           ALERT, ALR              =>      LOG_ALERT
16105           CRIT, CRI               =>      LOG_CRIT
16106           ERROR, ERR              =>      LOG_ERR
16107           WARNING, WARN, WAR      =>      LOG_WARNING
16108           NOTICE, NOTE, NOT       =>      LOG_NOTICE
16109           INFO, INF               =>      LOG_INFO
16110           DEBUG, DBG              =>      LOG_DEBUG
16111
16112   and as before, if none of those prefixes are present at the
16113   beginning of the string, LOG_ERR is chosen.
16114
16115   On Win32, the `LOG_*` levels are mapped according to this:
16116
16117           LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
16118           LOG_WARNING                             => EVENTLOG_WARNING_TYPE
16119           LOG_NOTICE, LOG_INFO, LOG_DEBUG         => EVENTLOG_INFORMATION_TYPE
16120
16121   *Richard Levitte*
16122
16123 * Made it possible to reconfigure with just the configuration
16124   argument "reconf" or "reconfigure".  The command line arguments
16125   are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
16126   and are retrieved from there when reconfiguring.
16127
16128   *Richard Levitte*
16129
16130 * MD4 implemented.
16131
16132   *Assar Westerlund <assar@sics.se>, Richard Levitte*
16133
16134 * Add the arguments -CAfile and -CApath to the pkcs12 utility.
16135
16136   *Richard Levitte*
16137
16138 * The obj_dat.pl script was messing up the sorting of object
16139   names. The reason was that it compared the quoted version
16140   of strings as a result "OCSP" > "OCSP Signing" because
16141   " > SPACE. Changed script to store unquoted versions of
16142   names and add quotes on output. It was also omitting some
16143   names from the lookup table if they were given a default
16144   value (that is if SN is missing it is given the same
16145   value as LN and vice versa), these are now added on the
16146   grounds that if an object has a name we should be able to
16147   look it up. Finally added warning output when duplicate
16148   short or long names are found.
16149
16150   *Steve Henson*
16151
16152 * Changes needed for Tandem NSK.
16153
16154   *Scott Uroff <scott@xypro.com>*
16155
16156 * Fix SSL 2.0 rollback checking: Due to an off-by-one error in
16157   RSA_padding_check_SSLv23(), special padding was never detected
16158   and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
16159   version rollback attacks was not effective.
16160
16161   In s23_clnt.c, don't use special rollback-attack detection padding
16162   (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
16163   client; similarly, in s23_srvr.c, don't do the rollback check if
16164   SSL 2.0 is the only protocol enabled in the server.
16165
16166   *Bodo Moeller*
16167
16168 * Make it possible to get hexdumps of unprintable data with 'openssl
16169   asn1parse'.  By implication, the functions ASN1_parse_dump() and
16170   BIO_dump_indent() are added.
16171
16172   *Richard Levitte*
16173
16174 * New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
16175   these print out strings and name structures based on various
16176   flags including RFC2253 support and proper handling of
16177   multibyte characters. Added options to the 'x509' utility
16178   to allow the various flags to be set.
16179
16180   *Steve Henson*
16181
16182 * Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
16183   Also change the functions X509_cmp_current_time() and
16184   X509_gmtime_adj() work with an ASN1_TIME structure,
16185   this will enable certificates using GeneralizedTime in validity
16186   dates to be checked.
16187
16188   *Steve Henson*
16189
16190 * Make the NEG_PUBKEY_BUG code (which tolerates invalid
16191   negative public key encodings) on by default,
16192   NO_NEG_PUBKEY_BUG can be set to disable it.
16193
16194   *Steve Henson*
16195
16196 * New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
16197   content octets. An i2c_ASN1_OBJECT is unnecessary because
16198   the encoding can be trivially obtained from the structure.
16199
16200   *Steve Henson*
16201
16202 * crypto/err.c locking bugfix: Use write locks (`CRYPTO_w_[un]lock`),
16203   not read locks (`CRYPTO_r_[un]lock`).
16204
16205   *Bodo Moeller*
16206
16207 * A first attempt at creating official support for shared
16208   libraries through configuration.  I've kept it so the
16209   default is static libraries only, and the OpenSSL programs
16210   are always statically linked for now, but there are
16211   preparations for dynamic linking in place.
16212   This has been tested on Linux and Tru64.
16213
16214   *Richard Levitte*
16215
16216 * Randomness polling function for Win9x, as described in:
16217   Peter Gutmann, Software Generation of Practically Strong
16218   Random Numbers.
16219
16220   *Ulf Möller*
16221
16222 * Fix so PRNG is seeded in req if using an already existing
16223   DSA key.
16224
16225   *Steve Henson*
16226
16227 * New options to smime application. -inform and -outform
16228   allow alternative formats for the S/MIME message including
16229   PEM and DER. The -content option allows the content to be
16230   specified separately. This should allow things like Netscape
16231   form signing output easier to verify.
16232
16233   *Steve Henson*
16234
16235 * Fix the ASN1 encoding of tags using the 'long form'.
16236
16237   *Steve Henson*
16238
16239 * New ASN1 functions, `i2c_*` and `c2i_*` for INTEGER and BIT
16240   STRING types. These convert content octets to and from the
16241   underlying type. The actual tag and length octets are
16242   already assumed to have been read in and checked. These
16243   are needed because all other string types have virtually
16244   identical handling apart from the tag. By having versions
16245   of the ASN1 functions that just operate on content octets
16246   IMPLICIT tagging can be handled properly. It also allows
16247   the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
16248   and ASN1_INTEGER are identical apart from the tag.
16249
16250   *Steve Henson*
16251
16252 * Change the handling of OID objects as follows:
16253
16254   - New object identifiers are inserted in objects.txt, following
16255     the syntax given in [crypto/objects/README.md](crypto/objects/README.md).
16256   - objects.pl is used to process obj_mac.num and create a new
16257     obj_mac.h.
16258   - obj_dat.pl is used to create a new obj_dat.h, using the data in
16259     obj_mac.h.
16260
16261   This is currently kind of a hack, and the perl code in objects.pl
16262   isn't very elegant, but it works as I intended.  The simplest way
16263   to check that it worked correctly is to look in obj_dat.h and
16264   check the array nid_objs and make sure the objects haven't moved
16265   around (this is important!).  Additions are OK, as well as
16266   consistent name changes.
16267
16268   *Richard Levitte*
16269
16270 * Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
16271
16272   *Bodo Moeller*
16273
16274 * Addition of the command line parameter '-rand file' to 'openssl req'.
16275   The given file adds to whatever has already been seeded into the
16276   random pool through the RANDFILE configuration file option or
16277   environment variable, or the default random state file.
16278
16279   *Richard Levitte*
16280
16281 * mkstack.pl now sorts each macro group into lexical order.
16282   Previously the output order depended on the order the files
16283   appeared in the directory, resulting in needless rewriting
16284   of safestack.h .
16285
16286   *Steve Henson*
16287
16288 * Patches to make OpenSSL compile under Win32 again. Mostly
16289   work arounds for the VC++ problem that it treats func() as
16290   func(void). Also stripped out the parts of mkdef.pl that
16291   added extra typesafe functions: these no longer exist.
16292
16293   *Steve Henson*
16294
16295 * Reorganisation of the stack code. The macros are now all
16296   collected in safestack.h . Each macro is defined in terms of
16297   a "stack macro" of the form `SKM_<name>(type, a, b)`. The
16298   DEBUG_SAFESTACK is now handled in terms of function casts,
16299   this has the advantage of retaining type safety without the
16300   use of additional functions. If DEBUG_SAFESTACK is not defined
16301   then the non typesafe macros are used instead. Also modified the
16302   mkstack.pl script to handle the new form. Needs testing to see
16303   if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
16304   the default if no major problems. Similar behaviour for ASN1_SET_OF
16305   and PKCS12_STACK_OF.
16306
16307   *Steve Henson*
16308
16309 * When some versions of IIS use the 'NET' form of private key the
16310   key derivation algorithm is different. Normally MD5(password) is
16311   used as a 128 bit RC4 key. In the modified case
16312   MD5(MD5(password) + "SGCKEYSALT")  is used instead. Added some
16313   new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
16314   as the old Netscape_RSA functions except they have an additional
16315   'sgckey' parameter which uses the modified algorithm. Also added
16316   an -sgckey command line option to the rsa utility. Thanks to
16317   Adrian Peck <bertie@ncipher.com> for posting details of the modified
16318   algorithm to openssl-dev.
16319
16320   *Steve Henson*
16321
16322 * The evp_local.h macros were using 'c.##kname' which resulted in
16323   invalid expansion on some systems (SCO 5.0.5 for example).
16324   Corrected to 'c.kname'.
16325
16326   *Phillip Porch <root@theporch.com>*
16327
16328 * New X509_get1_email() and X509_REQ_get1_email() functions that return
16329   a STACK of email addresses from a certificate or request, these look
16330   in the subject name and the subject alternative name extensions and
16331   omit any duplicate addresses.
16332
16333   *Steve Henson*
16334
16335 * Re-implement BN_mod_exp2_mont using independent (and larger) windows.
16336   This makes DSA verification about 2 % faster.
16337
16338   *Bodo Moeller*
16339
16340 * Increase maximum window size in `BN_mod_exp_...` to 6 bits instead of 5
16341   (meaning that now 2^5 values will be precomputed, which is only 4 KB
16342   plus overhead for 1024 bit moduli).
16343   This makes exponentiations about 0.5 % faster for 1024 bit
16344   exponents (as measured by "openssl speed rsa2048").
16345
16346   *Bodo Moeller*
16347
16348 * Rename memory handling macros to avoid conflicts with other
16349   software:
16350           Malloc         =>  OPENSSL_malloc
16351           Malloc_locked  =>  OPENSSL_malloc_locked
16352           Realloc        =>  OPENSSL_realloc
16353           Free           =>  OPENSSL_free
16354
16355   *Richard Levitte*
16356
16357 * New function BN_mod_exp_mont_word for small bases (roughly 15%
16358   faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
16359
16360   *Bodo Moeller*
16361
16362 * CygWin32 support.
16363
16364   *John Jarvie <jjarvie@newsguy.com>*
16365
16366 * The type-safe stack code has been rejigged. It is now only compiled
16367   in when OpenSSL is configured with the DEBUG_SAFESTACK option and
16368   by default all type-specific stack functions are "#define"d back to
16369   standard stack functions. This results in more streamlined output
16370   but retains the type-safety checking possibilities of the original
16371   approach.
16372
16373   *Geoff Thorpe*
16374
16375 * The STACK code has been cleaned up, and certain type declarations
16376   that didn't make a lot of sense have been brought in line. This has
16377   also involved a cleanup of sorts in safestack.h to more correctly
16378   map type-safe stack functions onto their plain stack counterparts.
16379   This work has also resulted in a variety of "const"ifications of
16380   lots of the code, especially `_cmp` operations which should normally
16381   be prototyped with "const" parameters anyway.
16382
16383   *Geoff Thorpe*
16384
16385 * When generating bytes for the first time in md_rand.c, 'stir the pool'
16386   by seeding with STATE_SIZE dummy bytes (with zero entropy count).
16387   (The PRNG state consists of two parts, the large pool 'state' and 'md',
16388   where all of 'md' is used each time the PRNG is used, but 'state'
16389   is used only indexed by a cyclic counter. As entropy may not be
16390   well distributed from the beginning, 'md' is important as a
16391   chaining variable. However, the output function chains only half
16392   of 'md', i.e. 80 bits.  ssleay_rand_add, on the other hand, chains
16393   all of 'md', and seeding with STATE_SIZE dummy bytes will result
16394   in all of 'state' being rewritten, with the new values depending
16395   on virtually all of 'md'.  This overcomes the 80 bit limitation.)
16396
16397   *Bodo Moeller*
16398
16399 * In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
16400   the handshake is continued after ssl_verify_cert_chain();
16401   otherwise, if SSL_VERIFY_NONE is set, remaining error codes
16402   can lead to 'unexplainable' connection aborts later.
16403
16404   *Bodo Moeller; problem tracked down by Lutz Jaenicke*
16405
16406 * Major EVP API cipher revision.
16407   Add hooks for extra EVP features. This allows various cipher
16408   parameters to be set in the EVP interface. Support added for variable
16409   key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
16410   setting of RC2 and RC5 parameters.
16411
16412   Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
16413   ciphers.
16414
16415   Remove lots of duplicated code from the EVP library. For example *every*
16416   cipher init() function handles the 'iv' in the same way according to the
16417   cipher mode. They also all do nothing if the 'key' parameter is NULL and
16418   for CFB and OFB modes they zero ctx->num.
16419
16420   New functionality allows removal of S/MIME code RC2 hack.
16421
16422   Most of the routines have the same form and so can be declared in terms
16423   of macros.
16424
16425   By shifting this to the top level EVP_CipherInit() it can be removed from
16426   all individual ciphers. If the cipher wants to handle IVs or keys
16427   differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
16428   flags.
16429
16430   Change lots of functions like EVP_EncryptUpdate() to now return a
16431   value: although software versions of the algorithms cannot fail
16432   any installed hardware versions can.
16433
16434   *Steve Henson*
16435
16436 * Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
16437   this option is set, tolerate broken clients that send the negotiated
16438   protocol version number instead of the requested protocol version
16439   number.
16440
16441   *Bodo Moeller*
16442
16443 * Call dh_tmp_cb (set by `..._TMP_DH_CB`) with correct 'is_export' flag;
16444   i.e. non-zero for export ciphersuites, zero otherwise.
16445   Previous versions had this flag inverted, inconsistent with
16446   rsa_tmp_cb (..._TMP_RSA_CB).
16447
16448   *Bodo Moeller; problem reported by Amit Chopra*
16449
16450 * Add missing DSA library text string. Work around for some IIS
16451   key files with invalid SEQUENCE encoding.
16452
16453   *Steve Henson*
16454
16455 * Add a document (doc/standards.txt) that list all kinds of standards
16456   and so on that are implemented in OpenSSL.
16457
16458   *Richard Levitte*
16459
16460 * Enhance c_rehash script. Old version would mishandle certificates
16461   with the same subject name hash and wouldn't handle CRLs at all.
16462   Added -fingerprint option to crl utility, to support new c_rehash
16463   features.
16464
16465   *Steve Henson*
16466
16467 * Eliminate non-ANSI declarations in crypto.h and stack.h.
16468
16469   *Ulf Möller*
16470
16471 * Fix for SSL server purpose checking. Server checking was
16472   rejecting certificates which had extended key usage present
16473   but no ssl client purpose.
16474
16475   *Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>*
16476
16477 * Make PKCS#12 code work with no password. The PKCS#12 spec
16478   is a little unclear about how a blank password is handled.
16479   Since the password in encoded as a BMPString with terminating
16480   double NULL a zero length password would end up as just the
16481   double NULL. However no password at all is different and is
16482   handled differently in the PKCS#12 key generation code. NS
16483   treats a blank password as zero length. MSIE treats it as no
16484   password on export: but it will try both on import. We now do
16485   the same: PKCS12_parse() tries zero length and no password if
16486   the password is set to "" or NULL (NULL is now a valid password:
16487   it wasn't before) as does the pkcs12 application.
16488
16489   *Steve Henson*
16490
16491 * Bugfixes in `apps/x509.c`: Avoid a memory leak; and don't use
16492   perror when PEM_read_bio_X509_REQ fails, the error message must
16493   be obtained from the error queue.
16494
16495   *Bodo Moeller*
16496
16497 * Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
16498   it in ERR_remove_state if appropriate, and change ERR_get_state
16499   accordingly to avoid race conditions (this is necessary because
16500   thread_hash is no longer constant once set).
16501
16502   *Bodo Moeller*
16503
16504 * Bugfix for linux-elf makefile.one.
16505
16506   *Ulf Möller*
16507
16508 * RSA_get_default_method() will now cause a default
16509   RSA_METHOD to be chosen if one doesn't exist already.
16510   Previously this was only set during a call to RSA_new()
16511   or RSA_new_method(NULL) meaning it was possible for
16512   RSA_get_default_method() to return NULL.
16513
16514   *Geoff Thorpe*
16515
16516 * Added native name translation to the existing DSO code
16517   that will convert (if the flag to do so is set) filenames
16518   that are sufficiently small and have no path information
16519   into a canonical native form. Eg. "blah" converted to
16520   "libblah.so" or "blah.dll" etc.
16521
16522   *Geoff Thorpe*
16523
16524 * New function ERR_error_string_n(e, buf, len) which is like
16525   ERR_error_string(e, buf), but writes at most 'len' bytes
16526   including the 0 terminator.  For ERR_error_string_n, 'buf'
16527   may not be NULL.
16528
16529   *Damien Miller <djm@mindrot.org>, Bodo Moeller*
16530
16531 * CONF library reworked to become more general.  A new CONF
16532   configuration file reader "class" is implemented as well as a
16533   new functions (`NCONF_*`, for "New CONF") to handle it.  The now
16534   old `CONF_*` functions are still there, but are reimplemented to
16535   work in terms of the new functions.  Also, a set of functions
16536   to handle the internal storage of the configuration data is
16537   provided to make it easier to write new configuration file
16538   reader "classes" (I can definitely see something reading a
16539   configuration file in XML format, for example), called `_CONF_*`,
16540   or "the configuration storage API"...
16541
16542   The new configuration file reading functions are:
16543
16544           NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
16545           NCONF_get_section, NCONF_get_string, NCONF_get_numbre
16546
16547           NCONF_default, NCONF_WIN32
16548
16549           NCONF_dump_fp, NCONF_dump_bio
16550
16551   NCONF_default and NCONF_WIN32 are method (or "class") choosers,
16552   NCONF_new creates a new CONF object.  This works in the same way
16553   as other interfaces in OpenSSL, like the BIO interface.
16554   `NCONF_dump_*` dump the internal storage of the configuration file,
16555   which is useful for debugging.  All other functions take the same
16556   arguments as the old `CONF_*` functions with the exception of the
16557   first that must be a `CONF *` instead of a `LHASH *`.
16558
16559   To make it easier to use the new classes with the old `CONF_*` functions,
16560   the function CONF_set_default_method is provided.
16561
16562   *Richard Levitte*
16563
16564 * Add '-tls1' option to 'openssl ciphers', which was already
16565   mentioned in the documentation but had not been implemented.
16566   (This option is not yet really useful because even the additional
16567   experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
16568
16569   *Bodo Moeller*
16570
16571 * Initial DSO code added into libcrypto for letting OpenSSL (and
16572   OpenSSL-based applications) load shared libraries and bind to
16573   them in a portable way.
16574
16575   *Geoff Thorpe, with contributions from Richard Levitte*
16576
16577### Changes between 0.9.5 and 0.9.5a  [1 Apr 2000]
16578
16579 * Make sure _lrotl and _lrotr are only used with MSVC.
16580
16581 * Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
16582   (the default implementation of RAND_status).
16583
16584 * Rename openssl x509 option '-crlext', which was added in 0.9.5,
16585   to '-clrext' (= clear extensions), as intended and documented.
16586   *Bodo Moeller; inconsistency pointed out by Michael Attili
16587   <attili@amaxo.com>*
16588
16589 * Fix for HMAC. It wasn't zeroing the rest of the block if the key length
16590   was larger than the MD block size.
16591
16592   *Steve Henson, pointed out by Yost William <YostW@tce.com>*
16593
16594 * Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
16595   fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
16596   using the passed key: if the passed key was a private key the result
16597   of X509_print(), for example, would be to print out all the private key
16598   components.
16599
16600   *Steve Henson*
16601
16602 * des_quad_cksum() byte order bug fix.
16603   *Ulf Möller, using the problem description in krb4-0.9.7, where
16604   the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>*
16605
16606 * Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
16607   discouraged.
16608
16609   *Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>*
16610
16611 * For easily testing in shell scripts whether some command
16612   'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
16613   returns with exit code 0 iff no command of the given name is available.
16614   'no-XXX' is printed in this case, 'XXX' otherwise.  In both cases,
16615   the output goes to stdout and nothing is printed to stderr.
16616   Additional arguments are always ignored.
16617
16618   Since for each cipher there is a command of the same name,
16619   the 'no-cipher' compilation switches can be tested this way.
16620
16621   ('openssl no-XXX' is not able to detect pseudo-commands such
16622   as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
16623
16624   *Bodo Moeller*
16625
16626 * Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
16627
16628   *Bodo Moeller*
16629
16630 * For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
16631   is set; it will be thrown away anyway because each handshake creates
16632   its own key.
16633   ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
16634   to parameters -- in previous versions (since OpenSSL 0.9.3) the
16635   'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning
16636   you effectively got SSL_OP_SINGLE_DH_USE when using this macro.
16637
16638   *Bodo Moeller*
16639
16640 * New s_client option -ign_eof: EOF at stdin is ignored, and
16641   'Q' and 'R' lose their special meanings (quit/renegotiate).
16642   This is part of what -quiet does; unlike -quiet, -ign_eof
16643   does not suppress any output.
16644
16645   *Richard Levitte*
16646
16647 * Add compatibility options to the purpose and trust code. The
16648   purpose X509_PURPOSE_ANY is "any purpose" which automatically
16649   accepts a certificate or CA, this was the previous behaviour,
16650   with all the associated security issues.
16651
16652   X509_TRUST_COMPAT is the old trust behaviour: only and
16653   automatically trust self signed roots in certificate store. A
16654   new trust setting X509_TRUST_DEFAULT is used to specify that
16655   a purpose has no associated trust setting and it should instead
16656   use the value in the default purpose.
16657
16658   *Steve Henson*
16659
16660 * Fix the PKCS#8 DSA private key code so it decodes keys again
16661   and fix a memory leak.
16662
16663   *Steve Henson*
16664
16665 * In util/mkerr.pl (which implements 'make errors'), preserve
16666   reason strings from the previous version of the .c file, as
16667   the default to have only downcase letters (and digits) in
16668   automatically generated reasons codes is not always appropriate.
16669
16670   *Bodo Moeller*
16671
16672 * In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
16673   using strerror.  Previously, ERR_reason_error_string() returned
16674   library names as reason strings for SYSerr; but SYSerr is a special
16675   case where small numbers are errno values, not library numbers.
16676
16677   *Bodo Moeller*
16678
16679 * Add '-dsaparam' option to 'openssl dhparam' application.  This
16680   converts DSA parameters into DH parameters. (When creating parameters,
16681   DSA_generate_parameters is used.)
16682
16683   *Bodo Moeller*
16684
16685 * Include 'length' (recommended exponent length) in C code generated
16686   by 'openssl dhparam -C'.
16687
16688   *Bodo Moeller*
16689
16690 * The second argument to set_label in perlasm was already being used
16691   so couldn't be used as a "file scope" flag. Moved to third argument
16692   which was free.
16693
16694   *Steve Henson*
16695
16696 * In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
16697   instead of RAND_bytes for encryption IVs and salts.
16698
16699   *Bodo Moeller*
16700
16701 * Include RAND_status() into RAND_METHOD instead of implementing
16702   it only for md_rand.c  Otherwise replacing the PRNG by calling
16703   RAND_set_rand_method would be impossible.
16704
16705   *Bodo Moeller*
16706
16707 * Don't let DSA_generate_key() enter an infinite loop if the random
16708   number generation fails.
16709
16710   *Bodo Moeller*
16711
16712 * New 'rand' application for creating pseudo-random output.
16713
16714   *Bodo Moeller*
16715
16716 * Added configuration support for Linux/IA64
16717
16718   *Rolf Haberrecker <rolf@suse.de>*
16719
16720 * Assembler module support for Mingw32.
16721
16722   *Ulf Möller*
16723
16724 * Shared library support for HPUX (in shlib/).
16725
16726   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous*
16727
16728 * Shared library support for Solaris gcc.
16729
16730   *Lutz Behnke <behnke@trustcenter.de>*
16731
16732### Changes between 0.9.4 and 0.9.5  [28 Feb 2000]
16733
16734 * PKCS7_encrypt() was adding text MIME headers twice because they
16735   were added manually and by SMIME_crlf_copy().
16736
16737   *Steve Henson*
16738
16739 * In bntest.c don't call BN_rand with zero bits argument.
16740
16741   *Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>*
16742
16743 * BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
16744   case was implemented. This caused BN_div_recp() to fail occasionally.
16745
16746   *Ulf Möller*
16747
16748 * Add an optional second argument to the set_label() in the perl
16749   assembly language builder. If this argument exists and is set
16750   to 1 it signals that the assembler should use a symbol whose
16751   scope is the entire file, not just the current function. This
16752   is needed with MASM which uses the format label:: for this scope.
16753
16754   *Steve Henson, pointed out by Peter Runestig <peter@runestig.com>*
16755
16756 * Change the ASN1 types so they are typedefs by default. Before
16757   almost all types were #define'd to ASN1_STRING which was causing
16758   STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
16759   for example.
16760
16761   *Steve Henson*
16762
16763 * Change names of new functions to the new get1/get0 naming
16764   convention: After 'get1', the caller owns a reference count
16765   and has to call `..._free`; 'get0' returns a pointer to some
16766   data structure without incrementing reference counters.
16767   (Some of the existing 'get' functions increment a reference
16768   counter, some don't.)
16769   Similarly, 'set1' and 'add1' functions increase reference
16770   counters or duplicate objects.
16771
16772   *Steve Henson*
16773
16774 * Allow for the possibility of temp RSA key generation failure:
16775   the code used to assume it always worked and crashed on failure.
16776
16777   *Steve Henson*
16778
16779 * Fix potential buffer overrun problem in BIO_printf().
16780   *Ulf Möller, using public domain code by Patrick Powell; problem
16781   pointed out by David Sacerdote <das33@cornell.edu>*
16782
16783 * Support EGD <http://www.lothar.com/tech/crypto/>.  New functions
16784   RAND_egd() and RAND_status().  In the command line application,
16785   the EGD socket can be specified like a seed file using RANDFILE
16786   or -rand.
16787
16788   *Ulf Möller*
16789
16790 * Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
16791   Some CAs (e.g. Verisign) distribute certificates in this form.
16792
16793   *Steve Henson*
16794
16795 * Remove the SSL_ALLOW_ADH compile option and set the default cipher
16796   list to exclude them. This means that no special compilation option
16797   is needed to use anonymous DH: it just needs to be included in the
16798   cipher list.
16799
16800   *Steve Henson*
16801
16802 * Change the EVP_MD_CTX_type macro so its meaning consistent with
16803   EVP_MD_type. The old functionality is available in a new macro called
16804   EVP_MD_md(). Change code that uses it and update docs.
16805
16806   *Steve Henson*
16807
16808 * `..._ctrl` functions now have corresponding `..._callback_ctrl` functions
16809   where the `void *` argument is replaced by a function pointer argument.
16810   Previously `void *` was abused to point to functions, which works on
16811   many platforms, but is not correct.  As these functions are usually
16812   called by macros defined in OpenSSL header files, most source code
16813   should work without changes.
16814
16815   *Richard Levitte*
16816
16817 * `<openssl/opensslconf.h>` (which is created by Configure) now contains
16818   sections with information on -D... compiler switches used for
16819   compiling the library so that applications can see them.  To enable
16820   one of these sections, a pre-processor symbol `OPENSSL_..._DEFINES`
16821   must be defined.  E.g.,
16822           #define OPENSSL_ALGORITHM_DEFINES
16823           #include <openssl/opensslconf.h>
16824   defines all pertinent `NO_<algo>` symbols, such as NO_IDEA, NO_RSA, etc.
16825
16826   *Richard Levitte, Ulf and Bodo Möller*
16827
16828 * Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
16829   record layer.
16830
16831   *Bodo Moeller*
16832
16833 * Change the 'other' type in certificate aux info to a STACK_OF
16834   X509_ALGOR. Although not an AlgorithmIdentifier as such it has
16835   the required ASN1 format: arbitrary types determined by an OID.
16836
16837   *Steve Henson*
16838
16839 * Add some PEM_write_X509_REQ_NEW() functions and a command line
16840   argument to 'req'. This is not because the function is newer or
16841   better than others it just uses the work 'NEW' in the certificate
16842   request header lines. Some software needs this.
16843
16844   *Steve Henson*
16845
16846 * Reorganise password command line arguments: now passwords can be
16847   obtained from various sources. Delete the PEM_cb function and make
16848   it the default behaviour: i.e. if the callback is NULL and the
16849   usrdata argument is not NULL interpret it as a null terminated pass
16850   phrase. If usrdata and the callback are NULL then the pass phrase
16851   is prompted for as usual.
16852
16853   *Steve Henson*
16854
16855 * Add support for the Compaq Atalla crypto accelerator. If it is installed,
16856   the support is automatically enabled. The resulting binaries will
16857   autodetect the card and use it if present.
16858
16859   *Ben Laurie and Compaq Inc.*
16860
16861 * Work around for Netscape hang bug. This sends certificate request
16862   and server done in one record. Since this is perfectly legal in the
16863   SSL/TLS protocol it isn't a "bug" option and is on by default. See
16864   the bugs/SSLv3 entry for more info.
16865
16866   *Steve Henson*
16867
16868 * HP-UX tune-up: new unified configs, HP C compiler bug workaround.
16869
16870   *Andy Polyakov*
16871
16872 * Add -rand argument to smime and pkcs12 applications and read/write
16873   of seed file.
16874
16875   *Steve Henson*
16876
16877 * New 'passwd' tool for crypt(3) and apr1 password hashes.
16878
16879   *Bodo Moeller*
16880
16881 * Add command line password options to the remaining applications.
16882
16883   *Steve Henson*
16884
16885 * Bug fix for BN_div_recp() for numerators with an even number of
16886   bits.
16887
16888   *Ulf Möller*
16889
16890 * More tests in bntest.c, and changed test_bn output.
16891
16892   *Ulf Möller*
16893
16894 * ./config recognizes MacOS X now.
16895
16896   *Andy Polyakov*
16897
16898 * Bug fix for BN_div() when the first words of num and divisor are
16899   equal (it gave wrong results if `(rem=(n1-q*d0)&BN_MASK2) < d0)`.
16900
16901   *Ulf Möller*
16902
16903 * Add support for various broken PKCS#8 formats, and command line
16904   options to produce them.
16905
16906   *Steve Henson*
16907
16908 * New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
16909   get temporary BIGNUMs from a BN_CTX.
16910
16911   *Ulf Möller*
16912
16913 * Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
16914   for p == 0.
16915
16916   *Ulf Möller*
16917
16918 * Change the `SSLeay_add_all_*()` functions to `OpenSSL_add_all_*()` and
16919   include a #define from the old name to the new. The original intent
16920   was that statically linked binaries could for example just call
16921   SSLeay_add_all_ciphers() to just add ciphers to the table and not
16922   link with digests. This never worked because SSLeay_add_all_digests()
16923   and SSLeay_add_all_ciphers() were in the same source file so calling
16924   one would link with the other. They are now in separate source files.
16925
16926   *Steve Henson*
16927
16928 * Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
16929
16930   *Steve Henson*
16931
16932 * Use a less unusual form of the Miller-Rabin primality test (it used
16933   a binary algorithm for exponentiation integrated into the Miller-Rabin
16934   loop, our standard modexp algorithms are faster).
16935
16936   *Bodo Moeller*
16937
16938 * Support for the EBCDIC character set completed.
16939
16940   *Martin Kraemer <Martin.Kraemer@Mch.SNI.De>*
16941
16942 * Source code cleanups: use const where appropriate, eliminate casts,
16943   use `void *` instead of `char *` in lhash.
16944
16945   *Ulf Möller*
16946
16947 * Bugfix: ssl3_send_server_key_exchange was not restartable
16948   (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
16949   this the server could overwrite ephemeral keys that the client
16950   has already seen).
16951
16952   *Bodo Moeller*
16953
16954 * Turn DSA_is_prime into a macro that calls BN_is_prime,
16955   using 50 iterations of the Rabin-Miller test.
16956
16957   DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
16958   iterations of the Rabin-Miller test as required by the appendix
16959   to FIPS PUB 186[-1]) instead of DSA_is_prime.
16960   As BN_is_prime_fasttest includes trial division, DSA parameter
16961   generation becomes much faster.
16962
16963   This implies a change for the callback functions in DSA_is_prime
16964   and DSA_generate_parameters: The callback function is called once
16965   for each positive witness in the Rabin-Miller test, not just
16966   occasionally in the inner loop; and the parameters to the
16967   callback function now provide an iteration count for the outer
16968   loop rather than for the current invocation of the inner loop.
16969   DSA_generate_parameters additionally can call the callback
16970   function with an 'iteration count' of -1, meaning that a
16971   candidate has passed the trial division test (when q is generated
16972   from an application-provided seed, trial division is skipped).
16973
16974   *Bodo Moeller*
16975
16976 * New function BN_is_prime_fasttest that optionally does trial
16977   division before starting the Rabin-Miller test and has
16978   an additional BN_CTX * argument (whereas BN_is_prime always
16979   has to allocate at least one BN_CTX).
16980   'callback(1, -1, cb_arg)' is called when a number has passed the
16981   trial division stage.
16982
16983   *Bodo Moeller*
16984
16985 * Fix for bug in CRL encoding. The validity dates weren't being handled
16986   as ASN1_TIME.
16987
16988   *Steve Henson*
16989
16990 * New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
16991
16992   *Steve Henson*
16993
16994 * New function BN_pseudo_rand().
16995
16996   *Ulf Möller*
16997
16998 * Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
16999   bignum version of BN_from_montgomery() with the working code from
17000   SSLeay 0.9.0 (the word based version is faster anyway), and clean up
17001   the comments.
17002
17003   *Ulf Möller*
17004
17005 * Avoid a race condition in s2_clnt.c (function get_server_hello) that
17006   made it impossible to use the same SSL_SESSION data structure in
17007   SSL2 clients in multiple threads.
17008
17009   *Bodo Moeller*
17010
17011 * The return value of RAND_load_file() no longer counts bytes obtained
17012   by stat().  RAND_load_file(..., -1) is new and uses the complete file
17013   to seed the PRNG (previously an explicit byte count was required).
17014
17015   *Ulf Möller, Bodo Möller*
17016
17017 * Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
17018   used `char *` instead of `void *` and had casts all over the place.
17019
17020   *Steve Henson*
17021
17022 * Make BN_generate_prime() return NULL on error if ret!=NULL.
17023
17024   *Ulf Möller*
17025
17026 * Retain source code compatibility for BN_prime_checks macro:
17027   BN_is_prime(..., BN_prime_checks, ...) now uses
17028   BN_prime_checks_for_size to determine the appropriate number of
17029   Rabin-Miller iterations.
17030
17031   *Ulf Möller*
17032
17033 * Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
17034   DH_CHECK_P_NOT_SAFE_PRIME.
17035   (Check if this is true? OpenPGP calls them "strong".)
17036
17037   *Ulf Möller*
17038
17039 * Merge the functionality of "dh" and "gendh" programs into a new program
17040   "dhparam". The old programs are retained for now but will handle DH keys
17041   (instead of parameters) in future.
17042
17043   *Steve Henson*
17044
17045 * Make the ciphers, s_server and s_client programs check the return values
17046   when a new cipher list is set.
17047
17048   *Steve Henson*
17049
17050 * Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
17051   ciphers. Before when the 56bit ciphers were enabled the sorting was
17052   wrong.
17053
17054   The syntax for the cipher sorting has been extended to support sorting by
17055   cipher-strength (using the strength_bits hard coded in the tables).
17056   The new command is `@STRENGTH` (see also `doc/apps/ciphers.pod`).
17057
17058   Fix a bug in the cipher-command parser: when supplying a cipher command
17059   string with an "undefined" symbol (neither command nor alphanumeric
17060   *A-Za-z0-9*, ssl_set_cipher_list used to hang in an endless loop. Now
17061   an error is flagged.
17062
17063   Due to the strength-sorting extension, the code of the
17064   ssl_create_cipher_list() function was completely rearranged. I hope that
17065   the readability was also increased :-)
17066
17067   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>*
17068
17069 * Minor change to 'x509' utility. The -CAcreateserial option now uses 1
17070   for the first serial number and places 2 in the serial number file. This
17071   avoids problems when the root CA is created with serial number zero and
17072   the first user certificate has the same issuer name and serial number
17073   as the root CA.
17074
17075   *Steve Henson*
17076
17077 * Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
17078   the new code. Add documentation for this stuff.
17079
17080   *Steve Henson*
17081
17082 * Changes to X509_ATTRIBUTE utilities. These have been renamed from
17083   `X509_*()` to `X509at_*()` on the grounds that they don't handle X509
17084   structures and behave in an analogous way to the X509v3 functions:
17085   they shouldn't be called directly but wrapper functions should be used
17086   instead.
17087
17088   So we also now have some wrapper functions that call the X509at functions
17089   when passed certificate requests. (TO DO: similar things can be done with
17090   PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
17091   things. Some of these need some d2i or i2d and print functionality
17092   because they handle more complex structures.)
17093
17094   *Steve Henson*
17095
17096 * Add missing #ifndefs that caused missing symbols when building libssl
17097   as a shared library without RSA.  Use #ifndef NO_SSL2 instead of
17098   NO_RSA in `ssl/s2*.c`.
17099
17100   *Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller*
17101
17102 * Precautions against using the PRNG uninitialized: RAND_bytes() now
17103   has a return value which indicates the quality of the random data
17104   (1 = ok, 0 = not seeded).  Also an error is recorded on the thread's
17105   error queue. New function RAND_pseudo_bytes() generates output that is
17106   guaranteed to be unique but not unpredictable. RAND_add is like
17107   RAND_seed, but takes an extra argument for an entropy estimate
17108   (RAND_seed always assumes full entropy).
17109
17110   *Ulf Möller*
17111
17112 * Do more iterations of Rabin-Miller probable prime test (specifically,
17113   3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
17114   instead of only 2 for all lengths; see BN_prime_checks_for_size definition
17115   in crypto/bn/bn_prime.c for the complete table).  This guarantees a
17116   false-positive rate of at most 2^-80 for random input.
17117
17118   *Bodo Moeller*
17119
17120 * Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
17121
17122   *Bodo Moeller*
17123
17124 * New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
17125   in the 0.9.5 release), this returns the chain
17126   from an X509_CTX structure with a dup of the stack and all
17127   the X509 reference counts upped: so the stack will exist
17128   after X509_CTX_cleanup() has been called. Modify pkcs12.c
17129   to use this.
17130
17131   Also make SSL_SESSION_print() print out the verify return
17132   code.
17133
17134   *Steve Henson*
17135
17136 * Add manpage for the pkcs12 command. Also change the default
17137   behaviour so MAC iteration counts are used unless the new
17138   -nomaciter option is used. This improves file security and
17139   only older versions of MSIE (4.0 for example) need it.
17140
17141   *Steve Henson*
17142
17143 * Honor the no-xxx Configure options when creating .DEF files.
17144
17145   *Ulf Möller*
17146
17147 * Add PKCS#10 attributes to field table: challengePassword,
17148   unstructuredName and unstructuredAddress. These are taken from
17149   draft PKCS#9 v2.0 but are compatible with v1.2 provided no
17150   international characters are used.
17151
17152   More changes to X509_ATTRIBUTE code: allow the setting of types
17153   based on strings. Remove the 'loc' parameter when adding
17154   attributes because these will be a SET OF encoding which is sorted
17155   in ASN1 order.
17156
17157   *Steve Henson*
17158
17159 * Initial changes to the 'req' utility to allow request generation
17160   automation. This will allow an application to just generate a template
17161   file containing all the field values and have req construct the
17162   request.
17163
17164   Initial support for X509_ATTRIBUTE handling. Stacks of these are
17165   used all over the place including certificate requests and PKCS#7
17166   structures. They are currently handled manually where necessary with
17167   some primitive wrappers for PKCS#7. The new functions behave in a
17168   manner analogous to the X509 extension functions: they allow
17169   attributes to be looked up by NID and added.
17170
17171   Later something similar to the X509V3 code would be desirable to
17172   automatically handle the encoding, decoding and printing of the
17173   more complex types. The string types like challengePassword can
17174   be handled by the string table functions.
17175
17176   Also modified the multi byte string table handling. Now there is
17177   a 'global mask' which masks out certain types. The table itself
17178   can use the flag STABLE_NO_MASK to ignore the mask setting: this
17179   is useful when for example there is only one permissible type
17180   (as in countryName) and using the mask might result in no valid
17181   types at all.
17182
17183   *Steve Henson*
17184
17185 * Clean up 'Finished' handling, and add functions SSL_get_finished and
17186   SSL_get_peer_finished to allow applications to obtain the latest
17187   Finished messages sent to the peer or expected from the peer,
17188   respectively.  (SSL_get_peer_finished is usually the Finished message
17189   actually received from the peer, otherwise the protocol will be aborted.)
17190
17191   As the Finished message are message digests of the complete handshake
17192   (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
17193   be used for external authentication procedures when the authentication
17194   provided by SSL/TLS is not desired or is not enough.
17195
17196   *Bodo Moeller*
17197
17198 * Enhanced support for Alpha Linux is added. Now ./config checks if
17199   the host supports BWX extension and if Compaq C is present on the
17200   $PATH. Just exploiting of the BWX extension results in 20-30%
17201   performance kick for some algorithms, e.g. DES and RC4 to mention
17202   a couple. Compaq C in turn generates ~20% faster code for MD5 and
17203   SHA1.
17204
17205   *Andy Polyakov*
17206
17207 * Add support for MS "fast SGC". This is arguably a violation of the
17208   SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
17209   weak crypto and after checking the certificate is SGC a second one
17210   with strong crypto. MS SGC stops the first handshake after receiving
17211   the server certificate message and sends a second client hello. Since
17212   a server will typically do all the time consuming operations before
17213   expecting any further messages from the client (server key exchange
17214   is the most expensive) there is little difference between the two.
17215
17216   To get OpenSSL to support MS SGC we have to permit a second client
17217   hello message after we have sent server done. In addition we have to
17218   reset the MAC if we do get this second client hello.
17219
17220   *Steve Henson*
17221
17222 * Add a function 'd2i_AutoPrivateKey()' this will automatically decide
17223   if a DER encoded private key is RSA or DSA traditional format. Changed
17224   d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
17225   format DER encoded private key. Newer code should use PKCS#8 format which
17226   has the key type encoded in the ASN1 structure. Added DER private key
17227   support to pkcs8 application.
17228
17229   *Steve Henson*
17230
17231 * SSL 3/TLS 1 servers now don't request certificates when an anonymous
17232   ciphersuites has been selected (as required by the SSL 3/TLS 1
17233   specifications).  Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
17234   is set, we interpret this as a request to violate the specification
17235   (the worst that can happen is a handshake failure, and 'correct'
17236   behaviour would result in a handshake failure anyway).
17237
17238   *Bodo Moeller*
17239
17240 * In SSL_CTX_add_session, take into account that there might be multiple
17241   SSL_SESSION structures with the same session ID (e.g. when two threads
17242   concurrently obtain them from an external cache).
17243   The internal cache can handle only one SSL_SESSION with a given ID,
17244   so if there's a conflict, we now throw out the old one to achieve
17245   consistency.
17246
17247   *Bodo Moeller*
17248
17249 * Add OIDs for idea and blowfish in CBC mode. This will allow both
17250   to be used in PKCS#5 v2.0 and S/MIME.  Also add checking to
17251   some routines that use cipher OIDs: some ciphers do not have OIDs
17252   defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
17253   example.
17254
17255   *Steve Henson*
17256
17257 * Simplify the trust setting structure and code. Now we just have
17258   two sequences of OIDs for trusted and rejected settings. These will
17259   typically have values the same as the extended key usage extension
17260   and any application specific purposes.
17261
17262   The trust checking code now has a default behaviour: it will just
17263   check for an object with the same NID as the passed id. Functions can
17264   be provided to override either the default behaviour or the behaviour
17265   for a given id. SSL client, server and email already have functions
17266   in place for compatibility: they check the NID and also return "trusted"
17267   if the certificate is self signed.
17268
17269   *Steve Henson*
17270
17271 * Add d2i,i2d bio/fp functions for PrivateKey: these convert the
17272   traditional format into an EVP_PKEY structure.
17273
17274   *Steve Henson*
17275
17276 * Add a password callback function PEM_cb() which either prompts for
17277   a password if usr_data is NULL or otherwise assumes it is a null
17278   terminated password. Allow passwords to be passed on command line
17279   environment or config files in a few more utilities.
17280
17281   *Steve Henson*
17282
17283 * Add a bunch of DER and PEM functions to handle PKCS#8 format private
17284   keys. Add some short names for PKCS#8 PBE algorithms and allow them
17285   to be specified on the command line for the pkcs8 and pkcs12 utilities.
17286   Update documentation.
17287
17288   *Steve Henson*
17289
17290 * Support for ASN1 "NULL" type. This could be handled before by using
17291   ASN1_TYPE but there wasn't any function that would try to read a NULL
17292   and produce an error if it couldn't. For compatibility we also have
17293   ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
17294   don't allocate anything because they don't need to.
17295
17296   *Steve Henson*
17297
17298 * Initial support for MacOS is now provided. Examine INSTALL.MacOS
17299   for details.
17300
17301   *Andy Polyakov, Roy Woods <roy@centicsystems.ca>*
17302
17303 * Rebuild of the memory allocation routines used by OpenSSL code and
17304   possibly others as well.  The purpose is to make an interface that
17305   provide hooks so anyone can build a separate set of allocation and
17306   deallocation routines to be used by OpenSSL, for example memory
17307   pool implementations, or something else, which was previously hard
17308   since Malloc(), Realloc() and Free() were defined as macros having
17309   the values malloc, realloc and free, respectively (except for Win32
17310   compilations).  The same is provided for memory debugging code.
17311   OpenSSL already comes with functionality to find memory leaks, but
17312   this gives people a chance to debug other memory problems.
17313
17314   With these changes, a new set of functions and macros have appeared:
17315
17316     CRYPTO_set_mem_debug_functions()         [F]
17317     CRYPTO_get_mem_debug_functions()         [F]
17318     CRYPTO_dbg_set_options()                 [F]
17319     CRYPTO_dbg_get_options()                 [F]
17320     CRYPTO_malloc_debug_init()               [M]
17321
17322   The memory debug functions are NULL by default, unless the library
17323   is compiled with CRYPTO_MDEBUG or friends is defined.  If someone
17324   wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
17325   gives the standard debugging functions that come with OpenSSL) or
17326   CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
17327   provided by the library user) must be used.  When the standard
17328   debugging functions are used, CRYPTO_dbg_set_options can be used to
17329   request additional information:
17330   CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
17331   the CRYPTO_MDEBUG_xxx macro when compiling the library.
17332
17333   Also, things like CRYPTO_set_mem_functions will always give the
17334   expected result (the new set of functions is used for allocation
17335   and deallocation) at all times, regardless of platform and compiler
17336   options.
17337
17338   To finish it up, some functions that were never use in any other
17339   way than through macros have a new API and new semantic:
17340
17341     CRYPTO_dbg_malloc()
17342     CRYPTO_dbg_realloc()
17343     CRYPTO_dbg_free()
17344
17345   All macros of value have retained their old syntax.
17346
17347   *Richard Levitte and Bodo Moeller*
17348
17349 * Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
17350   ordering of SMIMECapabilities wasn't in "strength order" and there
17351   was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
17352   algorithm.
17353
17354   *Steve Henson*
17355
17356 * Some ASN1 types with illegal zero length encoding (INTEGER,
17357   ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
17358
17359   *Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson*
17360
17361 * Merge in my S/MIME library for OpenSSL. This provides a simple
17362   S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
17363   functionality to handle multipart/signed properly) and a utility
17364   called 'smime' to call all this stuff. This is based on code I
17365   originally wrote for Celo who have kindly allowed it to be
17366   included in OpenSSL.
17367
17368   *Steve Henson*
17369
17370 * Add variants des_set_key_checked and des_set_key_unchecked of
17371   des_set_key (aka des_key_sched).  Global variable des_check_key
17372   decides which of these is called by des_set_key; this way
17373   des_check_key behaves as it always did, but applications and
17374   the library itself, which was buggy for des_check_key == 1,
17375   have a cleaner way to pick the version they need.
17376
17377   *Bodo Moeller*
17378
17379 * New function PKCS12_newpass() which changes the password of a
17380   PKCS12 structure.
17381
17382   *Steve Henson*
17383
17384 * Modify X509_TRUST and X509_PURPOSE so it also uses a static and
17385   dynamic mix. In both cases the ids can be used as an index into the
17386   table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
17387   functions so they accept a list of the field values and the
17388   application doesn't need to directly manipulate the X509_TRUST
17389   structure.
17390
17391   *Steve Henson*
17392
17393 * Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
17394   need initialising.
17395
17396   *Steve Henson*
17397
17398 * Modify the way the V3 extension code looks up extensions. This now
17399   works in a similar way to the object code: we have some "standard"
17400   extensions in a static table which is searched with OBJ_bsearch()
17401   and the application can add dynamic ones if needed. The file
17402   crypto/x509v3/ext_dat.h now has the info: this file needs to be
17403   updated whenever a new extension is added to the core code and kept
17404   in ext_nid order. There is a simple program 'tabtest.c' which checks
17405   this. New extensions are not added too often so this file can readily
17406   be maintained manually.
17407
17408   There are two big advantages in doing things this way. The extensions
17409   can be looked up immediately and no longer need to be "added" using
17410   X509V3_add_standard_extensions(): this function now does nothing.
17411   Side note: I get *lots* of email saying the extension code doesn't
17412   work because people forget to call this function.
17413   Also no dynamic allocation is done unless new extensions are added:
17414   so if we don't add custom extensions there is no need to call
17415   X509V3_EXT_cleanup().
17416
17417   *Steve Henson*
17418
17419 * Modify enc utility's salting as follows: make salting the default. Add a
17420   magic header, so unsalted files fail gracefully instead of just decrypting
17421   to garbage. This is because not salting is a big security hole, so people
17422   should be discouraged from doing it.
17423
17424   *Ben Laurie*
17425
17426 * Fixes and enhancements to the 'x509' utility. It allowed a message
17427   digest to be passed on the command line but it only used this
17428   parameter when signing a certificate. Modified so all relevant
17429   operations are affected by the digest parameter including the
17430   -fingerprint and -x509toreq options. Also -x509toreq choked if a
17431   DSA key was used because it didn't fix the digest.
17432
17433   *Steve Henson*
17434
17435 * Initial certificate chain verify code. Currently tests the untrusted
17436   certificates for consistency with the verify purpose (which is set
17437   when the X509_STORE_CTX structure is set up) and checks the pathlength.
17438
17439   There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
17440   this is because it will reject chains with invalid extensions whereas
17441   every previous version of OpenSSL and SSLeay made no checks at all.
17442
17443   Trust code: checks the root CA for the relevant trust settings. Trust
17444   settings have an initial value consistent with the verify purpose: e.g.
17445   if the verify purpose is for SSL client use it expects the CA to be
17446   trusted for SSL client use. However the default value can be changed to
17447   permit custom trust settings: one example of this would be to only trust
17448   certificates from a specific "secure" set of CAs.
17449
17450   Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
17451   which should be used for version portability: especially since the
17452   verify structure is likely to change more often now.
17453
17454   SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
17455   to set them. If not set then assume SSL clients will verify SSL servers
17456   and vice versa.
17457
17458   Two new options to the verify program: -untrusted allows a set of
17459   untrusted certificates to be passed in and -purpose which sets the
17460   intended purpose of the certificate. If a purpose is set then the
17461   new chain verify code is used to check extension consistency.
17462
17463   *Steve Henson*
17464
17465 * Support for the authority information access extension.
17466
17467   *Steve Henson*
17468
17469 * Modify RSA and DSA PEM read routines to transparently handle
17470   PKCS#8 format private keys. New *_PUBKEY_* functions that handle
17471   public keys in a format compatible with certificate
17472   SubjectPublicKeyInfo structures. Unfortunately there were already
17473   functions called *_PublicKey_* which used various odd formats so
17474   these are retained for compatibility: however the DSA variants were
17475   never in a public release so they have been deleted. Changed dsa/rsa
17476   utilities to handle the new format: note no releases ever handled public
17477   keys so we should be OK.
17478
17479   The primary motivation for this change is to avoid the same fiasco
17480   that dogs private keys: there are several incompatible private key
17481   formats some of which are standard and some OpenSSL specific and
17482   require various evil hacks to allow partial transparent handling and
17483   even then it doesn't work with DER formats. Given the option anything
17484   other than PKCS#8 should be dumped: but the other formats have to
17485   stay in the name of compatibility.
17486
17487   With public keys and the benefit of hindsight one standard format
17488   is used which works with EVP_PKEY, RSA or DSA structures: though
17489   it clearly returns an error if you try to read the wrong kind of key.
17490
17491   Added a -pubkey option to the 'x509' utility to output the public key.
17492   Also rename the `EVP_PKEY_get_*()` to `EVP_PKEY_rget_*()`
17493   (renamed to `EVP_PKEY_get1_*()` in the OpenSSL 0.9.5 release) and add
17494   `EVP_PKEY_rset_*()` functions (renamed to `EVP_PKEY_set1_*()`)
17495   that do the same as the `EVP_PKEY_assign_*()` except they up the
17496   reference count of the added key (they don't "swallow" the
17497   supplied key).
17498
17499   *Steve Henson*
17500
17501 * Fixes to crypto/x509/by_file.c the code to read in certificates and
17502   CRLs would fail if the file contained no certificates or no CRLs:
17503   added a new function to read in both types and return the number
17504   read: this means that if none are read it will be an error. The
17505   DER versions of the certificate and CRL reader would always fail
17506   because it isn't possible to mix certificates and CRLs in DER format
17507   without choking one or the other routine. Changed this to just read
17508   a certificate: this is the best we can do. Also modified the code
17509   in `apps/verify.c` to take notice of return codes: it was previously
17510   attempting to read in certificates from NULL pointers and ignoring
17511   any errors: this is one reason why the cert and CRL reader seemed
17512   to work. It doesn't check return codes from the default certificate
17513   routines: these may well fail if the certificates aren't installed.
17514
17515   *Steve Henson*
17516
17517 * Code to support otherName option in GeneralName.
17518
17519   *Steve Henson*
17520
17521 * First update to verify code. Change the verify utility
17522   so it warns if it is passed a self signed certificate:
17523   for consistency with the normal behaviour. X509_verify
17524   has been modified to it will now verify a self signed
17525   certificate if *exactly* the same certificate appears
17526   in the store: it was previously impossible to trust a
17527   single self signed certificate. This means that:
17528   openssl verify ss.pem
17529   now gives a warning about a self signed certificate but
17530   openssl verify -CAfile ss.pem ss.pem
17531   is OK.
17532
17533   *Steve Henson*
17534
17535 * For servers, store verify_result in SSL_SESSION data structure
17536   (and add it to external session representation).
17537   This is needed when client certificate verifications fails,
17538   but an application-provided verification callback (set by
17539   SSL_CTX_set_cert_verify_callback) allows accepting the session
17540   anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
17541   but returns 1): When the session is reused, we have to set
17542   ssl->verify_result to the appropriate error code to avoid
17543   security holes.
17544
17545   *Bodo Moeller, problem pointed out by Lutz Jaenicke*
17546
17547 * Fix a bug in the new PKCS#7 code: it didn't consider the
17548   case in PKCS7_dataInit() where the signed PKCS7 structure
17549   didn't contain any existing data because it was being created.
17550
17551   *Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson*
17552
17553 * Add a salt to the key derivation routines in enc.c. This
17554   forms the first 8 bytes of the encrypted file. Also add a
17555   -S option to allow a salt to be input on the command line.
17556
17557   *Steve Henson*
17558
17559 * New function X509_cmp(). Oddly enough there wasn't a function
17560   to compare two certificates. We do this by working out the SHA1
17561   hash and comparing that. X509_cmp() will be needed by the trust
17562   code.
17563
17564   *Steve Henson*
17565
17566 * SSL_get1_session() is like SSL_get_session(), but increments
17567   the reference count in the SSL_SESSION returned.
17568
17569   *Geoff Thorpe <geoff@eu.c2.net>*
17570
17571 * Fix for 'req': it was adding a null to request attributes.
17572   Also change the X509_LOOKUP and X509_INFO code to handle
17573   certificate auxiliary information.
17574
17575   *Steve Henson*
17576
17577 * Add support for 40 and 64 bit RC2 and RC4 algorithms: document
17578   the 'enc' command.
17579
17580   *Steve Henson*
17581
17582 * Add the possibility to add extra information to the memory leak
17583   detecting output, to form tracebacks, showing from where each
17584   allocation was originated: CRYPTO_push_info("constant string") adds
17585   the string plus current file name and line number to a per-thread
17586   stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
17587   is like calling CYRPTO_pop_info() until the stack is empty.
17588   Also updated memory leak detection code to be multi-thread-safe.
17589
17590   *Richard Levitte*
17591
17592 * Add options -text and -noout to pkcs7 utility and delete the
17593   encryption options which never did anything. Update docs.
17594
17595   *Steve Henson*
17596
17597 * Add options to some of the utilities to allow the pass phrase
17598   to be included on either the command line (not recommended on
17599   OSes like Unix) or read from the environment. Update the
17600   manpages and fix a few bugs.
17601
17602   *Steve Henson*
17603
17604 * Add a few manpages for some of the openssl commands.
17605
17606   *Steve Henson*
17607
17608 * Fix the -revoke option in ca. It was freeing up memory twice,
17609   leaking and not finding already revoked certificates.
17610
17611   *Steve Henson*
17612
17613 * Extensive changes to support certificate auxiliary information.
17614   This involves the use of X509_CERT_AUX structure and X509_AUX
17615   functions. An X509_AUX function such as PEM_read_X509_AUX()
17616   can still read in a certificate file in the usual way but it
17617   will also read in any additional "auxiliary information". By
17618   doing things this way a fair degree of compatibility can be
17619   retained: existing certificates can have this information added
17620   using the new 'x509' options.
17621
17622   Current auxiliary information includes an "alias" and some trust
17623   settings. The trust settings will ultimately be used in enhanced
17624   certificate chain verification routines: currently a certificate
17625   can only be trusted if it is self signed and then it is trusted
17626   for all purposes.
17627
17628   *Steve Henson*
17629
17630 * Fix assembler for Alpha (tested only on DEC OSF not Linux or `*BSD`).
17631   The problem was that one of the replacement routines had not been working
17632   since SSLeay releases.  For now the offending routine has been replaced
17633   with non-optimised assembler.  Even so, this now gives around 95%
17634   performance improvement for 1024 bit RSA signs.
17635
17636   *Mark Cox*
17637
17638 * Hack to fix PKCS#7 decryption when used with some unorthodox RC2
17639   handling. Most clients have the effective key size in bits equal to
17640   the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
17641   A few however don't do this and instead use the size of the decrypted key
17642   to determine the RC2 key length and the AlgorithmIdentifier to determine
17643   the effective key length. In this case the effective key length can still
17644   be 40 bits but the key length can be 168 bits for example. This is fixed
17645   by manually forcing an RC2 key into the EVP_PKEY structure because the
17646   EVP code can't currently handle unusual RC2 key sizes: it always assumes
17647   the key length and effective key length are equal.
17648
17649   *Steve Henson*
17650
17651 * Add a bunch of functions that should simplify the creation of
17652   X509_NAME structures. Now you should be able to do:
17653   X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
17654   and have it automatically work out the correct field type and fill in
17655   the structures. The more adventurous can try:
17656   X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
17657   and it will (hopefully) work out the correct multibyte encoding.
17658
17659   *Steve Henson*
17660
17661 * Change the 'req' utility to use the new field handling and multibyte
17662   copy routines. Before the DN field creation was handled in an ad hoc
17663   way in req, ca, and x509 which was rather broken and didn't support
17664   BMPStrings or UTF8Strings. Since some software doesn't implement
17665   BMPStrings or UTF8Strings yet, they can be enabled using the config file
17666   using the dirstring_type option. See the new comment in the default
17667   openssl.cnf for more info.
17668
17669   *Steve Henson*
17670
17671 * Make crypto/rand/md_rand.c more robust:
17672   - Assure unique random numbers after fork().
17673   - Make sure that concurrent threads access the global counter and
17674     md serializably so that we never lose entropy in them
17675     or use exactly the same state in multiple threads.
17676     Access to the large state is not always serializable because
17677     the additional locking could be a performance killer, and
17678     md should be large enough anyway.
17679
17680   *Bodo Moeller*
17681
17682 * New file `apps/app_rand.c` with commonly needed functionality
17683   for handling the random seed file.
17684
17685   Use the random seed file in some applications that previously did not:
17686           ca,
17687           dsaparam -genkey (which also ignored its '-rand' option),
17688           s_client,
17689           s_server,
17690           x509 (when signing).
17691   Except on systems with /dev/urandom, it is crucial to have a random
17692   seed file at least for key creation, DSA signing, and for DH exchanges;
17693   for RSA signatures we could do without one.
17694
17695   gendh and gendsa (unlike genrsa) used to read only the first byte
17696   of each file listed in the '-rand' option.  The function as previously
17697   found in genrsa is now in app_rand.c and is used by all programs
17698   that support '-rand'.
17699
17700   *Bodo Moeller*
17701
17702 * In RAND_write_file, use mode 0600 for creating files;
17703   don't just chmod when it may be too late.
17704
17705   *Bodo Moeller*
17706
17707 * Report an error from X509_STORE_load_locations
17708   when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
17709
17710   *Bill Perry*
17711
17712 * New function ASN1_mbstring_copy() this copies a string in either
17713   ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
17714   into an ASN1_STRING type. A mask of permissible types is passed
17715   and it chooses the "minimal" type to use or an error if not type
17716   is suitable.
17717
17718   *Steve Henson*
17719
17720 * Add function equivalents to the various macros in asn1.h. The old
17721   macros are retained with an `M_` prefix. Code inside the library can
17722   use the `M_` macros. External code (including the openssl utility)
17723   should *NOT* in order to be "shared library friendly".
17724
17725   *Steve Henson*
17726
17727 * Add various functions that can check a certificate's extensions
17728   to see if it usable for various purposes such as SSL client,
17729   server or S/MIME and CAs of these types. This is currently
17730   VERY EXPERIMENTAL but will ultimately be used for certificate chain
17731   verification. Also added a -purpose flag to x509 utility to
17732   print out all the purposes.
17733
17734   *Steve Henson*
17735
17736 * Add a CRYPTO_EX_DATA to X509 certificate structure and associated
17737   functions.
17738
17739   *Steve Henson*
17740
17741 * New `X509V3_{X509,CRL,REVOKED}_get_d2i()` functions. These will search
17742   for, obtain and decode and extension and obtain its critical flag.
17743   This allows all the necessary extension code to be handled in a
17744   single function call.
17745
17746   *Steve Henson*
17747
17748 * RC4 tune-up featuring 30-40% performance improvement on most RISC
17749   platforms. See crypto/rc4/rc4_enc.c for further details.
17750
17751   *Andy Polyakov*
17752
17753 * New -noout option to asn1parse. This causes no output to be produced
17754   its main use is when combined with -strparse and -out to extract data
17755   from a file (which may not be in ASN.1 format).
17756
17757   *Steve Henson*
17758
17759 * Fix for pkcs12 program. It was hashing an invalid certificate pointer
17760   when producing the local key id.
17761
17762   *Richard Levitte <levitte@stacken.kth.se>*
17763
17764 * New option -dhparam in s_server. This allows a DH parameter file to be
17765   stated explicitly. If it is not stated then it tries the first server
17766   certificate file. The previous behaviour hard coded the filename
17767   "server.pem".
17768
17769   *Steve Henson*
17770
17771 * Add -pubin and -pubout options to the rsa and dsa commands. These allow
17772   a public key to be input or output. For example:
17773   openssl rsa -in key.pem -pubout -out pubkey.pem
17774   Also added necessary DSA public key functions to handle this.
17775
17776   *Steve Henson*
17777
17778 * Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
17779   in the message. This was handled by allowing
17780   X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
17781
17782   *Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>*
17783
17784 * Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
17785   to the end of the strings whereas this didn't. This would cause problems
17786   if strings read with d2i_ASN1_bytes() were later modified.
17787
17788   *Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>*
17789
17790 * Fix for base64 decode bug. When a base64 bio reads only one line of
17791   data and it contains EOF it will end up returning an error. This is
17792   caused by input 46 bytes long. The cause is due to the way base64
17793   BIOs find the start of base64 encoded data. They do this by trying a
17794   trial decode on each line until they find one that works. When they
17795   do a flag is set and it starts again knowing it can pass all the
17796   data directly through the decoder. Unfortunately it doesn't reset
17797   the context it uses. This means that if EOF is reached an attempt
17798   is made to pass two EOFs through the context and this causes the
17799   resulting error. This can also cause other problems as well. As is
17800   usual with these problems it takes *ages* to find and the fix is
17801   trivial: move one line.
17802
17803   *Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer)*
17804
17805 * Ugly workaround to get s_client and s_server working under Windows. The
17806   old code wouldn't work because it needed to select() on sockets and the
17807   tty (for keypresses and to see if data could be written). Win32 only
17808   supports select() on sockets so we select() with a 1s timeout on the
17809   sockets and then see if any characters are waiting to be read, if none
17810   are present then we retry, we also assume we can always write data to
17811   the tty. This isn't nice because the code then blocks until we've
17812   received a complete line of data and it is effectively polling the
17813   keyboard at 1s intervals: however it's quite a bit better than not
17814   working at all :-) A dedicated Windows application might handle this
17815   with an event loop for example.
17816
17817   *Steve Henson*
17818
17819 * Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
17820   and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
17821   will be called when RSA_sign() and RSA_verify() are used. This is useful
17822   if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
17823   For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
17824   should *not* be used: RSA_sign() and RSA_verify() must be used instead.
17825   This necessitated the support of an extra signature type NID_md5_sha1
17826   for SSL signatures and modifications to the SSL library to use it instead
17827   of calling RSA_public_decrypt() and RSA_private_encrypt().
17828
17829   *Steve Henson*
17830
17831 * Add new -verify -CAfile and -CApath options to the crl program, these
17832   will lookup a CRL issuers certificate and verify the signature in a
17833   similar way to the verify program. Tidy up the crl program so it
17834   no longer accesses structures directly. Make the ASN1 CRL parsing a bit
17835   less strict. It will now permit CRL extensions even if it is not
17836   a V2 CRL: this will allow it to tolerate some broken CRLs.
17837
17838   *Steve Henson*
17839
17840 * Initialize all non-automatic variables each time one of the openssl
17841   sub-programs is started (this is necessary as they may be started
17842   multiple times from the "OpenSSL>" prompt).
17843
17844   *Lennart Bang, Bodo Moeller*
17845
17846 * Preliminary compilation option RSA_NULL which disables RSA crypto without
17847   removing all other RSA functionality (this is what NO_RSA does). This
17848   is so (for example) those in the US can disable those operations covered
17849   by the RSA patent while allowing storage and parsing of RSA keys and RSA
17850   key generation.
17851
17852   *Steve Henson*
17853
17854 * Non-copying interface to BIO pairs.
17855   (still largely untested)
17856
17857   *Bodo Moeller*
17858
17859 * New function ASN1_tag2str() to convert an ASN1 tag to a descriptive
17860   ASCII string. This was handled independently in various places before.
17861
17862   *Steve Henson*
17863
17864 * New functions UTF8_getc() and UTF8_putc() that parse and generate
17865   UTF8 strings a character at a time.
17866
17867   *Steve Henson*
17868
17869 * Use client_version from client hello to select the protocol
17870   (s23_srvr.c) and for RSA client key exchange verification
17871   (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
17872
17873   *Bodo Moeller*
17874
17875 * Add various utility functions to handle SPKACs, these were previously
17876   handled by poking round in the structure internals. Added new function
17877   NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
17878   print, verify and generate SPKACs. Based on an original idea from
17879   Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
17880
17881   *Steve Henson*
17882
17883 * RIPEMD160 is operational on all platforms and is back in 'make test'.
17884
17885   *Andy Polyakov*
17886
17887 * Allow the config file extension section to be overwritten on the
17888   command line. Based on an original idea from Massimiliano Pala
17889   <madwolf@comune.modena.it>. The new option is called -extensions
17890   and can be applied to ca, req and x509. Also -reqexts to override
17891   the request extensions in req and -crlexts to override the crl extensions
17892   in ca.
17893
17894   *Steve Henson*
17895
17896 * Add new feature to the SPKAC handling in ca.  Now you can include
17897   the same field multiple times by preceding it by "XXXX." for example:
17898   1.OU="Unit name 1"
17899   2.OU="Unit name 2"
17900   this is the same syntax as used in the req config file.
17901
17902   *Steve Henson*
17903
17904 * Allow certificate extensions to be added to certificate requests. These
17905   are specified in a 'req_extensions' option of the req section of the
17906   config file. They can be printed out with the -text option to req but
17907   are otherwise ignored at present.
17908
17909   *Steve Henson*
17910
17911 * Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
17912   data read consists of only the final block it would not decrypted because
17913   EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
17914   A misplaced 'break' also meant the decrypted final block might not be
17915   copied until the next read.
17916
17917   *Steve Henson*
17918
17919 * Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
17920   a few extra parameters to the DH structure: these will be useful if
17921   for example we want the value of 'q' or implement X9.42 DH.
17922
17923   *Steve Henson*
17924
17925 * Initial support for DSA_METHOD. This is based on the RSA_METHOD and
17926   provides hooks that allow the default DSA functions or functions on a
17927   "per key" basis to be replaced. This allows hardware acceleration and
17928   hardware key storage to be handled without major modification to the
17929   library. Also added low-level modexp hooks and CRYPTO_EX structure and
17930   associated functions.
17931
17932   *Steve Henson*
17933
17934 * Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
17935   as "read only": it can't be written to and the buffer it points to will
17936   not be freed. Reading from a read only BIO is much more efficient than
17937   a normal memory BIO. This was added because there are several times when
17938   an area of memory needs to be read from a BIO. The previous method was
17939   to create a memory BIO and write the data to it, this results in two
17940   copies of the data and an O(n^2) reading algorithm. There is a new
17941   function BIO_new_mem_buf() which creates a read only memory BIO from
17942   an area of memory. Also modified the PKCS#7 routines to use read only
17943   memory BIOs.
17944
17945   *Steve Henson*
17946
17947 * Bugfix: ssl23_get_client_hello did not work properly when called in
17948   state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
17949   a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
17950   but a retry condition occurred while trying to read the rest.
17951
17952   *Bodo Moeller*
17953
17954 * The PKCS7_ENC_CONTENT_new() function was setting the content type as
17955   NID_pkcs7_encrypted by default: this was wrong since this should almost
17956   always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
17957   the encrypted data type: this is a more sensible place to put it and it
17958   allows the PKCS#12 code to be tidied up that duplicated this
17959   functionality.
17960
17961   *Steve Henson*
17962
17963 * Changed obj_dat.pl script so it takes its input and output files on
17964   the command line. This should avoid shell escape redirection problems
17965   under Win32.
17966
17967   *Steve Henson*
17968
17969 * Initial support for certificate extension requests, these are included
17970   in things like Xenroll certificate requests. Included functions to allow
17971   extensions to be obtained and added.
17972
17973   *Steve Henson*
17974
17975 * -crlf option to s_client and s_server for sending newlines as
17976   CRLF (as required by many protocols).
17977
17978   *Bodo Moeller*
17979
17980### Changes between 0.9.3a and 0.9.4  [09 Aug 1999]
17981
17982 * Install libRSAglue.a when OpenSSL is built with RSAref.
17983
17984   *Ralf S. Engelschall*
17985
17986 * A few more `#ifndef NO_FP_API / #endif` pairs for consistency.
17987
17988   *Andrija Antonijevic <TheAntony2@bigfoot.com>*
17989
17990 * Fix -startdate and -enddate (which was missing) arguments to 'ca'
17991   program.
17992
17993   *Steve Henson*
17994
17995 * New function DSA_dup_DH, which duplicates DSA parameters/keys as
17996   DH parameters/keys (q is lost during that conversion, but the resulting
17997   DH parameters contain its length).
17998
17999   For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
18000   much faster than DH_generate_parameters (which creates parameters
18001   where `p = 2*q + 1`), and also the smaller q makes DH computations
18002   much more efficient (160-bit exponentiation instead of 1024-bit
18003   exponentiation); so this provides a convenient way to support DHE
18004   ciphersuites in SSL/TLS servers (see ssl/ssltest.c).  It is of
18005   utter importance to use
18006           SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
18007   or
18008           SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
18009   when such DH parameters are used, because otherwise small subgroup
18010   attacks may become possible!
18011
18012   *Bodo Moeller*
18013
18014 * Avoid memory leak in i2d_DHparams.
18015
18016   *Bodo Moeller*
18017
18018 * Allow the -k option to be used more than once in the enc program:
18019   this allows the same encrypted message to be read by multiple recipients.
18020
18021   *Steve Henson*
18022
18023 * New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
18024   an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
18025   it will always use the numerical form of the OID, even if it has a short
18026   or long name.
18027
18028   *Steve Henson*
18029
18030 * Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
18031   method only got called if p,q,dmp1,dmq1,iqmp components were present,
18032   otherwise bn_mod_exp was called. In the case of hardware keys for example
18033   no private key components need be present and it might store extra data
18034   in the RSA structure, which cannot be accessed from bn_mod_exp.
18035   By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
18036   private key operations.
18037
18038   *Steve Henson*
18039
18040 * Added support for SPARC Linux.
18041
18042   *Andy Polyakov*
18043
18044 * pem_password_cb function type incompatibly changed from
18045           typedef int pem_password_cb(char *buf, int size, int rwflag);
18046   to
18047           ....(char *buf, int size, int rwflag, void *userdata);
18048   so that applications can pass data to their callbacks:
18049   The `PEM[_ASN1]_{read,write}...` functions and macros now take an
18050   additional void * argument, which is just handed through whenever
18051   the password callback is called.
18052
18053   *Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller*
18054
18055   New function SSL_CTX_set_default_passwd_cb_userdata.
18056
18057   Compatibility note: As many C implementations push function arguments
18058   onto the stack in reverse order, the new library version is likely to
18059   interoperate with programs that have been compiled with the old
18060   pem_password_cb definition (PEM_whatever takes some data that
18061   happens to be on the stack as its last argument, and the callback
18062   just ignores this garbage); but there is no guarantee whatsoever that
18063   this will work.
18064
18065 * The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
18066   (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
18067   problems not only on Windows, but also on some Unix platforms.
18068   To avoid problematic command lines, these definitions are now in an
18069   auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
18070   for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
18071
18072   *Bodo Moeller*
18073
18074 * MIPS III/IV assembler module is reimplemented.
18075
18076   *Andy Polyakov*
18077
18078 * More DES library cleanups: remove references to srand/rand and
18079   delete an unused file.
18080
18081   *Ulf Möller*
18082
18083 * Add support for the free Netwide assembler (NASM) under Win32,
18084   since not many people have MASM (ml) and it can be hard to obtain.
18085   This is currently experimental but it seems to work OK and pass all
18086   the tests. Check out INSTALL.W32 for info.
18087
18088   *Steve Henson*
18089
18090 * Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
18091   without temporary keys kept an extra copy of the server key,
18092   and connections with temporary keys did not free everything in case
18093   of an error.
18094
18095   *Bodo Moeller*
18096
18097 * New function RSA_check_key and new openssl rsa option -check
18098   for verifying the consistency of RSA keys.
18099
18100   *Ulf Moeller, Bodo Moeller*
18101
18102 * Various changes to make Win32 compile work:
18103   1. Casts to avoid "loss of data" warnings in p5_crpt2.c
18104   2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
18105      comparison" warnings.
18106   3. Add `sk_<TYPE>_sort` to DEF file generator and do make update.
18107
18108   *Steve Henson*
18109
18110 * Add a debugging option to PKCS#5 v2 key generation function: when
18111   you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
18112   derived keys are printed to stderr.
18113
18114   *Steve Henson*
18115
18116 * Copy the flags in ASN1_STRING_dup().
18117
18118   *Roman E. Pavlov <pre@mo.msk.ru>*
18119
18120 * The x509 application mishandled signing requests containing DSA
18121   keys when the signing key was also DSA and the parameters didn't match.
18122
18123   It was supposed to omit the parameters when they matched the signing key:
18124   the verifying software was then supposed to automatically use the CA's
18125   parameters if they were absent from the end user certificate.
18126
18127   Omitting parameters is no longer recommended. The test was also
18128   the wrong way round! This was probably due to unusual behaviour in
18129   EVP_cmp_parameters() which returns 1 if the parameters match.
18130   This meant that parameters were omitted when they *didn't* match and
18131   the certificate was useless. Certificates signed with 'ca' didn't have
18132   this bug.
18133
18134   *Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>*
18135
18136 * Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
18137   The interface is as follows:
18138   Applications can use
18139           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
18140           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
18141   "off" is now the default.
18142   The library internally uses
18143           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
18144           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
18145   to disable memory-checking temporarily.
18146
18147   Some inconsistent states that previously were possible (and were
18148   even the default) are now avoided.
18149
18150   -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
18151   with each memory chunk allocated; this is occasionally more helpful
18152   than just having a counter.
18153
18154   -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
18155
18156   -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
18157   extensions.
18158
18159   *Bodo Moeller*
18160
18161 * Introduce "mode" for SSL structures (with defaults in SSL_CTX),
18162   which largely parallels "options", but is for changing API behaviour,
18163   whereas "options" are about protocol behaviour.
18164   Initial "mode" flags are:
18165
18166   SSL_MODE_ENABLE_PARTIAL_WRITE   Allow SSL_write to report success when
18167                                   a single record has been written.
18168   SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER  Don't insist that SSL_write
18169                                   retries use the same buffer location.
18170                                   (But all of the contents must be
18171                                   copied!)
18172
18173   *Bodo Moeller*
18174
18175 * Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
18176   worked.
18177
18178 * Fix problems with no-hmac etc.
18179
18180   *Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>*
18181
18182 * New functions RSA_get_default_method(), RSA_set_method() and
18183   RSA_get_method(). These allows replacement of RSA_METHODs without having
18184   to mess around with the internals of an RSA structure.
18185
18186   *Steve Henson*
18187
18188 * Fix memory leaks in DSA_do_sign and DSA_is_prime.
18189   Also really enable memory leak checks in openssl.c and in some
18190   test programs.
18191
18192   *Chad C. Mulligan, Bodo Moeller*
18193
18194 * Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
18195   up the length of negative integers. This has now been simplified to just
18196   store the length when it is first determined and use it later, rather
18197   than trying to keep track of where data is copied and updating it to
18198   point to the end.
18199   *Steve Henson, reported by Brien Wheeler <bwheeler@authentica-security.com>*
18200
18201 * Add a new function PKCS7_signatureVerify. This allows the verification
18202   of a PKCS#7 signature but with the signing certificate passed to the
18203   function itself. This contrasts with PKCS7_dataVerify which assumes the
18204   certificate is present in the PKCS#7 structure. This isn't always the
18205   case: certificates can be omitted from a PKCS#7 structure and be
18206   distributed by "out of band" means (such as a certificate database).
18207
18208   *Steve Henson*
18209
18210 * Complete the `PEM_*` macros with DECLARE_PEM versions to replace the
18211   function prototypes in pem.h, also change util/mkdef.pl to add the
18212   necessary function names.
18213
18214   *Steve Henson*
18215
18216 * mk1mf.pl (used by Windows builds) did not properly read the
18217   options set by Configure in the top level Makefile, and Configure
18218   was not even able to write more than one option correctly.
18219   Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
18220
18221   *Bodo Moeller*
18222
18223 * New functions CONF_load_bio() and CONF_load_fp() to allow a config
18224   file to be loaded from a BIO or FILE pointer. The BIO version will
18225   for example allow memory BIOs to contain config info.
18226
18227   *Steve Henson*
18228
18229 * New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
18230   Whoever hopes to achieve shared-library compatibility across versions
18231   must use this, not the compile-time macro.
18232   (Exercise 0.9.4: Which is the minimum library version required by
18233   such programs?)
18234   Note: All this applies only to multi-threaded programs, others don't
18235   need locks.
18236
18237   *Bodo Moeller*
18238
18239 * Add missing case to s3_clnt.c state machine -- one of the new SSL tests
18240   through a BIO pair triggered the default case, i.e.
18241   SSLerr(...,SSL_R_UNKNOWN_STATE).
18242
18243   *Bodo Moeller*
18244
18245 * New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
18246   can use the SSL library even if none of the specific BIOs is
18247   appropriate.
18248
18249   *Bodo Moeller*
18250
18251 * Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
18252   for the encoded length.
18253
18254   *Jeon KyoungHo <khjeon@sds.samsung.co.kr>*
18255
18256 * Add initial documentation of the X509V3 functions.
18257
18258   *Steve Henson*
18259
18260 * Add a new pair of functions PEM_write_PKCS8PrivateKey() and
18261   PEM_write_bio_PKCS8PrivateKey() that are equivalent to
18262   PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
18263   secure PKCS#8 private key format with a high iteration count.
18264
18265   *Steve Henson*
18266
18267 * Fix determination of Perl interpreter: A perl or perl5
18268   *directory* in $PATH was also accepted as the interpreter.
18269
18270   *Ralf S. Engelschall*
18271
18272 * Fix demos/sign/sign.c: well there wasn't anything strictly speaking
18273   wrong with it but it was very old and did things like calling
18274   PEM_ASN1_read() directly and used MD5 for the hash not to mention some
18275   unusual formatting.
18276
18277   *Steve Henson*
18278
18279 * Fix demos/selfsign.c: it used obsolete and deleted functions, changed
18280   to use the new extension code.
18281
18282   *Steve Henson*
18283
18284 * Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
18285   with macros. This should make it easier to change their form, add extra
18286   arguments etc. Fix a few PEM prototypes which didn't have cipher as a
18287   constant.
18288
18289   *Steve Henson*
18290
18291 * Add to configuration table a new entry that can specify an alternative
18292   name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
18293   according to Mark Crispin <MRC@Panda.COM>.
18294
18295   *Bodo Moeller*
18296
18297 * DES CBC did not update the IV. Weird.
18298
18299   *Ben Laurie*
18300lse
18301   des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
18302   Changing the behaviour of the former might break existing programs --
18303   where IV updating is needed, des_ncbc_encrypt can be used.
18304ndif
18305
18306 * When bntest is run from "make test" it drives bc to check its
18307   calculations, as well as internally checking them. If an internal check
18308   fails, it needs to cause bc to give a non-zero result or make test carries
18309   on without noticing the failure. Fixed.
18310
18311   *Ben Laurie*
18312
18313 * DES library cleanups.
18314
18315   *Ulf Möller*
18316
18317 * Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
18318   used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
18319   ciphers. NOTE: although the key derivation function has been verified
18320   against some published test vectors it has not been extensively tested
18321   yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
18322   of v2.0.
18323
18324   *Steve Henson*
18325
18326 * Instead of "mkdir -p", which is not fully portable, use new
18327   Perl script "util/mkdir-p.pl".
18328
18329   *Bodo Moeller*
18330
18331 * Rewrite the way password based encryption (PBE) is handled. It used to
18332   assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
18333   structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
18334   but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
18335   the 'parameter' field of the AlgorithmIdentifier is passed to the
18336   underlying key generation function so it must do its own ASN1 parsing.
18337   This has also changed the EVP_PBE_CipherInit() function which now has a
18338   'parameter' argument instead of literal salt and iteration count values
18339   and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
18340
18341   *Steve Henson*
18342
18343 * Support for PKCS#5 v1.5 compatible password based encryption algorithms
18344   and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
18345   Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
18346   KEY" because this clashed with PKCS#8 unencrypted string. Since this
18347   value was just used as a "magic string" and not used directly its
18348   value doesn't matter.
18349
18350   *Steve Henson*
18351
18352 * Introduce some semblance of const correctness to BN. Shame C doesn't
18353   support mutable.
18354
18355   *Ben Laurie*
18356
18357 * "linux-sparc64" configuration (ultrapenguin).
18358
18359   *Ray Miller <ray.miller@oucs.ox.ac.uk>*
18360   "linux-sparc" configuration.
18361
18362   *Christian Forster <fo@hawo.stw.uni-erlangen.de>*
18363
18364 * config now generates no-xxx options for missing ciphers.
18365
18366   *Ulf Möller*
18367
18368 * Support the EBCDIC character set (work in progress).
18369   File ebcdic.c not yet included because it has a different license.
18370
18371   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
18372
18373 * Support BS2000/OSD-POSIX.
18374
18375   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
18376
18377 * Make callbacks for key generation use `void *` instead of `char *`.
18378
18379   *Ben Laurie*
18380
18381 * Make S/MIME samples compile (not yet tested).
18382
18383   *Ben Laurie*
18384
18385 * Additional typesafe stacks.
18386
18387   *Ben Laurie*
18388
18389 * New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
18390
18391   *Bodo Moeller*
18392
18393### Changes between 0.9.3 and 0.9.3a  [29 May 1999]
18394
18395 * New configuration variant "sco5-gcc".
18396
18397 * Updated some demos.
18398
18399   *Sean O Riordain, Wade Scholine*
18400
18401 * Add missing BIO_free at exit of pkcs12 application.
18402
18403   *Wu Zhigang*
18404
18405 * Fix memory leak in conf.c.
18406
18407   *Steve Henson*
18408
18409 * Updates for Win32 to assembler version of MD5.
18410
18411   *Steve Henson*
18412
18413 * Set #! path to perl in `apps/der_chop` to where we found it
18414   instead of using a fixed path.
18415
18416   *Bodo Moeller*
18417
18418 * SHA library changes for irix64-mips4-cc.
18419
18420   *Andy Polyakov*
18421
18422 * Improvements for VMS support.
18423
18424   *Richard Levitte*
18425
18426### Changes between 0.9.2b and 0.9.3  [24 May 1999]
18427
18428 * Bignum library bug fix. IRIX 6 passes "make test" now!
18429   This also avoids the problems with SC4.2 and unpatched SC5.
18430
18431   *Andy Polyakov <appro@fy.chalmers.se>*
18432
18433 * New functions sk_num, sk_value and sk_set to replace the previous macros.
18434   These are required because of the typesafe stack would otherwise break
18435   existing code. If old code used a structure member which used to be STACK
18436   and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
18437   sk_num or sk_value it would produce an error because the num, data members
18438   are not present in STACK_OF. Now it just produces a warning. sk_set
18439   replaces the old method of assigning a value to sk_value
18440   (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
18441   that does this will no longer work (and should use sk_set instead) but
18442   this could be regarded as a "questionable" behaviour anyway.
18443
18444   *Steve Henson*
18445
18446 * Fix most of the other PKCS#7 bugs. The "experimental" code can now
18447   correctly handle encrypted S/MIME data.
18448
18449   *Steve Henson*
18450
18451 * Change type of various DES function arguments from des_cblock
18452   (which means, in function argument declarations, pointer to char)
18453   to des_cblock * (meaning pointer to array with 8 char elements),
18454   which allows the compiler to do more typechecking; it was like
18455   that back in SSLeay, but with lots of ugly casts.
18456
18457   Introduce new type const_des_cblock.
18458
18459   *Bodo Moeller*
18460
18461 * Reorganise the PKCS#7 library and get rid of some of the more obvious
18462   problems: find RecipientInfo structure that matches recipient certificate
18463   and initialise the ASN1 structures properly based on passed cipher.
18464
18465   *Steve Henson*
18466
18467 * Belatedly make the BN tests actually check the results.
18468
18469   *Ben Laurie*
18470
18471 * Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
18472   to and from BNs: it was completely broken. New compilation option
18473   NEG_PUBKEY_BUG to allow for some broken certificates that encode public
18474   key elements as negative integers.
18475
18476   *Steve Henson*
18477
18478 * Reorganize and speed up MD5.
18479
18480   *Andy Polyakov <appro@fy.chalmers.se>*
18481
18482 * VMS support.
18483
18484   *Richard Levitte <richard@levitte.org>*
18485
18486 * New option -out to asn1parse to allow the parsed structure to be
18487   output to a file. This is most useful when combined with the -strparse
18488   option to examine the output of things like OCTET STRINGS.
18489
18490   *Steve Henson*
18491
18492 * Make SSL library a little more fool-proof by not requiring any longer
18493   that `SSL_set_{accept,connect}_state` be called before
18494   `SSL_{accept,connect}` may be used (`SSL_set_..._state` is omitted
18495   in many applications because usually everything *appeared* to work as
18496   intended anyway -- now it really works as intended).
18497
18498   *Bodo Moeller*
18499
18500 * Move openssl.cnf out of lib/.
18501
18502   *Ulf Möller*
18503
18504 * Fix various things to let OpenSSL even pass "egcc -pipe -O2 -Wall
18505   -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
18506   -Wmissing-declarations -Wnested-externs -Winline" with EGCS 1.1.2+
18507
18508   *Ralf S. Engelschall*
18509
18510 * Various fixes to the EVP and PKCS#7 code. It may now be able to
18511   handle PKCS#7 enveloped data properly.
18512
18513   *Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve*
18514
18515 * Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
18516   copying pointers.  The cert_st handling is changed by this in
18517   various ways (and thus what used to be known as ctx->default_cert
18518   is now called ctx->cert, since we don't resort to `s->ctx->[default_]cert`
18519   any longer when s->cert does not give us what we need).
18520   ssl_cert_instantiate becomes obsolete by this change.
18521   As soon as we've got the new code right (possibly it already is?),
18522   we have solved a couple of bugs of the earlier code where s->cert
18523   was used as if it could not have been shared with other SSL structures.
18524
18525   Note that using the SSL API in certain dirty ways now will result
18526   in different behaviour than observed with earlier library versions:
18527   Changing settings for an `SSL_CTX *ctx` after having done s = SSL_new(ctx)
18528   does not influence s as it used to.
18529
18530   In order to clean up things more thoroughly, inside SSL_SESSION
18531   we don't use CERT any longer, but a new structure SESS_CERT
18532   that holds per-session data (if available); currently, this is
18533   the peer's certificate chain and, for clients, the server's certificate
18534   and temporary key.  CERT holds only those values that can have
18535   meaningful defaults in an SSL_CTX.
18536
18537   *Bodo Moeller*
18538
18539 * New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
18540   from the internal representation. Various PKCS#7 fixes: remove some
18541   evil casts and set the enc_dig_alg field properly based on the signing
18542   key type.
18543
18544   *Steve Henson*
18545
18546 * Allow PKCS#12 password to be set from the command line or the
18547   environment. Let 'ca' get its config file name from the environment
18548   variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
18549   and 'x509').
18550
18551   *Steve Henson*
18552
18553 * Allow certificate policies extension to use an IA5STRING for the
18554   organization field. This is contrary to the PKIX definition but
18555   VeriSign uses it and IE5 only recognises this form. Document 'x509'
18556   extension option.
18557
18558   *Steve Henson*
18559
18560 * Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
18561   without disallowing inline assembler and the like for non-pedantic builds.
18562
18563   *Ben Laurie*
18564
18565 * Support Borland C++ builder.
18566
18567   *Janez Jere <jj@void.si>, modified by Ulf Möller*
18568
18569 * Support Mingw32.
18570
18571   *Ulf Möller*
18572
18573 * SHA-1 cleanups and performance enhancements.
18574
18575   *Andy Polyakov <appro@fy.chalmers.se>*
18576
18577 * Sparc v8plus assembler for the bignum library.
18578
18579   *Andy Polyakov <appro@fy.chalmers.se>*
18580
18581 * Accept any -xxx and +xxx compiler options in Configure.
18582
18583   *Ulf Möller*
18584
18585 * Update HPUX configuration.
18586
18587   *Anonymous*
18588
18589 * Add missing `sk_<type>_unshift()` function to safestack.h
18590
18591   *Ralf S. Engelschall*
18592
18593 * New function SSL_CTX_use_certificate_chain_file that sets the
18594   "extra_cert"s in addition to the certificate.  (This makes sense
18595   only for "PEM" format files, as chains as a whole are not
18596   DER-encoded.)
18597
18598   *Bodo Moeller*
18599
18600 * Support verify_depth from the SSL API.
18601   x509_vfy.c had what can be considered an off-by-one-error:
18602   Its depth (which was not part of the external interface)
18603   was actually counting the number of certificates in a chain;
18604   now it really counts the depth.
18605
18606   *Bodo Moeller*
18607
18608 * Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
18609   instead of X509err, which often resulted in confusing error
18610   messages since the error codes are not globally unique
18611   (e.g. an alleged error in ssl3_accept when a certificate
18612   didn't match the private key).
18613
18614 * New function SSL_CTX_set_session_id_context that allows to set a default
18615   value (so that you don't need SSL_set_session_id_context for each
18616   connection using the SSL_CTX).
18617
18618   *Bodo Moeller*
18619
18620 * OAEP decoding bug fix.
18621
18622   *Ulf Möller*
18623
18624 * Support INSTALL_PREFIX for package builders, as proposed by
18625   David Harris.
18626
18627   *Bodo Moeller*
18628
18629 * New Configure options "threads" and "no-threads".  For systems
18630   where the proper compiler options are known (currently Solaris
18631   and Linux), "threads" is the default.
18632
18633   *Bodo Moeller*
18634
18635 * New script util/mklink.pl as a faster substitute for util/mklink.sh.
18636
18637   *Bodo Moeller*
18638
18639 * Install various scripts to $(OPENSSLDIR)/misc, not to
18640   $(INSTALLTOP)/bin -- they shouldn't clutter directories
18641   such as /usr/local/bin.
18642
18643   *Bodo Moeller*
18644
18645 * "make linux-shared" to build shared libraries.
18646
18647   *Niels Poppe <niels@netbox.org>*
18648
18649 * New Configure option `no-<cipher>` (rsa, idea, rc5, ...).
18650
18651   *Ulf Möller*
18652
18653 * Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
18654   extension adding in x509 utility.
18655
18656   *Steve Henson*
18657
18658 * Remove NOPROTO sections and error code comments.
18659
18660   *Ulf Möller*
18661
18662 * Partial rewrite of the DEF file generator to now parse the ANSI
18663   prototypes.
18664
18665   *Steve Henson*
18666
18667 * New Configure options --prefix=DIR and --openssldir=DIR.
18668
18669   *Ulf Möller*
18670
18671 * Complete rewrite of the error code script(s). It is all now handled
18672   by one script at the top level which handles error code gathering,
18673   header rewriting and C source file generation. It should be much better
18674   than the old method: it now uses a modified version of Ulf's parser to
18675   read the ANSI prototypes in all header files (thus the old K&R definitions
18676   aren't needed for error creation any more) and do a better job of
18677   translating function codes into names. The old 'ASN1 error code embedded
18678   in a comment' is no longer necessary and it doesn't use .err files which
18679   have now been deleted. Also the error code call doesn't have to appear all
18680   on one line (which resulted in some large lines...).
18681
18682   *Steve Henson*
18683
18684 * Change #include filenames from `<foo.h>` to `<openssl/foo.h>`.
18685
18686   *Bodo Moeller*
18687
18688 * Change behaviour of ssl2_read when facing length-0 packets: Don't return
18689   0 (which usually indicates a closed connection), but continue reading.
18690
18691   *Bodo Moeller*
18692
18693 * Fix some race conditions.
18694
18695   *Bodo Moeller*
18696
18697 * Add support for CRL distribution points extension. Add Certificate
18698   Policies and CRL distribution points documentation.
18699
18700   *Steve Henson*
18701
18702 * Move the autogenerated header file parts to crypto/opensslconf.h.
18703
18704   *Ulf Möller*
18705
18706 * Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
18707   8 of keying material. Merlin has also confirmed interop with this fix
18708   between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
18709
18710   *Merlin Hughes <merlin@baltimore.ie>*
18711
18712 * Fix lots of warnings.
18713
18714   *Richard Levitte <levitte@stacken.kth.se>*
18715
18716 * In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
18717   the directory spec didn't end with a LIST_SEPARATOR_CHAR.
18718
18719   *Richard Levitte <levitte@stacken.kth.se>*
18720
18721 * Fix problems with sizeof(long) == 8.
18722
18723   *Andy Polyakov <appro@fy.chalmers.se>*
18724
18725 * Change functions to ANSI C.
18726
18727   *Ulf Möller*
18728
18729 * Fix typos in error codes.
18730
18731   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller*
18732
18733 * Remove defunct assembler files from Configure.
18734
18735   *Ulf Möller*
18736
18737 * SPARC v8 assembler BIGNUM implementation.
18738
18739   *Andy Polyakov <appro@fy.chalmers.se>*
18740
18741 * Support for Certificate Policies extension: both print and set.
18742   Various additions to support the r2i method this uses.
18743
18744   *Steve Henson*
18745
18746 * A lot of constification, and fix a bug in X509_NAME_oneline() that could
18747   return a const string when you are expecting an allocated buffer.
18748
18749   *Ben Laurie*
18750
18751 * Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
18752   types DirectoryString and DisplayText.
18753
18754   *Steve Henson*
18755
18756 * Add code to allow r2i extensions to access the configuration database,
18757   add an LHASH database driver and add several ctx helper functions.
18758
18759   *Steve Henson*
18760
18761 * Fix an evil bug in bn_expand2() which caused various BN functions to
18762   fail when they extended the size of a BIGNUM.
18763
18764   *Steve Henson*
18765
18766 * Various utility functions to handle SXNet extension. Modify mkdef.pl to
18767   support typesafe stack.
18768
18769   *Steve Henson*
18770
18771 * Fix typo in SSL_[gs]et_options().
18772
18773   *Nils Frostberg <nils@medcom.se>*
18774
18775 * Delete various functions and files that belonged to the (now obsolete)
18776   old X509V3 handling code.
18777
18778   *Steve Henson*
18779
18780 * New Configure option "rsaref".
18781
18782   *Ulf Möller*
18783
18784 * Don't auto-generate pem.h.
18785
18786   *Bodo Moeller*
18787
18788 * Introduce type-safe ASN.1 SETs.
18789
18790   *Ben Laurie*
18791
18792 * Convert various additional casted stacks to type-safe STACK_OF() variants.
18793
18794   *Ben Laurie, Ralf S. Engelschall, Steve Henson*
18795
18796 * Introduce type-safe STACKs. This will almost certainly break lots of code
18797   that links with OpenSSL (well at least cause lots of warnings), but fear
18798   not: the conversion is trivial, and it eliminates loads of evil casts. A
18799   few STACKed things have been converted already. Feel free to convert more.
18800   In the fullness of time, I'll do away with the STACK type altogether.
18801
18802   *Ben Laurie*
18803
18804 * Add `openssl ca -revoke <certfile>` facility which revokes a certificate
18805   specified in `<certfile>` by updating the entry in the index.txt file.
18806   This way one no longer has to edit the index.txt file manually for
18807   revoking a certificate. The -revoke option does the gory details now.
18808
18809   *Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall*
18810
18811 * Fix `openssl crl -noout -text` combination where `-noout` killed the
18812   `-text` option at all and this way the `-noout -text` combination was
18813   inconsistent in `openssl crl` with the friends in `openssl x509|rsa|dsa`.
18814
18815   *Ralf S. Engelschall*
18816
18817 * Make sure a corresponding plain text error message exists for the
18818   X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
18819   verify callback function determined that a certificate was revoked.
18820
18821   *Ralf S. Engelschall*
18822
18823 * Bugfix: In test/testenc, don't test `openssl <cipher>` for
18824   ciphers that were excluded, e.g. by -DNO_IDEA.  Also, test
18825   all available ciphers including rc5, which was forgotten until now.
18826   In order to let the testing shell script know which algorithms
18827   are available, a new (up to now undocumented) command
18828   `openssl list-cipher-commands` is used.
18829
18830   *Bodo Moeller*
18831
18832 * Bugfix: s_client occasionally would sleep in select() when
18833   it should have checked SSL_pending() first.
18834
18835   *Bodo Moeller*
18836
18837 * New functions DSA_do_sign and DSA_do_verify to provide access to
18838   the raw DSA values prior to ASN.1 encoding.
18839
18840   *Ulf Möller*
18841
18842 * Tweaks to Configure
18843
18844   *Niels Poppe <niels@netbox.org>*
18845
18846 * Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
18847   yet...
18848
18849   *Steve Henson*
18850
18851 * New variables $(RANLIB) and $(PERL) in the Makefiles.
18852
18853   *Ulf Möller*
18854
18855 * New config option to avoid instructions that are illegal on the 80386.
18856   The default code is faster, but requires at least a 486.
18857
18858   *Ulf Möller*
18859
18860 * Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
18861   SSL2_SERVER_VERSION (not used at all) macros, which are now the
18862   same as SSL2_VERSION anyway.
18863
18864   *Bodo Moeller*
18865
18866 * New "-showcerts" option for s_client.
18867
18868   *Bodo Moeller*
18869
18870 * Still more PKCS#12 integration. Add pkcs12 application to openssl
18871   application. Various cleanups and fixes.
18872
18873   *Steve Henson*
18874
18875 * More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
18876   modify error routines to work internally. Add error codes and PBE init
18877   to library startup routines.
18878
18879   *Steve Henson*
18880
18881 * Further PKCS#12 integration. Added password based encryption, PKCS#8 and
18882   packing functions to asn1 and evp. Changed function names and error
18883   codes along the way.
18884
18885   *Steve Henson*
18886
18887 * PKCS12 integration: and so it begins... First of several patches to
18888   slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
18889   objects to objects.h
18890
18891   *Steve Henson*
18892
18893 * Add a new 'indent' option to some X509V3 extension code. Initial ASN1
18894   and display support for Thawte strong extranet extension.
18895
18896   *Steve Henson*
18897
18898 * Add LinuxPPC support.
18899
18900   *Jeff Dubrule <igor@pobox.org>*
18901
18902 * Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
18903   bn_div_words in alpha.s.
18904
18905   *Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie*
18906
18907 * Make sure the RSA OAEP test is skipped under -DRSAref because
18908   OAEP isn't supported when OpenSSL is built with RSAref.
18909
18910   *Ulf Moeller <ulf@fitug.de>*
18911
18912 * Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h
18913   so they no longer are missing under -DNOPROTO.
18914
18915   *Soren S. Jorvang <soren@t.dk>*
18916
18917### Changes between 0.9.1c and 0.9.2b  [22 Mar 1999]
18918
18919 * Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
18920   doesn't work when the session is reused. Coming soon!
18921
18922   *Ben Laurie*
18923
18924 * Fix a security hole, that allows sessions to be reused in the wrong
18925   context thus bypassing client cert protection! All software that uses
18926   client certs and session caches in multiple contexts NEEDS PATCHING to
18927   allow session reuse! A fuller solution is in the works.
18928
18929   *Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)*
18930
18931 * Some more source tree cleanups (removed obsolete files
18932   crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
18933   permission on "config" script to be executable) and a fix for the INSTALL
18934   document.
18935
18936   *Ulf Moeller <ulf@fitug.de>*
18937
18938 * Remove some legacy and erroneous uses of malloc, free instead of
18939   Malloc, Free.
18940
18941   *Lennart Bang <lob@netstream.se>, with minor changes by Steve*
18942
18943 * Make rsa_oaep_test return non-zero on error.
18944
18945   *Ulf Moeller <ulf@fitug.de>*
18946
18947 * Add support for native Solaris shared libraries. Configure
18948   solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
18949   if someone would make that last step automatic.
18950
18951   *Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>*
18952
18953 * ctx_size was not built with the right compiler during "make links". Fixed.
18954
18955   *Ben Laurie*
18956
18957 * Change the meaning of 'ALL' in the cipher list. It now means "everything
18958   except NULL ciphers". This means the default cipher list will no longer
18959   enable NULL ciphers. They need to be specifically enabled e.g. with
18960   the string "DEFAULT:eNULL".
18961
18962   *Steve Henson*
18963
18964 * Fix to RSA private encryption routines: if p < q then it would
18965   occasionally produce an invalid result. This will only happen with
18966   externally generated keys because OpenSSL (and SSLeay) ensure p > q.
18967
18968   *Steve Henson*
18969
18970 * Be less restrictive and allow also `perl util/perlpath.pl
18971   /path/to/bin/perl` in addition to `perl util/perlpath.pl /path/to/bin`,
18972   because this way one can also use an interpreter named `perl5` (which is
18973   usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
18974   installed as `perl`).
18975
18976   *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
18977
18978 * Let util/clean-depend.pl work also with older Perl 5.00x versions.
18979
18980   *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
18981
18982 * Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
18983   advapi32.lib to Win32 build and change the pem test comparison
18984   to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
18985   suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
18986   and crypto/des/ede_cbcm_enc.c.
18987
18988   *Steve Henson*
18989
18990 * DES quad checksum was broken on big-endian architectures. Fixed.
18991
18992   *Ben Laurie*
18993
18994 * Comment out two functions in bio.h that aren't implemented. Fix up the
18995   Win32 test batch file so it (might) work again. The Win32 test batch file
18996   is horrible: I feel ill....
18997
18998   *Steve Henson*
18999
19000 * Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
19001   in e_os.h. Audit of header files to check ANSI and non ANSI
19002   sections: 10 functions were absent from non ANSI section and not exported
19003   from Windows DLLs. Fixed up libeay.num for new functions.
19004
19005   *Steve Henson*
19006
19007 * Make `openssl version` output lines consistent.
19008
19009   *Ralf S. Engelschall*
19010
19011 * Fix Win32 symbol export lists for BIO functions: Added
19012   BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
19013   to ms/libeay{16,32}.def.
19014
19015   *Ralf S. Engelschall*
19016
19017 * Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
19018   fine under Unix and passes some trivial tests I've now added. But the
19019   whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
19020   added to make sure no one expects that this stuff really works in the
19021   OpenSSL 0.9.2 release.  Additionally I've started to clean the XS sources
19022   up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
19023   openssl_bio.xs.
19024
19025   *Ralf S. Engelschall*
19026
19027 * Fix the generation of two part addresses in perl.
19028
19029   *Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie*
19030
19031 * Add config entry for Linux on MIPS.
19032
19033   *John Tobey <jtobey@channel1.com>*
19034
19035 * Make links whenever Configure is run, unless we are on Windoze.
19036
19037   *Ben Laurie*
19038
19039 * Permit extensions to be added to CRLs using crl_section in openssl.cnf.
19040   Currently only issuerAltName and AuthorityKeyIdentifier make any sense
19041   in CRLs.
19042
19043   *Steve Henson*
19044
19045 * Add a useful kludge to allow package maintainers to specify compiler and
19046   other platforms details on the command line without having to patch the
19047   Configure script every time: One now can use
19048   `perl Configure <id>:<details>`,
19049   i.e. platform ids are allowed to have details appended
19050   to them (separated by colons). This is treated as there would be a static
19051   pre-configured entry in Configure's %table under key `<id>` with value
19052   `<details>` and `perl Configure <id>` is called.  So, when you want to
19053   perform a quick test-compile under FreeBSD 3.1 with pgcc and without
19054   assembler stuff you can use `perl Configure "FreeBSD-elf:pgcc:-O6:::"`
19055   now, which overrides the FreeBSD-elf entry on-the-fly.
19056
19057   *Ralf S. Engelschall*
19058
19059 * Disable new TLS1 ciphersuites by default: they aren't official yet.
19060
19061   *Ben Laurie*
19062
19063 * Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
19064   on the `perl Configure ...` command line. This way one can compile
19065   OpenSSL libraries with Position Independent Code (PIC) which is needed
19066   for linking it into DSOs.
19067
19068   *Ralf S. Engelschall*
19069
19070 * Remarkably, export ciphers were totally broken and no-one had noticed!
19071   Fixed.
19072
19073   *Ben Laurie*
19074
19075 * Cleaned up the LICENSE document: The official contact for any license
19076   questions now is the OpenSSL core team under openssl-core@openssl.org.
19077   And add a paragraph about the dual-license situation to make sure people
19078   recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
19079   to the OpenSSL toolkit.
19080
19081   *Ralf S. Engelschall*
19082
19083 * General source tree makefile cleanups: Made `making xxx in yyy...`
19084   display consistent in the source tree and replaced `/bin/rm` by `rm`.
19085   Additionally cleaned up the `make links` target: Remove unnecessary
19086   semicolons, subsequent redundant removes, inline point.sh into mklink.sh
19087   to speed processing and no longer clutter the display with confusing
19088   stuff. Instead only the actually done links are displayed.
19089
19090   *Ralf S. Engelschall*
19091
19092 * Permit null encryption ciphersuites, used for authentication only. It used
19093   to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
19094   It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
19095   encryption.
19096
19097   *Ben Laurie*
19098
19099 * Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
19100   signed attributes when verifying signatures (this would break them),
19101   the detached data encoding was wrong and public keys obtained using
19102   X509_get_pubkey() weren't freed.
19103
19104   *Steve Henson*
19105
19106 * Add text documentation for the BUFFER functions. Also added a work around
19107   to a Win95 console bug. This was triggered by the password read stuff: the
19108   last character typed gets carried over to the next fread(). If you were
19109   generating a new cert request using 'req' for example then the last
19110   character of the passphrase would be CR which would then enter the first
19111   field as blank.
19112
19113   *Steve Henson*
19114
19115 * Added the new 'Includes OpenSSL Cryptography Software' button as
19116   doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
19117   button and can be used by applications based on OpenSSL to show the
19118   relationship to the OpenSSL project.
19119
19120   *Ralf S. Engelschall*
19121
19122 * Remove confusing variables in function signatures in files
19123   ssl/ssl_lib.c and ssl/ssl.h.
19124
19125   *Lennart Bong <lob@kulthea.stacken.kth.se>*
19126
19127 * Don't install bss_file.c under PREFIX/include/
19128
19129   *Lennart Bong <lob@kulthea.stacken.kth.se>*
19130
19131 * Get the Win32 compile working again. Modify mkdef.pl so it can handle
19132   functions that return function pointers and has support for NT specific
19133   stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
19134   #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
19135   unsigned to signed types: this was killing the Win32 compile.
19136
19137   *Steve Henson*
19138
19139 * Add new certificate file to stack functions,
19140   SSL_add_dir_cert_subjects_to_stack() and
19141   SSL_add_file_cert_subjects_to_stack().  These largely supplant
19142   SSL_load_client_CA_file(), and can be used to add multiple certs easily
19143   to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
19144   This means that Apache-SSL and similar packages don't have to mess around
19145   to add as many CAs as they want to the preferred list.
19146
19147   *Ben Laurie*
19148
19149 * Experiment with doxygen documentation. Currently only partially applied to
19150   ssl/ssl_lib.c.
19151   See <http://www.stack.nl/~dimitri/doxygen/index.html>, and run doxygen with
19152   openssl.doxy as the configuration file.
19153
19154   *Ben Laurie*
19155
19156 * Get rid of remaining C++-style comments which strict C compilers hate.
19157
19158   *Ralf S. Engelschall, pointed out by Carlos Amengual*
19159
19160 * Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
19161   compiled in by default: it has problems with large keys.
19162
19163   *Steve Henson*
19164
19165 * Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
19166   DH private keys and/or callback functions which directly correspond to
19167   their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
19168   is needed for applications which have to configure certificates on a
19169   per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
19170   (e.g. s_server).
19171      For the RSA certificate situation is makes no difference, but
19172   for the DSA certificate situation this fixes the "no shared cipher"
19173   problem where the OpenSSL cipher selection procedure failed because the
19174   temporary keys were not overtaken from the context and the API provided
19175   no way to reconfigure them.
19176      The new functions now let applications reconfigure the stuff and they
19177   are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
19178   SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback.  Additionally a new
19179   non-public-API function ssl_cert_instantiate() is used as a helper
19180   function and also to reduce code redundancy inside ssl_rsa.c.
19181
19182   *Ralf S. Engelschall*
19183
19184 * Move s_server -dcert and -dkey options out of the undocumented feature
19185   area because they are useful for the DSA situation and should be
19186   recognized by the users.
19187
19188   *Ralf S. Engelschall*
19189
19190 * Fix the cipher decision scheme for export ciphers: the export bits are
19191   *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
19192   SSL_EXP_MASK.  So, the original variable has to be used instead of the
19193   already masked variable.
19194
19195   *Richard Levitte <levitte@stacken.kth.se>*
19196
19197 * Fix `port` variable from `int` to `unsigned int` in crypto/bio/b_sock.c
19198
19199   *Richard Levitte <levitte@stacken.kth.se>*
19200
19201 * Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
19202   from `int` to `unsigned int` because it is a length and initialized by
19203   EVP_DigestFinal() which expects an `unsigned int *`.
19204
19205   *Richard Levitte <levitte@stacken.kth.se>*
19206
19207 * Don't hard-code path to Perl interpreter on shebang line of Configure
19208   script. Instead use the usual Shell->Perl transition trick.
19209
19210   *Ralf S. Engelschall*
19211
19212 * Make `openssl x509 -noout -modulus`' functional also for DSA certificates
19213   (in addition to RSA certificates) to match the behaviour of `openssl dsa
19214   -noout -modulus` as it's already the case for `openssl rsa -noout
19215   -modulus`.  For RSA the -modulus is the real "modulus" while for DSA
19216   currently the public key is printed (a decision which was already done by
19217   `openssl dsa -modulus` in the past) which serves a similar purpose.
19218   Additionally the NO_RSA no longer completely removes the whole -modulus
19219   option; it now only avoids using the RSA stuff. Same applies to NO_DSA
19220   now, too.
19221
19222   *Ralf S.  Engelschall*
19223
19224 * Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
19225   BIO. See the source (crypto/evp/bio_ok.c) for more info.
19226
19227   *Arne Ansper <arne@ats.cyber.ee>*
19228
19229 * Dump the old yucky req code that tried (and failed) to allow raw OIDs
19230   to be added. Now both 'req' and 'ca' can use new objects defined in the
19231   config file.
19232
19233   *Steve Henson*
19234
19235 * Add cool BIO that does syslog (or event log on NT).
19236
19237   *Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie*
19238
19239 * Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
19240   TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
19241   TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
19242   Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
19243
19244   *Ben Laurie*
19245
19246 * Add preliminary config info for new extension code.
19247
19248   *Steve Henson*
19249
19250 * Make RSA_NO_PADDING really use no padding.
19251
19252   *Ulf Moeller <ulf@fitug.de>*
19253
19254 * Generate errors when private/public key check is done.
19255
19256   *Ben Laurie*
19257
19258 * Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
19259   for some CRL extensions and new objects added.
19260
19261   *Steve Henson*
19262
19263 * Really fix the ASN1 IMPLICIT bug this time... Partial support for private
19264   key usage extension and fuller support for authority key id.
19265
19266   *Steve Henson*
19267
19268 * Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
19269   padding method for RSA, which is recommended for new applications in PKCS
19270   #1 v2.0 (RFC 2437, October 1998).
19271   OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
19272   foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
19273   against Bleichbacher's attack on RSA.
19274   *Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
19275   Ben Laurie*
19276
19277 * Updates to the new SSL compression code
19278
19279   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19280
19281 * Fix so that the version number in the master secret, when passed
19282   via RSA, checks that if TLS was proposed, but we roll back to SSLv3
19283   (because the server will not accept higher), that the version number
19284   is 0x03,0x01, not 0x03,0x00
19285
19286   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19287
19288 * Run extensive memory leak checks on SSL commands. Fixed *lots* of memory
19289   leaks in `ssl/` relating to new `X509_get_pubkey()` behaviour. Also fixes
19290   in `apps/` and an unrelated leak in `crypto/dsa/dsa_vrf.c`.
19291
19292   *Steve Henson*
19293
19294 * Support for RAW extensions where an arbitrary extension can be
19295   created by including its DER encoding. See `apps/openssl.cnf` for
19296   an example.
19297
19298   *Steve Henson*
19299
19300 * Make sure latest Perl versions don't interpret some generated C array
19301   code as Perl array code in the crypto/err/err_genc.pl script.
19302
19303   *Lars Weber <3weber@informatik.uni-hamburg.de>*
19304
19305 * Modify ms/do_ms.bat to not generate assembly language makefiles since
19306   not many people have the assembler. Various Win32 compilation fixes and
19307   update to the INSTALL.W32 file with (hopefully) more accurate Win32
19308   build instructions.
19309
19310   *Steve Henson*
19311
19312 * Modify configure script 'Configure' to automatically create crypto/date.h
19313   file under Win32 and also build pem.h from pem.org. New script
19314   util/mkfiles.pl to create the MINFO file on environments that can't do a
19315   'make files': perl util/mkfiles.pl >MINFO should work.
19316
19317   *Steve Henson*
19318
19319 * Major rework of DES function declarations, in the pursuit of correctness
19320   and purity. As a result, many evil casts evaporated, and some weirdness,
19321   too. You may find this causes warnings in your code. Zapping your evil
19322   casts will probably fix them. Mostly.
19323
19324   *Ben Laurie*
19325
19326 * Fix for a typo in asn1.h. Bug fix to object creation script
19327   obj_dat.pl. It considered a zero in an object definition to mean
19328   "end of object": none of the objects in objects.h have any zeros
19329   so it wasn't spotted.
19330
19331   *Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>*
19332
19333 * Add support for Triple DES Cipher Block Chaining with Output Feedback
19334   Masking (CBCM). In the absence of test vectors, the best I have been able
19335   to do is check that the decrypt undoes the encrypt, so far. Send me test
19336   vectors if you have them.
19337
19338   *Ben Laurie*
19339
19340 * Correct calculation of key length for export ciphers (too much space was
19341   allocated for null ciphers). This has not been tested!
19342
19343   *Ben Laurie*
19344
19345 * Modifications to the mkdef.pl for Win32 DEF file creation. The usage
19346   message is now correct (it understands "crypto" and "ssl" on its
19347   command line). There is also now an "update" option. This will update
19348   the util/ssleay.num and util/libeay.num files with any new functions.
19349   If you do a:
19350   perl util/mkdef.pl crypto ssl update
19351   it will update them.
19352
19353   *Steve Henson*
19354
19355 * Overhauled the Perl interface:
19356   - ported BN stuff to OpenSSL's different BN library
19357   - made the perl/ source tree CVS-aware
19358   - renamed the package from SSLeay to OpenSSL (the files still contain
19359     their history because I've copied them in the repository)
19360   - removed obsolete files (the test scripts will be replaced
19361     by better Test::Harness variants in the future)
19362
19363   *Ralf S. Engelschall*
19364
19365 * First cut for a very conservative source tree cleanup:
19366   1. merge various obsolete readme texts into doc/ssleay.txt
19367   where we collect the old documents and readme texts.
19368   2. remove the first part of files where I'm already sure that we no
19369   longer need them because of three reasons: either they are just temporary
19370   files which were left by Eric or they are preserved original files where
19371   I've verified that the diff is also available in the CVS via "cvs diff
19372   -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
19373   the crypto/md/ stuff).
19374
19375   *Ralf S. Engelschall*
19376
19377 * More extension code. Incomplete support for subject and issuer alt
19378   name, issuer and authority key id. Change the i2v function parameters
19379   and add an extra 'crl' parameter in the X509V3_CTX structure: guess
19380   what that's for :-) Fix to ASN1 macro which messed up
19381   IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
19382
19383   *Steve Henson*
19384
19385 * Preliminary support for ENUMERATED type. This is largely copied from the
19386   INTEGER code.
19387
19388   *Steve Henson*
19389
19390 * Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
19391
19392   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19393
19394 * Make sure `make rehash` target really finds the `openssl` program.
19395
19396   *Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
19397
19398 * Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
19399   like to hear about it if this slows down other processors.
19400
19401   *Ben Laurie*
19402
19403 * Add CygWin32 platform information to Configure script.
19404
19405   *Alan Batie <batie@aahz.jf.intel.com>*
19406
19407 * Fixed ms/32all.bat script: `no_asm` -> `no-asm`
19408
19409   *Rainer W. Gerling <gerling@mpg-gv.mpg.de>*
19410
19411 * New program nseq to manipulate netscape certificate sequences
19412
19413   *Steve Henson*
19414
19415 * Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
19416   few typos.
19417
19418   *Steve Henson*
19419
19420 * Fixes to BN code.  Previously the default was to define BN_RECURSION
19421   but the BN code had some problems that would cause failures when
19422   doing certificate verification and some other functions.
19423
19424   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19425
19426 * Add ASN1 and PEM code to support netscape certificate sequences.
19427
19428   *Steve Henson*
19429
19430 * Add ASN1 and PEM code to support netscape certificate sequences.
19431
19432   *Steve Henson*
19433
19434 * Add several PKIX and private extended key usage OIDs.
19435
19436   *Steve Henson*
19437
19438 * Modify the 'ca' program to handle the new extension code. Modify
19439   openssl.cnf for new extension format, add comments.
19440
19441   *Steve Henson*
19442
19443 * More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
19444   and add a sample to openssl.cnf so req -x509 now adds appropriate
19445   CA extensions.
19446
19447   *Steve Henson*
19448
19449 * Continued X509 V3 changes. Add to other makefiles, integrate with the
19450   error code, add initial support to X509_print() and x509 application.
19451
19452   *Steve Henson*
19453
19454 * Takes a deep breath and start adding X509 V3 extension support code. Add
19455   files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
19456   stuff is currently isolated and isn't even compiled yet.
19457
19458   *Steve Henson*
19459
19460 * Continuing patches for GeneralizedTime. Fix up certificate and CRL
19461   ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
19462   Removed the versions check from X509 routines when loading extensions:
19463   this allows certain broken certificates that don't set the version
19464   properly to be processed.
19465
19466   *Steve Henson*
19467
19468 * Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
19469   Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
19470   can still be regenerated with "make depend".
19471
19472   *Ben Laurie*
19473
19474 * Spelling mistake in C version of CAST-128.
19475
19476   *Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>*
19477
19478 * Changes to the error generation code. The perl script err-code.pl
19479   now reads in the old error codes and retains the old numbers, only
19480   adding new ones if necessary. It also only changes the .err files if new
19481   codes are added. The makefiles have been modified to only insert errors
19482   when needed (to avoid needlessly modifying header files). This is done
19483   by only inserting errors if the .err file is newer than the auto generated
19484   C file. To rebuild all the error codes from scratch (the old behaviour)
19485   either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
19486   or delete all the .err files.
19487
19488   *Steve Henson*
19489
19490 * CAST-128 was incorrectly implemented for short keys. The C version has
19491   been fixed, but is untested. The assembler versions are also fixed, but
19492   new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
19493   to regenerate it if needed.
19494   *Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
19495    Hagino <itojun@kame.net>*
19496
19497 * File was opened incorrectly in randfile.c.
19498
19499   *Ulf Möller <ulf@fitug.de>*
19500
19501 * Beginning of support for GeneralizedTime. d2i, i2d, check and print
19502   functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
19503   GeneralizedTime. ASN1_TIME is the proper type used in certificates et
19504   al: it's just almost always a UTCTime. Note this patch adds new error
19505   codes so do a "make errors" if there are problems.
19506
19507   *Steve Henson*
19508
19509 * Correct Linux 1 recognition in config.
19510
19511   *Ulf Möller <ulf@fitug.de>*
19512
19513 * Remove pointless MD5 hash when using DSA keys in ca.
19514
19515   *Anonymous <nobody@replay.com>*
19516
19517 * Generate an error if given an empty string as a cert directory. Also
19518   generate an error if handed NULL (previously returned 0 to indicate an
19519   error, but didn't set one).
19520
19521   *Ben Laurie, reported by Anonymous <nobody@replay.com>*
19522
19523 * Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
19524
19525   *Ben Laurie*
19526
19527 * Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
19528   parameters. This was causing a warning which killed off the Win32 compile.
19529
19530   *Steve Henson*
19531
19532 * Remove C++ style comments from crypto/bn/bn_local.h.
19533
19534   *Neil Costigan <neil.costigan@celocom.com>*
19535
19536 * The function OBJ_txt2nid was broken. It was supposed to return a nid
19537   based on a text string, looking up short and long names and finally
19538   "dot" format. The "dot" format stuff didn't work. Added new function
19539   OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote
19540   OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
19541   OID is not part of the table.
19542
19543   *Steve Henson*
19544
19545 * Add prototypes to X509 lookup/verify methods, fixing a bug in
19546   X509_LOOKUP_by_alias().
19547
19548   *Ben Laurie*
19549
19550 * Sort openssl functions by name.
19551
19552   *Ben Laurie*
19553
19554 * Get the `gendsa` command working and add it to the `list` command. Remove
19555   encryption from sample DSA keys (in case anyone is interested the password
19556   was "1234").
19557
19558   *Steve Henson*
19559
19560 * Make *all* `*_free` functions accept a NULL pointer.
19561
19562   *Frans Heymans <fheymans@isaserver.be>*
19563
19564 * If a DH key is generated in s3_srvr.c, don't blow it by trying to use
19565   NULL pointers.
19566
19567   *Anonymous <nobody@replay.com>*
19568
19569 * s_server should send the CAfile as acceptable CAs, not its own cert.
19570
19571   *Bodo Moeller <3moeller@informatik.uni-hamburg.de>*
19572
19573 * Don't blow it for numeric `-newkey` arguments to `apps/req`.
19574
19575   *Bodo Moeller <3moeller@informatik.uni-hamburg.de>*
19576
19577 * Temp key "for export" tests were wrong in s3_srvr.c.
19578
19579   *Anonymous <nobody@replay.com>*
19580
19581 * Add prototype for temp key callback functions
19582   SSL_CTX_set_tmp_{rsa,dh}_callback().
19583
19584   *Ben Laurie*
19585
19586 * Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
19587   DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
19588
19589   *Steve Henson*
19590
19591 * X509_name_add_entry() freed the wrong thing after an error.
19592
19593   *Arne Ansper <arne@ats.cyber.ee>*
19594
19595 * rsa_eay.c would attempt to free a NULL context.
19596
19597   *Arne Ansper <arne@ats.cyber.ee>*
19598
19599 * BIO_s_socket() had a broken should_retry() on Windoze.
19600
19601   *Arne Ansper <arne@ats.cyber.ee>*
19602
19603 * BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
19604
19605   *Arne Ansper <arne@ats.cyber.ee>*
19606
19607 * Make sure the already existing X509_STORE->depth variable is initialized
19608   in X509_STORE_new(), but document the fact that this variable is still
19609   unused in the certificate verification process.
19610
19611   *Ralf S. Engelschall*
19612
19613 * Fix the various library and `apps/` files to free up pkeys obtained from
19614   X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
19615
19616   *Steve Henson*
19617
19618 * Fix reference counting in X509_PUBKEY_get(). This makes
19619   demos/maurice/example2.c work, amongst others, probably.
19620
19621   *Steve Henson and Ben Laurie*
19622
19623 * First cut of a cleanup for `apps/`. First the `ssleay` program is now named
19624   `openssl` and second, the shortcut symlinks for the `openssl <command>`
19625   are no longer created. This way we have a single and consistent command
19626   line interface `openssl <command>`, similar to `cvs <command>`.
19627
19628   *Ralf S. Engelschall, Paul Sutton and Ben Laurie*
19629
19630 * ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
19631   BIT STRING wrapper always have zero unused bits.
19632
19633   *Steve Henson*
19634
19635 * Add CA.pl, perl version of CA.sh, add extended key usage OID.
19636
19637   *Steve Henson*
19638
19639 * Make the top-level INSTALL documentation easier to understand.
19640
19641   *Paul Sutton*
19642
19643 * Makefiles updated to exit if an error occurs in a sub-directory
19644   make (including if user presses ^C) [Paul Sutton]
19645
19646 * Make Montgomery context stuff explicit in RSA data structure.
19647
19648   *Ben Laurie*
19649
19650 * Fix build order of pem and err to allow for generated pem.h.
19651
19652   *Ben Laurie*
19653
19654 * Fix renumbering bug in X509_NAME_delete_entry().
19655
19656   *Ben Laurie*
19657
19658 * Enhanced the err-ins.pl script so it makes the error library number
19659   global and can add a library name. This is needed for external ASN1 and
19660   other error libraries.
19661
19662   *Steve Henson*
19663
19664 * Fixed sk_insert which never worked properly.
19665
19666   *Steve Henson*
19667
19668 * Fix ASN1 macros so they can handle indefinite length constructed
19669   EXPLICIT tags. Some non standard certificates use these: they can now
19670   be read in.
19671
19672   *Steve Henson*
19673
19674 * Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
19675   into a single doc/ssleay.txt bundle. This way the information is still
19676   preserved but no longer messes up this directory. Now it's new room for
19677   the new set of documentation files.
19678
19679   *Ralf S. Engelschall*
19680
19681 * SETs were incorrectly DER encoded. This was a major pain, because they
19682   shared code with SEQUENCEs, which aren't coded the same. This means that
19683   almost everything to do with SETs or SEQUENCEs has either changed name or
19684   number of arguments.
19685
19686   *Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>*
19687
19688 * Fix test data to work with the above.
19689
19690   *Ben Laurie*
19691
19692 * Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
19693   was already fixed by Eric for 0.9.1 it seems.
19694
19695   *Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>*
19696
19697 * Autodetect FreeBSD3.
19698
19699   *Ben Laurie*
19700
19701 * Fix various bugs in Configure. This affects the following platforms:
19702   nextstep
19703   ncr-scde
19704   unixware-2.0
19705   unixware-2.0-pentium
19706   sco5-cc.
19707
19708   *Ben Laurie*
19709
19710 * Eliminate generated files from CVS. Reorder tests to regenerate files
19711   before they are needed.
19712
19713   *Ben Laurie*
19714
19715 * Generate Makefile.ssl from Makefile.org (to keep CVS happy).
19716
19717   *Ben Laurie*
19718
19719### Changes between 0.9.1b and 0.9.1c  [23-Dec-1998]
19720
19721 * Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and
19722   changed SSLeay to OpenSSL in version strings.
19723
19724   *Ralf S. Engelschall*
19725
19726 * Some fixups to the top-level documents.
19727
19728   *Paul Sutton*
19729
19730 * Fixed the nasty bug where rsaref.h was not found under compile-time
19731   because the symlink to include/ was missing.
19732
19733   *Ralf S. Engelschall*
19734
19735 * Incorporated the popular no-RSA/DSA-only patches
19736   which allow to compile an RSA-free SSLeay.
19737
19738   *Andrew Cooke / Interrader Ldt., Ralf S. Engelschall*
19739
19740 * Fixed nasty rehash problem under `make -f Makefile.ssl links`
19741   when "ssleay" is still not found.
19742
19743   *Ralf S. Engelschall*
19744
19745 * Added more platforms to Configure: Cray T3E, HPUX 11,
19746
19747   *Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>*
19748
19749 * Updated the README file.
19750
19751   *Ralf S. Engelschall*
19752
19753 * Added various .cvsignore files in the CVS repository subdirs
19754   to make a "cvs update" really silent.
19755
19756   *Ralf S. Engelschall*
19757
19758 * Recompiled the error-definition header files and added
19759   missing symbols to the Win32 linker tables.
19760
19761   *Ralf S. Engelschall*
19762
19763 * Cleaned up the top-level documents;
19764   o new files: CHANGES and LICENSE
19765   o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay
19766   o merged COPYRIGHT into LICENSE
19767   o removed obsolete TODO file
19768   o renamed MICROSOFT to INSTALL.W32
19769
19770   *Ralf S. Engelschall*
19771
19772 * Removed dummy files from the 0.9.1b source tree:
19773   crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
19774   crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
19775   crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
19776   crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
19777   util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
19778
19779   *Ralf S. Engelschall*
19780
19781 * Added various platform portability fixes.
19782
19783   *Mark J. Cox*
19784
19785 * The Genesis of the OpenSSL rpject:
19786   We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
19787   Young and Tim J. Hudson created while they were working for C2Net until
19788   summer 1998.
19789
19790   *The OpenSSL Project*
19791
19792### Changes between 0.9.0b and 0.9.1b  [not released]
19793
19794 * Updated a few CA certificates under certs/
19795
19796   *Eric A. Young*
19797
19798 * Changed some BIGNUM api stuff.
19799
19800   *Eric A. Young*
19801
19802 * Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD,
19803   DGUX x86, Linux Alpha, etc.
19804
19805   *Eric A. Young*
19806
19807 * New COMP library [crypto/comp/] for SSL Record Layer Compression:
19808   RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
19809   available).
19810
19811   *Eric A. Young*
19812
19813 * Add -strparse option to asn1pars program which parses nested
19814   binary structures
19815
19816   *Dr Stephen Henson <shenson@bigfoot.com>*
19817
19818 * Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
19819
19820   *Eric A. Young*
19821
19822 * DSA fix for "ca" program.
19823
19824   *Eric A. Young*
19825
19826 * Added "-genkey" option to "dsaparam" program.
19827
19828   *Eric A. Young*
19829
19830 * Added RIPE MD160 (rmd160) message digest.
19831
19832   *Eric A. Young*
19833
19834 * Added -a (all) option to "ssleay version" command.
19835
19836   *Eric A. Young*
19837
19838 * Added PLATFORM define which is the id given to Configure.
19839
19840   *Eric A. Young*
19841
19842 * Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
19843
19844   *Eric A. Young*
19845
19846 * Extended the ASN.1 parser routines.
19847
19848   *Eric A. Young*
19849
19850 * Extended BIO routines to support REUSEADDR, seek, tell, etc.
19851
19852   *Eric A. Young*
19853
19854 * Added a BN_CTX to the BN library.
19855
19856   *Eric A. Young*
19857
19858 * Fixed the weak key values in DES library
19859
19860   *Eric A. Young*
19861
19862 * Changed API in EVP library for cipher aliases.
19863
19864   *Eric A. Young*
19865
19866 * Added support for RC2/64bit cipher.
19867
19868   *Eric A. Young*
19869
19870 * Converted the lhash library to the crypto/mem.c functions.
19871
19872   *Eric A. Young*
19873
19874 * Added more recognized ASN.1 object ids.
19875
19876   *Eric A. Young*
19877
19878 * Added more RSA padding checks for SSL/TLS.
19879
19880   *Eric A. Young*
19881
19882 * Added BIO proxy/filter functionality.
19883
19884   *Eric A. Young*
19885
19886 * Added extra_certs to SSL_CTX which can be used
19887   send extra CA certificates to the client in the CA cert chain sending
19888   process. It can be configured with SSL_CTX_add_extra_chain_cert().
19889
19890   *Eric A. Young*
19891
19892 * Now Fortezza is denied in the authentication phase because
19893   this is key exchange mechanism is not supported by SSLeay at all.
19894
19895   *Eric A. Young*
19896
19897 * Additional PKCS1 checks.
19898
19899   *Eric A. Young*
19900
19901 * Support the string "TLSv1" for all TLS v1 ciphers.
19902
19903   *Eric A. Young*
19904
19905 * Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
19906   ex_data index of the SSL context in the X509_STORE_CTX ex_data.
19907
19908   *Eric A. Young*
19909
19910 * Fixed a few memory leaks.
19911
19912   *Eric A. Young*
19913
19914 * Fixed various code and comment typos.
19915
19916   *Eric A. Young*
19917
19918 * A minor bug in ssl/s3_clnt.c where there would always be 4 0
19919   bytes sent in the client random.
19920
19921   *Edward Bishop <ebishop@spyglass.com>*
19922
19923<!-- Links -->
19924
19925[CVE-2024-6119]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-6119
19926[CVE-2024-5535]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-5535
19927[CVE-2024-4741]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-4741
19928[CVE-2024-4603]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-4603
19929[CVE-2024-2511]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-2511
19930[CVE-2024-0727]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-0727
19931[CVE-2023-6237]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-6237
19932[CVE-2023-6129]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-6129
19933[CVE-2023-5678]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-5678
19934[CVE-2023-5363]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-5363
19935[CVE-2023-4807]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-4807
19936[CVE-2023-3817]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-3817
19937[CVE-2023-3446]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-3446
19938[CVE-2023-2975]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-2975
19939[RFC 2578 (STD 58), section 3.5]: https://datatracker.ietf.org/doc/html/rfc2578#section-3.5
19940[CVE-2023-2650]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-2650
19941[CVE-2023-1255]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-1255
19942[CVE-2023-0466]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0466
19943[CVE-2023-0465]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0465
19944[CVE-2023-0464]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0464
19945[CVE-2023-0401]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0401
19946[CVE-2023-0286]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0286
19947[CVE-2023-0217]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0217
19948[CVE-2023-0216]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0216
19949[CVE-2023-0215]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0215
19950[CVE-2022-4450]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4450
19951[CVE-2022-4304]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4304
19952[CVE-2022-4203]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4203
19953[CVE-2022-3996]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-3996
19954[CVE-2022-2274]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-2274
19955[CVE-2022-2097]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-2097
19956[CVE-2020-1971]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1971
19957[CVE-2020-1967]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1967
19958[CVE-2019-1563]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1563
19959[CVE-2019-1559]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1559
19960[CVE-2019-1552]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1552
19961[CVE-2019-1551]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1551
19962[CVE-2019-1549]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1549
19963[CVE-2019-1547]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1547
19964[CVE-2019-1543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1543
19965[CVE-2018-5407]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-5407
19966[CVE-2018-0739]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0739
19967[CVE-2018-0737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0737
19968[CVE-2018-0735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0735
19969[CVE-2018-0734]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0734
19970[CVE-2018-0733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0733
19971[CVE-2018-0732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0732
19972[CVE-2017-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3738
19973[CVE-2017-3737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3737
19974[CVE-2017-3736]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3736
19975[CVE-2017-3735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3735
19976[CVE-2017-3733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3733
19977[CVE-2017-3732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3732
19978[CVE-2017-3731]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3731
19979[CVE-2017-3730]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3730
19980[CVE-2016-7055]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7055
19981[CVE-2016-7054]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7054
19982[CVE-2016-7053]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7053
19983[CVE-2016-7052]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7052
19984[CVE-2016-6309]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6309
19985[CVE-2016-6308]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6308
19986[CVE-2016-6307]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6307
19987[CVE-2016-6306]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6306
19988[CVE-2016-6305]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6305
19989[CVE-2016-6304]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6304
19990[CVE-2016-6303]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6303
19991[CVE-2016-6302]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6302
19992[CVE-2016-2183]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2183
19993[CVE-2016-2182]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2182
19994[CVE-2016-2181]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2181
19995[CVE-2016-2180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2180
19996[CVE-2016-2179]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2179
19997[CVE-2016-2178]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2178
19998[CVE-2016-2177]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2177
19999[CVE-2016-2176]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2176
20000[CVE-2016-2109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2109
20001[CVE-2016-2107]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2107
20002[CVE-2016-2106]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2106
20003[CVE-2016-2105]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2105
20004[CVE-2016-0800]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0800
20005[CVE-2016-0799]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0799
20006[CVE-2016-0798]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0798
20007[CVE-2016-0797]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0797
20008[CVE-2016-0705]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0705
20009[CVE-2016-0702]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0702
20010[CVE-2016-0701]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0701
20011[CVE-2015-3197]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3197
20012[CVE-2015-3196]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3196
20013[CVE-2015-3195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3195
20014[CVE-2015-3194]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3194
20015[CVE-2015-3193]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3193
20016[CVE-2015-1793]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1793
20017[CVE-2015-1792]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1792
20018[CVE-2015-1791]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1791
20019[CVE-2015-1790]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1790
20020[CVE-2015-1789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1789
20021[CVE-2015-1788]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1788
20022[CVE-2015-1787]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1787
20023[CVE-2015-0293]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0293
20024[CVE-2015-0291]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0291
20025[CVE-2015-0290]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0290
20026[CVE-2015-0289]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0289
20027[CVE-2015-0288]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0288
20028[CVE-2015-0287]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0287
20029[CVE-2015-0286]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0286
20030[CVE-2015-0285]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0285
20031[CVE-2015-0209]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0209
20032[CVE-2015-0208]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0208
20033[CVE-2015-0207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0207
20034[CVE-2015-0206]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0206
20035[CVE-2015-0205]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0205
20036[CVE-2015-0204]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0204
20037[CVE-2014-8275]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-8275
20038[CVE-2014-5139]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-5139
20039[CVE-2014-3572]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3572
20040[CVE-2014-3571]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3571
20041[CVE-2014-3570]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3570
20042[CVE-2014-3569]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3569
20043[CVE-2014-3568]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3568
20044[CVE-2014-3567]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3567
20045[CVE-2014-3566]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3566
20046[CVE-2014-3513]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3513
20047[CVE-2014-3512]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3512
20048[CVE-2014-3511]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3511
20049[CVE-2014-3510]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3510
20050[CVE-2014-3509]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3509
20051[CVE-2014-3508]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3508
20052[CVE-2014-3507]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3507
20053[CVE-2014-3506]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3506
20054[CVE-2014-3505]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3505
20055[CVE-2014-3470]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3470
20056[CVE-2014-0224]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0224
20057[CVE-2014-0221]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0221
20058[CVE-2014-0195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0195
20059[CVE-2014-0160]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0160
20060[CVE-2014-0076]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0076
20061[CVE-2013-6450]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-6450
20062[CVE-2013-4353]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-4353
20063[CVE-2013-0169]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0169
20064[CVE-2013-0166]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0166
20065[CVE-2012-2686]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2686
20066[CVE-2012-2333]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2333
20067[CVE-2012-2110]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2110
20068[CVE-2012-0884]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0884
20069[CVE-2012-0050]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0050
20070[CVE-2012-0027]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0027
20071[CVE-2011-4619]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4619
20072[CVE-2011-4577]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4577
20073[CVE-2011-4576]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4576
20074[CVE-2011-4109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4109
20075[CVE-2011-4108]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4108
20076[CVE-2011-3210]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3210
20077[CVE-2011-3207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3207
20078[CVE-2011-0014]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-0014
20079[CVE-2010-4252]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4252
20080[CVE-2010-4180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4180
20081[CVE-2010-3864]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-3864
20082[CVE-2010-1633]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-1633
20083[CVE-2010-0740]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0740
20084[CVE-2010-0433]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0433
20085[CVE-2009-4355]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-4355
20086[CVE-2009-3555]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3555
20087[CVE-2009-3245]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3245
20088[CVE-2009-1386]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1386
20089[CVE-2009-1379]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1379
20090[CVE-2009-1378]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1378
20091[CVE-2009-1377]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1377
20092[CVE-2009-0789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0789
20093[CVE-2009-0591]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0591
20094[CVE-2009-0590]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0590
20095[CVE-2008-5077]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-5077
20096[CVE-2008-1678]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1678
20097[CVE-2008-1672]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1672
20098[CVE-2008-0891]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-0891
20099[CVE-2007-5135]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-5135
20100[CVE-2007-4995]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-4995
20101[CVE-2006-4343]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4343
20102[CVE-2006-4339]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4339
20103[CVE-2006-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-3738
20104[CVE-2006-2940]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2940
20105[CVE-2006-2937]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2937
20106[CVE-2005-2969]: https://www.openssl.org/news/vulnerabilities.html#CVE-2005-2969
20107[CVE-2004-0112]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0112
20108[CVE-2004-0079]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0079
20109[CVE-2003-0851]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0851
20110[CVE-2003-0545]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0545
20111[CVE-2003-0544]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0544
20112[CVE-2003-0543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0543
20113[CVE-2003-0078]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0078
20114[CVE-2002-0659]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0659
20115[CVE-2002-0657]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0657
20116[CVE-2002-0656]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0656
20117[CVE-2002-0655]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0655
20118