CHANGES revision 284285
1
2 OpenSSL CHANGES
3 _______________
4
5 Changes between 1.0.1m and 1.0.1n [11 Jun 2015]
6
7  *) Malformed ECParameters causes infinite loop
8
9     When processing an ECParameters structure OpenSSL enters an infinite loop
10     if the curve specified is over a specially malformed binary polynomial
11     field.
12
13     This can be used to perform denial of service against any
14     system which processes public keys, certificate requests or
15     certificates.  This includes TLS clients and TLS servers with
16     client authentication enabled.
17
18     This issue was reported to OpenSSL by Joseph Barr-Pixton.
19     (CVE-2015-1788)
20     [Andy Polyakov]
21
22  *) Exploitable out-of-bounds read in X509_cmp_time
23
24     X509_cmp_time does not properly check the length of the ASN1_TIME
25     string and can read a few bytes out of bounds. In addition,
26     X509_cmp_time accepts an arbitrary number of fractional seconds in the
27     time string.
28
29     An attacker can use this to craft malformed certificates and CRLs of
30     various sizes and potentially cause a segmentation fault, resulting in
31     a DoS on applications that verify certificates or CRLs. TLS clients
32     that verify CRLs are affected. TLS clients and servers with client
33     authentication enabled may be affected if they use custom verification
34     callbacks.
35
36     This issue was reported to OpenSSL by Robert Swiecki (Google), and
37     independently by Hanno B�ck.
38     (CVE-2015-1789)
39     [Emilia K�sper]
40
41  *) PKCS7 crash with missing EnvelopedContent
42
43     The PKCS#7 parsing code does not handle missing inner EncryptedContent
44     correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
45     with missing content and trigger a NULL pointer dereference on parsing.
46
47     Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
48     structures from untrusted sources are affected. OpenSSL clients and
49     servers are not affected.
50
51     This issue was reported to OpenSSL by Michal Zalewski (Google).
52     (CVE-2015-1790)
53     [Emilia K�sper]
54
55  *) CMS verify infinite loop with unknown hash function
56
57     When verifying a signedData message the CMS code can enter an infinite loop
58     if presented with an unknown hash function OID. This can be used to perform
59     denial of service against any system which verifies signedData messages using
60     the CMS code.
61     This issue was reported to OpenSSL by Johannes Bauer.
62     (CVE-2015-1792)
63     [Stephen Henson]
64
65  *) Race condition handling NewSessionTicket
66
67     If a NewSessionTicket is received by a multi-threaded client when attempting to
68     reuse a previous ticket then a race condition can occur potentially leading to
69     a double free of the ticket data.
70     (CVE-2015-1791)
71     [Matt Caswell]
72
73  *) Reject DH handshakes with parameters shorter than 768 bits.
74     [Kurt Roeckx and Emilia Kasper]
75
76 Changes between 1.0.1l and 1.0.1m [19 Mar 2015]
77
78  *) Segmentation fault in ASN1_TYPE_cmp fix
79
80     The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
81     made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
82     certificate signature algorithm consistency this can be used to crash any
83     certificate verification operation and exploited in a DoS attack. Any
84     application which performs certificate verification is vulnerable including
85     OpenSSL clients and servers which enable client authentication.
86     (CVE-2015-0286)
87     [Stephen Henson]
88
89  *) ASN.1 structure reuse memory corruption fix
90
91     Reusing a structure in ASN.1 parsing may allow an attacker to cause
92     memory corruption via an invalid write. Such reuse is and has been
93     strongly discouraged and is believed to be rare.
94
95     Applications that parse structures containing CHOICE or ANY DEFINED BY
96     components may be affected. Certificate parsing (d2i_X509 and related
97     functions) are however not affected. OpenSSL clients and servers are
98     not affected.
99     (CVE-2015-0287)
100     [Stephen Henson]
101
102  *) PKCS7 NULL pointer dereferences fix
103
104     The PKCS#7 parsing code does not handle missing outer ContentInfo
105     correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
106     missing content and trigger a NULL pointer dereference on parsing.
107
108     Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
109     otherwise parse PKCS#7 structures from untrusted sources are
110     affected. OpenSSL clients and servers are not affected.
111
112     This issue was reported to OpenSSL by Michal Zalewski (Google).
113     (CVE-2015-0289)
114     [Emilia K�sper]
115
116  *) DoS via reachable assert in SSLv2 servers fix
117
118     A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
119     servers that both support SSLv2 and enable export cipher suites by sending
120     a specially crafted SSLv2 CLIENT-MASTER-KEY message.
121
122     This issue was discovered by Sean Burford (Google) and Emilia K�sper
123     (OpenSSL development team).
124     (CVE-2015-0293)
125     [Emilia K�sper]
126
127  *) Use After Free following d2i_ECPrivatekey error fix
128
129     A malformed EC private key file consumed via the d2i_ECPrivateKey function
130     could cause a use after free condition. This, in turn, could cause a double
131     free in several private key parsing functions (such as d2i_PrivateKey
132     or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
133     for applications that receive EC private keys from untrusted
134     sources. This scenario is considered rare.
135
136     This issue was discovered by the BoringSSL project and fixed in their
137     commit 517073cd4b.
138     (CVE-2015-0209)
139     [Matt Caswell]
140
141  *) X509_to_X509_REQ NULL pointer deref fix
142
143     The function X509_to_X509_REQ will crash with a NULL pointer dereference if
144     the certificate key is invalid. This function is rarely used in practice.
145
146     This issue was discovered by Brian Carpenter.
147     (CVE-2015-0288)
148     [Stephen Henson]
149
150  *) Removed the export ciphers from the DEFAULT ciphers
151     [Kurt Roeckx]
152
153 Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
154
155  *) Build fixes for the Windows and OpenVMS platforms
156     [Matt Caswell and Richard Levitte]
157
158 Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
159
160  *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
161     message can cause a segmentation fault in OpenSSL due to a NULL pointer
162     dereference. This could lead to a Denial Of Service attack. Thanks to
163     Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
164     (CVE-2014-3571)
165     [Steve Henson]
166
167  *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
168     dtls1_buffer_record function under certain conditions. In particular this
169     could occur if an attacker sent repeated DTLS records with the same
170     sequence number but for the next epoch. The memory leak could be exploited
171     by an attacker in a Denial of Service attack through memory exhaustion.
172     Thanks to Chris Mueller for reporting this issue.
173     (CVE-2015-0206)
174     [Matt Caswell]
175
176  *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
177     built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
178     method would be set to NULL which could later result in a NULL pointer
179     dereference. Thanks to Frank Schmirler for reporting this issue.
180     (CVE-2014-3569)
181     [Kurt Roeckx]
182
183  *) Abort handshake if server key exchange message is omitted for ephemeral
184     ECDH ciphersuites.
185
186     Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
187     reporting this issue.
188     (CVE-2014-3572)
189     [Steve Henson]
190
191  *) Remove non-export ephemeral RSA code on client and server. This code
192     violated the TLS standard by allowing the use of temporary RSA keys in
193     non-export ciphersuites and could be used by a server to effectively
194     downgrade the RSA key length used to a value smaller than the server
195     certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
196     INRIA or reporting this issue.
197     (CVE-2015-0204)
198     [Steve Henson]
199
200  *) Fixed issue where DH client certificates are accepted without verification.
201     An OpenSSL server will accept a DH certificate for client authentication
202     without the certificate verify message. This effectively allows a client to
203     authenticate without the use of a private key. This only affects servers
204     which trust a client certificate authority which issues certificates
205     containing DH keys: these are extremely rare and hardly ever encountered.
206     Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
207     this issue.
208     (CVE-2015-0205)
209     [Steve Henson]
210
211  *) Ensure that the session ID context of an SSL is updated when its
212     SSL_CTX is updated via SSL_set_SSL_CTX.
213
214     The session ID context is typically set from the parent SSL_CTX,
215     and can vary with the CTX.
216     [Adam Langley]
217
218  *) Fix various certificate fingerprint issues.
219
220     By using non-DER or invalid encodings outside the signed portion of a
221     certificate the fingerprint can be changed without breaking the signature.
222     Although no details of the signed portion of the certificate can be changed
223     this can cause problems with some applications: e.g. those using the
224     certificate fingerprint for blacklists.
225
226     1. Reject signatures with non zero unused bits.
227
228     If the BIT STRING containing the signature has non zero unused bits reject
229     the signature. All current signature algorithms require zero unused bits.
230
231     2. Check certificate algorithm consistency.
232
233     Check the AlgorithmIdentifier inside TBS matches the one in the
234     certificate signature. NB: this will result in signature failure
235     errors for some broken certificates.
236
237     Thanks to Konrad Kraszewski from Google for reporting this issue.
238
239     3. Check DSA/ECDSA signatures use DER.
240
241     Reencode DSA/ECDSA signatures and compare with the original received
242     signature. Return an error if there is a mismatch.
243
244     This will reject various cases including garbage after signature
245     (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
246     program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
247     (negative or with leading zeroes).
248
249     Further analysis was conducted and fixes were developed by Stephen Henson
250     of the OpenSSL core team.
251
252     (CVE-2014-8275)
253     [Steve Henson]
254
255   *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
256      results on some platforms, including x86_64. This bug occurs at random
257      with a very low probability, and is not known to be exploitable in any
258      way, though its exact impact is difficult to determine. Thanks to Pieter
259      Wuille (Blockstream) who reported this issue and also suggested an initial
260      fix. Further analysis was conducted by the OpenSSL development team and
261      Adam Langley of Google. The final fix was developed by Andy Polyakov of
262      the OpenSSL core team.
263      (CVE-2014-3570)
264      [Andy Polyakov]
265
266   *) Do not resume sessions on the server if the negotiated protocol
267      version does not match the session's version. Resuming with a different
268      version, while not strictly forbidden by the RFC, is of questionable
269      sanity and breaks all known clients.
270      [David Benjamin, Emilia K�sper]
271
272   *) Tighten handling of the ChangeCipherSpec (CCS) message: reject
273      early CCS messages during renegotiation. (Note that because
274      renegotiation is encrypted, this early CCS was not exploitable.)
275      [Emilia K�sper]
276
277   *) Tighten client-side session ticket handling during renegotiation:
278      ensure that the client only accepts a session ticket if the server sends
279      the extension anew in the ServerHello. Previously, a TLS client would
280      reuse the old extension state and thus accept a session ticket if one was
281      announced in the initial ServerHello.
282
283      Similarly, ensure that the client requires a session ticket if one
284      was advertised in the ServerHello. Previously, a TLS client would
285      ignore a missing NewSessionTicket message.
286      [Emilia K�sper]
287
288 Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
289
290  *) SRTP Memory Leak.
291
292     A flaw in the DTLS SRTP extension parsing code allows an attacker, who
293     sends a carefully crafted handshake message, to cause OpenSSL to fail
294     to free up to 64k of memory causing a memory leak. This could be
295     exploited in a Denial Of Service attack. This issue affects OpenSSL
296     1.0.1 server implementations for both SSL/TLS and DTLS regardless of
297     whether SRTP is used or configured. Implementations of OpenSSL that
298     have been compiled with OPENSSL_NO_SRTP defined are not affected.
299
300     The fix was developed by the OpenSSL team.
301     (CVE-2014-3513)
302     [OpenSSL team]
303
304  *) Session Ticket Memory Leak.
305
306     When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
307     integrity of that ticket is first verified. In the event of a session
308     ticket integrity check failing, OpenSSL will fail to free memory
309     causing a memory leak. By sending a large number of invalid session
310     tickets an attacker could exploit this issue in a Denial Of Service
311     attack.
312     (CVE-2014-3567)
313     [Steve Henson]
314
315  *) Build option no-ssl3 is incomplete.
316
317     When OpenSSL is configured with "no-ssl3" as a build option, servers
318     could accept and complete a SSL 3.0 handshake, and clients could be
319     configured to send them.
320     (CVE-2014-3568)
321     [Akamai and the OpenSSL team]
322
323  *) Add support for TLS_FALLBACK_SCSV.
324     Client applications doing fallback retries should call
325     SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
326     (CVE-2014-3566)
327     [Adam Langley, Bodo Moeller]
328
329  *) Add additional DigestInfo checks.
330 
331     Reencode DigestInto in DER and check against the original when
332     verifying RSA signature: this will reject any improperly encoded
333     DigestInfo structures.
334
335     Note: this is a precautionary measure and no attacks are currently known.
336
337     [Steve Henson]
338
339 Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
340
341  *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
342     SRP code can be overrun an internal buffer. Add sanity check that
343     g, A, B < N to SRP code.
344
345     Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
346     Group for discovering this issue.
347     (CVE-2014-3512)
348     [Steve Henson]
349
350  *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
351     TLS 1.0 instead of higher protocol versions when the ClientHello message
352     is badly fragmented. This allows a man-in-the-middle attacker to force a
353     downgrade to TLS 1.0 even if both the server and the client support a
354     higher protocol version, by modifying the client's TLS records.
355
356     Thanks to David Benjamin and Adam Langley (Google) for discovering and
357     researching this issue.
358     (CVE-2014-3511)
359     [David Benjamin]
360
361  *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
362     to a denial of service attack. A malicious server can crash the client
363     with a null pointer dereference (read) by specifying an anonymous (EC)DH
364     ciphersuite and sending carefully crafted handshake messages.
365
366     Thanks to Felix Gr�bert (Google) for discovering and researching this
367     issue.
368     (CVE-2014-3510)
369     [Emilia K�sper]
370
371  *) By sending carefully crafted DTLS packets an attacker could cause openssl
372     to leak memory. This can be exploited through a Denial of Service attack.
373     Thanks to Adam Langley for discovering and researching this issue.
374     (CVE-2014-3507)
375     [Adam Langley]
376
377  *) An attacker can force openssl to consume large amounts of memory whilst
378     processing DTLS handshake messages. This can be exploited through a
379     Denial of Service attack.
380     Thanks to Adam Langley for discovering and researching this issue.
381     (CVE-2014-3506)
382     [Adam Langley]
383
384  *) An attacker can force an error condition which causes openssl to crash
385     whilst processing DTLS packets due to memory being freed twice. This
386     can be exploited through a Denial of Service attack.
387     Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
388     this issue.
389     (CVE-2014-3505)
390     [Adam Langley]
391
392  *) If a multithreaded client connects to a malicious server using a resumed
393     session and the server sends an ec point format extension it could write
394     up to 255 bytes to freed memory.
395
396     Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
397     issue.
398     (CVE-2014-3509)
399     [Gabor Tyukasz]
400
401  *) A malicious server can crash an OpenSSL client with a null pointer
402     dereference (read) by specifying an SRP ciphersuite even though it was not
403     properly negotiated with the client. This can be exploited through a
404     Denial of Service attack.
405
406     Thanks to Joonas Kuorilehto and Riku Hietam�ki (Codenomicon) for
407     discovering and researching this issue.
408     (CVE-2014-5139)
409     [Steve Henson]
410
411  *) A flaw in OBJ_obj2txt may cause pretty printing functions such as
412     X509_name_oneline, X509_name_print_ex et al. to leak some information
413     from the stack. Applications may be affected if they echo pretty printing
414     output to the attacker.
415
416     Thanks to Ivan Fratric (Google) for discovering this issue.
417     (CVE-2014-3508)
418     [Emilia K�sper, and Steve Henson]
419
420  *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
421     for corner cases. (Certain input points at infinity could lead to
422     bogus results, with non-infinity inputs mapped to infinity too.)
423     [Bodo Moeller]
424
425 Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
426
427  *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
428     handshake can force the use of weak keying material in OpenSSL
429     SSL/TLS clients and servers.
430
431     Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
432     researching this issue. (CVE-2014-0224)
433     [KIKUCHI Masashi, Steve Henson]
434
435  *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
436     OpenSSL DTLS client the code can be made to recurse eventually crashing
437     in a DoS attack.
438
439     Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
440     (CVE-2014-0221)
441     [Imre Rad, Steve Henson]
442
443  *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
444     be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
445     client or server. This is potentially exploitable to run arbitrary
446     code on a vulnerable client or server.
447
448     Thanks to J�ri Aedla for reporting this issue. (CVE-2014-0195)
449     [J�ri Aedla, Steve Henson]
450
451  *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
452     are subject to a denial of service attack.
453
454     Thanks to Felix Gr�bert and Ivan Fratric at Google for discovering
455     this issue. (CVE-2014-3470)
456     [Felix Gr�bert, Ivan Fratric, Steve Henson]
457
458  *) Harmonize version and its documentation. -f flag is used to display
459     compilation flags.
460     [mancha <mancha1@zoho.com>]
461
462  *) Fix eckey_priv_encode so it immediately returns an error upon a failure
463     in i2d_ECPrivateKey.
464     [mancha <mancha1@zoho.com>]
465
466  *) Fix some double frees. These are not thought to be exploitable.
467     [mancha <mancha1@zoho.com>]
468
469 Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
470
471  *) A missing bounds check in the handling of the TLS heartbeat extension
472     can be used to reveal up to 64k of memory to a connected client or
473     server.
474
475     Thanks for Neel Mehta of Google Security for discovering this bug and to
476     Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
477     preparing the fix (CVE-2014-0160)
478     [Adam Langley, Bodo Moeller]
479
480  *) Fix for the attack described in the paper "Recovering OpenSSL
481     ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
482     by Yuval Yarom and Naomi Benger. Details can be obtained from:
483     http://eprint.iacr.org/2014/140
484
485     Thanks to Yuval Yarom and Naomi Benger for discovering this
486     flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
487     [Yuval Yarom and Naomi Benger]
488
489  *) TLS pad extension: draft-agl-tls-padding-03
490
491     Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
492     TLS client Hello record length value would otherwise be > 255 and
493     less that 512 pad with a dummy extension containing zeroes so it
494     is at least 512 bytes long.
495
496     [Adam Langley, Steve Henson]
497
498 Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
499
500  *) Fix for TLS record tampering bug. A carefully crafted invalid 
501     handshake could crash OpenSSL with a NULL pointer exception.
502     Thanks to Anton Johansson for reporting this issues.
503     (CVE-2013-4353)
504
505  *) Keep original DTLS digest and encryption contexts in retransmission
506     structures so we can use the previous session parameters if they need
507     to be resent. (CVE-2013-6450)
508     [Steve Henson]
509
510  *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
511     avoids preferring ECDHE-ECDSA ciphers when the client appears to be
512     Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
513     several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
514     is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
515     10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
516     [Rob Stradling, Adam Langley]
517
518 Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
519
520  *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI
521     supporting platforms or when small records were transferred.
522     [Andy Polyakov, Steve Henson]
523
524 Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
525
526  *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
527
528     This addresses the flaw in CBC record processing discovered by 
529     Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
530     at: http://www.isg.rhul.ac.uk/tls/     
531
532     Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
533     Security Group at Royal Holloway, University of London
534     (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
535     Emilia K�sper for the initial patch.
536     (CVE-2013-0169)
537     [Emilia K�sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson]
538
539  *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
540     ciphersuites which can be exploited in a denial of service attack.
541     Thanks go to and to Adam Langley <agl@chromium.org> for discovering
542     and detecting this bug and to Wolfgang Ettlinger
543     <wolfgang.ettlinger@gmail.com> for independently discovering this issue.
544     (CVE-2012-2686)
545     [Adam Langley]
546
547  *) Return an error when checking OCSP signatures when key is NULL.
548     This fixes a DoS attack. (CVE-2013-0166)
549     [Steve Henson]
550
551  *) Make openssl verify return errors.
552     [Chris Palmer <palmer@google.com> and Ben Laurie]
553
554  *) Call OCSP Stapling callback after ciphersuite has been chosen, so
555     the right response is stapled. Also change SSL_get_certificate()
556     so it returns the certificate actually sent.
557     See http://rt.openssl.org/Ticket/Display.html?id=2836.
558     [Rob Stradling <rob.stradling@comodo.com>]
559
560  *) Fix possible deadlock when decoding public keys.
561     [Steve Henson]
562
563  *) Don't use TLS 1.0 record version number in initial client hello
564     if renegotiating.
565     [Steve Henson]
566
567 Changes between 1.0.1b and 1.0.1c [10 May 2012]
568
569  *) Sanity check record length before skipping explicit IV in TLS
570     1.2, 1.1 and DTLS to fix DoS attack.
571
572     Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
573     fuzzing as a service testing platform.
574     (CVE-2012-2333)
575     [Steve Henson]
576
577  *) Initialise tkeylen properly when encrypting CMS messages.
578     Thanks to Solar Designer of Openwall for reporting this issue.
579     [Steve Henson]
580
581  *) In FIPS mode don't try to use composite ciphers as they are not
582     approved.
583     [Steve Henson]
584
585 Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
586
587  *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
588     1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
589     mean any application compiled against OpenSSL 1.0.0 headers setting
590     SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng
591     TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
592     0x10000000L Any application which was previously compiled against
593     OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
594     will need to be recompiled as a result. Letting be results in
595     inability to disable specifically TLS 1.1 and in client context,
596     in unlike event, limit maximum offered version to TLS 1.0 [see below].
597     [Steve Henson]
598
599  *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not
600     disable just protocol X, but all protocols above X *if* there are
601     protocols *below* X still enabled. In more practical terms it means
602     that if application wants to disable TLS1.0 in favor of TLS1.1 and
603     above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass
604     SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to
605     client side.
606     [Andy Polyakov]
607
608 Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
609
610  *) Check for potentially exploitable overflows in asn1_d2i_read_bio
611     BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
612     in CRYPTO_realloc_clean.
613
614     Thanks to Tavis Ormandy, Google Security Team, for discovering this
615     issue and to Adam Langley <agl@chromium.org> for fixing it.
616     (CVE-2012-2110)
617     [Adam Langley (Google), Tavis Ormandy, Google Security Team]
618
619  *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
620     [Adam Langley]
621
622  *) Workarounds for some broken servers that "hang" if a client hello
623     record length exceeds 255 bytes.
624
625     1. Do not use record version number > TLS 1.0 in initial client
626        hello: some (but not all) hanging servers will now work.
627     2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
628	the number of ciphers sent in the client hello. This should be
629        set to an even number, such as 50, for example by passing:
630        -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
631        Most broken servers should now work.
632     3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
633	TLS 1.2 client support entirely.
634     [Steve Henson]
635
636  *) Fix SEGV in Vector Permutation AES module observed in OpenSSH.
637     [Andy Polyakov]
638
639 Changes between 1.0.0h and 1.0.1  [14 Mar 2012]
640
641  *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET
642     STRING form instead of a DigestInfo.
643     [Steve Henson]
644
645  *) The format used for MDC2 RSA signatures is inconsistent between EVP
646     and the RSA_sign/RSA_verify functions. This was made more apparent when
647     OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
648     those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 
649     the correct format in RSA_verify so both forms transparently work.
650     [Steve Henson]
651
652  *) Some servers which support TLS 1.0 can choke if we initially indicate
653     support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
654     encrypted premaster secret. As a workaround use the maximum pemitted
655     client version in client hello, this should keep such servers happy
656     and still work with previous versions of OpenSSL.
657     [Steve Henson]
658
659  *) Add support for TLS/DTLS heartbeats.
660     [Robin Seggelmann <seggelmann@fh-muenster.de>]
661
662  *) Add support for SCTP.
663     [Robin Seggelmann <seggelmann@fh-muenster.de>]
664
665  *) Improved PRNG seeding for VOS.
666     [Paul Green <Paul.Green@stratus.com>]
667
668  *) Extensive assembler packs updates, most notably:
669
670	- x86[_64]:     AES-NI, PCLMULQDQ, RDRAND support;
671	- x86[_64]:     SSSE3 support (SHA1, vector-permutation AES);
672	- x86_64:       bit-sliced AES implementation;
673	- ARM:          NEON support, contemporary platforms optimizations;
674	- s390x:        z196 support;
675	- *:            GHASH and GF(2^m) multiplication implementations;
676
677     [Andy Polyakov]
678
679  *) Make TLS-SRP code conformant with RFC 5054 API cleanup
680     (removal of unnecessary code)
681     [Peter Sylvester <peter.sylvester@edelweb.fr>]
682
683  *) Add TLS key material exporter from RFC 5705.
684     [Eric Rescorla]
685
686  *) Add DTLS-SRTP negotiation from RFC 5764.
687     [Eric Rescorla]
688
689  *) Add Next Protocol Negotiation,
690     http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be
691     disabled with a no-npn flag to config or Configure. Code donated
692     by Google.
693     [Adam Langley <agl@google.com> and Ben Laurie]
694
695  *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
696     NIST-P256, NIST-P521, with constant-time single point multiplication on
697     typical inputs. Compiler support for the nonstandard type __uint128_t is
698     required to use this (present in gcc 4.4 and later, for 64-bit builds).
699     Code made available under Apache License version 2.0.
700
701     Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
702     line to include this in your build of OpenSSL, and run "make depend" (or
703     "make update"). This enables the following EC_METHODs:
704
705         EC_GFp_nistp224_method()
706         EC_GFp_nistp256_method()
707         EC_GFp_nistp521_method()
708
709     EC_GROUP_new_by_curve_name() will automatically use these (while
710     EC_GROUP_new_curve_GFp() currently prefers the more flexible
711     implementations).
712     [Emilia K�sper, Adam Langley, Bodo Moeller (Google)]
713
714  *) Use type ossl_ssize_t instad of ssize_t which isn't available on
715     all platforms. Move ssize_t definition from e_os.h to the public
716     header file e_os2.h as it now appears in public header file cms.h
717     [Steve Henson]
718
719  *) New -sigopt option to the ca, req and x509 utilities. Additional
720     signature parameters can be passed using this option and in
721     particular PSS. 
722     [Steve Henson]
723
724  *) Add RSA PSS signing function. This will generate and set the
725     appropriate AlgorithmIdentifiers for PSS based on those in the
726     corresponding EVP_MD_CTX structure. No application support yet.
727     [Steve Henson]
728
729  *) Support for companion algorithm specific ASN1 signing routines.
730     New function ASN1_item_sign_ctx() signs a pre-initialised
731     EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
732     the appropriate parameters.
733     [Steve Henson]
734
735  *) Add new algorithm specific ASN1 verification initialisation function
736     to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
737     handling will be the same no matter what EVP_PKEY_METHOD is used.
738     Add a PSS handler to support verification of PSS signatures: checked
739     against a number of sample certificates.
740     [Steve Henson]
741
742  *) Add signature printing for PSS. Add PSS OIDs.
743     [Steve Henson, Martin Kaiser <lists@kaiser.cx>]
744
745  *) Add algorithm specific signature printing. An individual ASN1 method
746     can now print out signatures instead of the standard hex dump. 
747
748     More complex signatures (e.g. PSS) can print out more meaningful
749     information. Include DSA version that prints out the signature
750     parameters r, s.
751     [Steve Henson]
752
753  *) Password based recipient info support for CMS library: implementing
754     RFC3211.
755     [Steve Henson]
756
757  *) Split password based encryption into PBES2 and PBKDF2 functions. This
758     neatly separates the code into cipher and PBE sections and is required
759     for some algorithms that split PBES2 into separate pieces (such as
760     password based CMS).
761     [Steve Henson]
762
763  *) Session-handling fixes:
764     - Fix handling of connections that are resuming with a session ID,
765       but also support Session Tickets.
766     - Fix a bug that suppressed issuing of a new ticket if the client
767       presented a ticket with an expired session.
768     - Try to set the ticket lifetime hint to something reasonable.
769     - Make tickets shorter by excluding irrelevant information.
770     - On the client side, don't ignore renewed tickets.
771     [Adam Langley, Bodo Moeller (Google)]
772
773  *) Fix PSK session representation.
774     [Bodo Moeller]
775
776  *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
777
778     This work was sponsored by Intel.
779     [Andy Polyakov]
780
781  *) Add GCM support to TLS library. Some custom code is needed to split
782     the IV between the fixed (from PRF) and explicit (from TLS record)
783     portions. This adds all GCM ciphersuites supported by RFC5288 and 
784     RFC5289. Generalise some AES* cipherstrings to inlclude GCM and
785     add a special AESGCM string for GCM only.
786     [Steve Henson]
787
788  *) Expand range of ctrls for AES GCM. Permit setting invocation
789     field on decrypt and retrieval of invocation field only on encrypt.
790     [Steve Henson]
791
792  *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
793     As required by RFC5289 these ciphersuites cannot be used if for
794     versions of TLS earlier than 1.2.
795     [Steve Henson]
796
797  *) For FIPS capable OpenSSL interpret a NULL default public key method
798     as unset and return the appopriate default but do *not* set the default.
799     This means we can return the appopriate method in applications that
800     swicth between FIPS and non-FIPS modes.
801     [Steve Henson]
802
803  *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
804     ENGINE is used then we cannot handle that in the FIPS module so we
805     keep original code iff non-FIPS operations are allowed.
806     [Steve Henson]
807
808  *) Add -attime option to openssl utilities.
809     [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson]
810
811  *) Redirect DSA and DH operations to FIPS module in FIPS mode.
812     [Steve Henson]
813
814  *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
815     FIPS EC methods unconditionally for now.
816     [Steve Henson]
817
818  *) New build option no-ec2m to disable characteristic 2 code.
819     [Steve Henson]
820
821  *) Backport libcrypto audit of return value checking from 1.1.0-dev; not
822     all cases can be covered as some introduce binary incompatibilities.
823     [Steve Henson]
824
825  *) Redirect RSA operations to FIPS module including keygen,
826     encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
827     [Steve Henson]
828
829  *) Add similar low level API blocking to ciphers.
830     [Steve Henson]
831
832  *) Low level digest APIs are not approved in FIPS mode: any attempt
833     to use these will cause a fatal error. Applications that *really* want
834     to use them can use the private_* version instead.
835     [Steve Henson]
836
837  *) Redirect cipher operations to FIPS module for FIPS builds. 
838     [Steve Henson]
839
840  *) Redirect digest operations to FIPS module for FIPS builds. 
841     [Steve Henson]
842
843  *) Update build system to add "fips" flag which will link in fipscanister.o
844     for static and shared library builds embedding a signature if needed.
845     [Steve Henson]
846
847  *) Output TLS supported curves in preference order instead of numerical
848     order. This is currently hardcoded for the highest order curves first.
849     This should be configurable so applications can judge speed vs strength.
850     [Steve Henson]
851
852  *) Add TLS v1.2 server support for client authentication. 
853     [Steve Henson]
854
855  *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
856     and enable MD5.
857     [Steve Henson]
858
859  *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying
860     FIPS modules versions.
861     [Steve Henson]
862
863  *) Add TLS v1.2 client side support for client authentication. Keep cache
864     of handshake records longer as we don't know the hash algorithm to use
865     until after the certificate request message is received.
866     [Steve Henson]
867
868  *) Initial TLS v1.2 client support. Add a default signature algorithms
869     extension including all the algorithms we support. Parse new signature
870     format in client key exchange. Relax some ECC signing restrictions for
871     TLS v1.2 as indicated in RFC5246.
872     [Steve Henson]
873
874  *) Add server support for TLS v1.2 signature algorithms extension. Switch
875     to new signature format when needed using client digest preference.
876     All server ciphersuites should now work correctly in TLS v1.2. No client
877     support yet and no support for client certificates.
878     [Steve Henson]
879
880  *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
881     to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
882     ciphersuites. At present only RSA key exchange ciphersuites work with
883     TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
884     SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
885     and version checking.
886     [Steve Henson]
887
888  *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
889     with this defined it will not be affected by any changes to ssl internal
890     structures. Add several utility functions to allow openssl application
891     to work with OPENSSL_NO_SSL_INTERN defined.
892     [Steve Henson]
893
894  *) Add SRP support.
895     [Tom Wu <tjw@cs.stanford.edu> and Ben Laurie]
896
897  *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
898     [Steve Henson]
899
900  *) Permit abbreviated handshakes when renegotiating using the function
901     SSL_renegotiate_abbreviated().
902     [Robin Seggelmann <seggelmann@fh-muenster.de>]
903
904  *) Add call to ENGINE_register_all_complete() to
905     ENGINE_load_builtin_engines(), so some implementations get used
906     automatically instead of needing explicit application support.
907     [Steve Henson]
908
909  *) Add support for TLS key exporter as described in RFC5705.
910     [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson]
911
912  *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
913     a few changes are required:
914
915       Add SSL_OP_NO_TLSv1_1 flag.
916       Add TLSv1_1 methods.
917       Update version checking logic to handle version 1.1.
918       Add explicit IV handling (ported from DTLS code).
919       Add command line options to s_client/s_server.
920     [Steve Henson]
921
922 Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
923
924  *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
925     in CMS and PKCS7 code. When RSA decryption fails use a random key for
926     content decryption and always return the same error. Note: this attack
927     needs on average 2^20 messages so it only affects automated senders. The
928     old behaviour can be reenabled in the CMS code by setting the
929     CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
930     an MMA defence is not necessary.
931     Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
932     this issue. (CVE-2012-0884)
933     [Steve Henson]
934
935  *) Fix CVE-2011-4619: make sure we really are receiving a 
936     client hello before rejecting multiple SGC restarts. Thanks to
937     Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
938     [Steve Henson]
939
940 Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
941
942  *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
943     Thanks to Antonio Martin, Enterprise Secure Access Research and
944     Development, Cisco Systems, Inc. for discovering this bug and
945     preparing a fix. (CVE-2012-0050)
946     [Antonio Martin]
947
948 Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
949
950  *) Nadhem Alfardan and Kenny Paterson have discovered an extension
951     of the Vaudenay padding oracle attack on CBC mode encryption
952     which enables an efficient plaintext recovery attack against
953     the OpenSSL implementation of DTLS. Their attack exploits timing
954     differences arising during decryption processing. A research
955     paper describing this attack can be found at:
956                  http://www.isg.rhul.ac.uk/~kp/dtls.pdf
957     Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
958     Security Group at Royal Holloway, University of London
959     (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
960     <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
961     for preparing the fix. (CVE-2011-4108)
962     [Robin Seggelmann, Michael Tuexen]
963
964  *) Clear bytes used for block padding of SSL 3.0 records.
965     (CVE-2011-4576)
966     [Adam Langley (Google)]
967
968  *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George
969     Kadianakis <desnacked@gmail.com> for discovering this issue and
970     Adam Langley for preparing the fix. (CVE-2011-4619)
971     [Adam Langley (Google)]
972
973  *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027)
974     [Andrey Kulikov <amdeich@gmail.com>]
975
976  *) Prevent malformed RFC3779 data triggering an assertion failure.
977     Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
978     and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577)
979     [Rob Austein <sra@hactrn.net>]
980
981  *) Improved PRNG seeding for VOS.
982     [Paul Green <Paul.Green@stratus.com>]
983
984  *) Fix ssl_ciph.c set-up race.
985     [Adam Langley (Google)]
986
987  *) Fix spurious failures in ecdsatest.c.
988     [Emilia K�sper (Google)]
989
990  *) Fix the BIO_f_buffer() implementation (which was mixing different
991     interpretations of the '..._len' fields).
992     [Adam Langley (Google)]
993
994  *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
995     BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
996     threads won't reuse the same blinding coefficients.
997
998     This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
999     lock to call BN_BLINDING_invert_ex, and avoids one use of
1000     BN_BLINDING_update for each BN_BLINDING structure (previously,
1001     the last update always remained unused).
1002     [Emilia K�sper (Google)]
1003
1004  *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
1005     [Bob Buckholz (Google)]
1006
1007 Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
1008
1009  *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted
1010     by initialising X509_STORE_CTX properly. (CVE-2011-3207)
1011     [Kaspar Brand <ossl@velox.ch>]
1012
1013  *) Fix SSL memory handling for (EC)DH ciphersuites, in particular
1014     for multi-threaded use of ECDH. (CVE-2011-3210)
1015     [Adam Langley (Google)]
1016
1017  *) Fix x509_name_ex_d2i memory leak on bad inputs.
1018     [Bodo Moeller]
1019
1020  *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
1021     signature public key algorithm by using OID xref utilities instead.
1022     Before this you could only use some ECC ciphersuites with SHA1 only.
1023     [Steve Henson]
1024
1025  *) Add protection against ECDSA timing attacks as mentioned in the paper
1026     by Billy Bob Brumley and Nicola Tuveri, see:
1027
1028	http://eprint.iacr.org/2011/232.pdf
1029
1030     [Billy Bob Brumley and Nicola Tuveri]
1031
1032 Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
1033
1034  *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
1035     [Neel Mehta, Adam Langley, Bodo Moeller (Google)]
1036
1037  *) Fix bug in string printing code: if *any* escaping is enabled we must
1038     escape the escape character (backslash) or the resulting string is
1039     ambiguous.
1040     [Steve Henson]
1041
1042 Changes between 1.0.0b and 1.0.0c  [2 Dec 2010]
1043
1044  *) Disable code workaround for ancient and obsolete Netscape browsers
1045     and servers: an attacker can use it in a ciphersuite downgrade attack.
1046     Thanks to Martin Rex for discovering this bug. CVE-2010-4180
1047     [Steve Henson]
1048
1049  *) Fixed J-PAKE implementation error, originally discovered by
1050     Sebastien Martini, further info and confirmation from Stefan
1051     Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
1052     [Ben Laurie]
1053
1054 Changes between 1.0.0a and 1.0.0b  [16 Nov 2010]
1055
1056  *) Fix extension code to avoid race conditions which can result in a buffer
1057     overrun vulnerability: resumed sessions must not be modified as they can
1058     be shared by multiple threads. CVE-2010-3864
1059     [Steve Henson]
1060
1061  *) Fix WIN32 build system to correctly link an ENGINE directory into
1062     a DLL. 
1063     [Steve Henson]
1064
1065 Changes between 1.0.0 and 1.0.0a  [01 Jun 2010]
1066
1067  *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 
1068     (CVE-2010-1633)
1069     [Steve Henson, Peter-Michael Hager <hager@dortmund.net>]
1070
1071 Changes between 0.9.8n and 1.0.0  [29 Mar 2010]
1072
1073  *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
1074     context. The operation can be customised via the ctrl mechanism in
1075     case ENGINEs want to include additional functionality.
1076     [Steve Henson]
1077
1078  *) Tolerate yet another broken PKCS#8 key format: private key value negative.
1079     [Steve Henson]
1080
1081  *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
1082     output hashes compatible with older versions of OpenSSL.
1083     [Willy Weisz <weisz@vcpc.univie.ac.at>]
1084
1085  *) Fix compression algorithm handling: if resuming a session use the
1086     compression algorithm of the resumed session instead of determining
1087     it from client hello again. Don't allow server to change algorithm.
1088     [Steve Henson]
1089
1090  *) Add load_crls() function to apps tidying load_certs() too. Add option
1091     to verify utility to allow additional CRLs to be included.
1092     [Steve Henson]
1093
1094  *) Update OCSP request code to permit adding custom headers to the request:
1095     some responders need this.
1096     [Steve Henson]
1097
1098  *) The function EVP_PKEY_sign() returns <=0 on error: check return code
1099     correctly.
1100     [Julia Lawall <julia@diku.dk>]
1101
1102  *) Update verify callback code in apps/s_cb.c and apps/verify.c, it
1103     needlessly dereferenced structures, used obsolete functions and
1104     didn't handle all updated verify codes correctly.
1105     [Steve Henson]
1106
1107  *) Disable MD2 in the default configuration.
1108     [Steve Henson]
1109
1110  *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
1111     indicate the initial BIO being pushed or popped. This makes it possible
1112     to determine whether the BIO is the one explicitly called or as a result
1113     of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
1114     it handles reference counts correctly and doesn't zero out the I/O bio
1115     when it is not being explicitly popped. WARNING: applications which
1116     included workarounds for the old buggy behaviour will need to be modified
1117     or they could free up already freed BIOs.
1118     [Steve Henson]
1119
1120  *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
1121     renaming to all platforms (within the 0.9.8 branch, this was
1122     done conditionally on Netware platforms to avoid a name clash).
1123     [Guenter <lists@gknw.net>]
1124
1125  *) Add ECDHE and PSK support to DTLS.
1126     [Michael Tuexen <tuexen@fh-muenster.de>]
1127
1128  *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
1129     be used on C++.
1130     [Steve Henson]
1131
1132  *) Add "missing" function EVP_MD_flags() (without this the only way to
1133     retrieve a digest flags is by accessing the structure directly. Update
1134     EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest
1135     or cipher is registered as in the "from" argument. Print out all
1136     registered digests in the dgst usage message instead of manually 
1137     attempting to work them out.
1138     [Steve Henson]
1139
1140  *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
1141     this allows the use of compression and extensions. Change default cipher
1142     string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
1143     by default unless an application cipher string requests it.
1144     [Steve Henson]
1145
1146  *) Alter match criteria in PKCS12_parse(). It used to try to use local
1147     key ids to find matching certificates and keys but some PKCS#12 files
1148     don't follow the (somewhat unwritten) rules and this strategy fails.
1149     Now just gather all certificates together and the first private key
1150     then look for the first certificate that matches the key.
1151     [Steve Henson]
1152
1153  *) Support use of registered digest and cipher names for dgst and cipher
1154     commands instead of having to add each one as a special case. So now
1155     you can do:
1156
1157        openssl sha256 foo
1158
1159     as well as:
1160
1161        openssl dgst -sha256 foo
1162
1163     and this works for ENGINE based algorithms too.
1164
1165     [Steve Henson]
1166
1167  *) Update Gost ENGINE to support parameter files.
1168     [Victor B. Wagner <vitus@cryptocom.ru>]
1169
1170  *) Support GeneralizedTime in ca utility. 
1171     [Oliver Martin <oliver@volatilevoid.net>, Steve Henson]
1172
1173  *) Enhance the hash format used for certificate directory links. The new
1174     form uses the canonical encoding (meaning equivalent names will work
1175     even if they aren't identical) and uses SHA1 instead of MD5. This form
1176     is incompatible with the older format and as a result c_rehash should
1177     be used to rebuild symbolic links.
1178     [Steve Henson]
1179
1180  *) Make PKCS#8 the default write format for private keys, replacing the
1181     traditional format. This form is standardised, more secure and doesn't
1182     include an implicit MD5 dependency.
1183     [Steve Henson]
1184
1185  *) Add a $gcc_devteam_warn option to Configure. The idea is that any code
1186     committed to OpenSSL should pass this lot as a minimum.
1187     [Steve Henson]
1188
1189  *) Add session ticket override functionality for use by EAP-FAST.
1190     [Jouni Malinen <j@w1.fi>]
1191
1192  *) Modify HMAC functions to return a value. Since these can be implemented
1193     in an ENGINE errors can occur.
1194     [Steve Henson]
1195
1196  *) Type-checked OBJ_bsearch_ex.
1197     [Ben Laurie]
1198
1199  *) Type-checked OBJ_bsearch. Also some constification necessitated
1200     by type-checking.  Still to come: TXT_DB, bsearch(?),
1201     OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
1202     CONF_VALUE.
1203     [Ben Laurie]
1204
1205  *) New function OPENSSL_gmtime_adj() to add a specific number of days and
1206     seconds to a tm structure directly, instead of going through OS
1207     specific date routines. This avoids any issues with OS routines such
1208     as the year 2038 bug. New *_adj() functions for ASN1 time structures
1209     and X509_time_adj_ex() to cover the extended range. The existing
1210     X509_time_adj() is still usable and will no longer have any date issues.
1211     [Steve Henson]
1212
1213  *) Delta CRL support. New use deltas option which will attempt to locate
1214     and search any appropriate delta CRLs available.
1215
1216     This work was sponsored by Google.
1217     [Steve Henson]
1218
1219  *) Support for CRLs partitioned by reason code. Reorganise CRL processing
1220     code and add additional score elements. Validate alternate CRL paths
1221     as part of the CRL checking and indicate a new error "CRL path validation
1222     error" in this case. Applications wanting additional details can use
1223     the verify callback and check the new "parent" field. If this is not
1224     NULL CRL path validation is taking place. Existing applications wont
1225     see this because it requires extended CRL support which is off by
1226     default.
1227
1228     This work was sponsored by Google.
1229     [Steve Henson]
1230
1231  *) Support for freshest CRL extension.
1232
1233     This work was sponsored by Google.
1234     [Steve Henson]
1235
1236  *) Initial indirect CRL support. Currently only supported in the CRLs
1237     passed directly and not via lookup. Process certificate issuer
1238     CRL entry extension and lookup CRL entries by bother issuer name
1239     and serial number. Check and process CRL issuer entry in IDP extension.
1240
1241     This work was sponsored by Google.
1242     [Steve Henson]
1243
1244  *) Add support for distinct certificate and CRL paths. The CRL issuer
1245     certificate is validated separately in this case. Only enabled if
1246     an extended CRL support flag is set: this flag will enable additional
1247     CRL functionality in future.
1248
1249     This work was sponsored by Google.
1250     [Steve Henson]
1251
1252  *) Add support for policy mappings extension.
1253
1254     This work was sponsored by Google.
1255     [Steve Henson]
1256
1257  *) Fixes to pathlength constraint, self issued certificate handling,
1258     policy processing to align with RFC3280 and PKITS tests.
1259
1260     This work was sponsored by Google.
1261     [Steve Henson]
1262
1263  *) Support for name constraints certificate extension. DN, email, DNS
1264     and URI types are currently supported.
1265
1266     This work was sponsored by Google.
1267     [Steve Henson]
1268
1269  *) To cater for systems that provide a pointer-based thread ID rather
1270     than numeric, deprecate the current numeric thread ID mechanism and
1271     replace it with a structure and associated callback type. This
1272     mechanism allows a numeric "hash" to be extracted from a thread ID in
1273     either case, and on platforms where pointers are larger than 'long',
1274     mixing is done to help ensure the numeric 'hash' is usable even if it
1275     can't be guaranteed unique. The default mechanism is to use "&errno"
1276     as a pointer-based thread ID to distinguish between threads.
1277
1278     Applications that want to provide their own thread IDs should now use
1279     CRYPTO_THREADID_set_callback() to register a callback that will call
1280     either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
1281
1282     Note that ERR_remove_state() is now deprecated, because it is tied
1283     to the assumption that thread IDs are numeric.  ERR_remove_state(0)
1284     to free the current thread's error state should be replaced by
1285     ERR_remove_thread_state(NULL).
1286
1287     (This new approach replaces the functions CRYPTO_set_idptr_callback(),
1288     CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
1289     OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
1290     application was previously providing a numeric thread callback that
1291     was inappropriate for distinguishing threads, then uniqueness might
1292     have been obtained with &errno that happened immediately in the
1293     intermediate development versions of OpenSSL; this is no longer the
1294     case, the numeric thread callback will now override the automatic use
1295     of &errno.)
1296     [Geoff Thorpe, with help from Bodo Moeller]
1297
1298  *) Initial support for different CRL issuing certificates. This covers a
1299     simple case where the self issued certificates in the chain exist and
1300     the real CRL issuer is higher in the existing chain.
1301
1302     This work was sponsored by Google.
1303     [Steve Henson]
1304
1305  *) Removed effectively defunct crypto/store from the build.
1306     [Ben Laurie]
1307
1308  *) Revamp of STACK to provide stronger type-checking. Still to come:
1309     TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
1310     ASN1_STRING, CONF_VALUE.
1311     [Ben Laurie]
1312
1313  *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
1314     RAM on SSL connections.  This option can save about 34k per idle SSL.
1315     [Nick Mathewson]
1316
1317  *) Revamp of LHASH to provide stronger type-checking. Still to come:
1318     STACK, TXT_DB, bsearch, qsort.
1319     [Ben Laurie]
1320
1321  *) Initial support for Cryptographic Message Syntax (aka CMS) based
1322     on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
1323     support for data, signedData, compressedData, digestedData and
1324     encryptedData, envelopedData types included. Scripts to check against
1325     RFC4134 examples draft and interop and consistency checks of many
1326     content types and variants.
1327     [Steve Henson]
1328
1329  *) Add options to enc utility to support use of zlib compression BIO.
1330     [Steve Henson]
1331
1332  *) Extend mk1mf to support importing of options and assembly language
1333     files from Configure script, currently only included in VC-WIN32.
1334     The assembly language rules can now optionally generate the source
1335     files from the associated perl scripts.
1336     [Steve Henson]
1337
1338  *) Implement remaining functionality needed to support GOST ciphersuites.
1339     Interop testing has been performed using CryptoPro implementations.
1340     [Victor B. Wagner <vitus@cryptocom.ru>]
1341
1342  *) s390x assembler pack.
1343     [Andy Polyakov]
1344
1345  *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
1346     "family."
1347     [Andy Polyakov]
1348
1349  *) Implement Opaque PRF Input TLS extension as specified in
1350     draft-rescorla-tls-opaque-prf-input-00.txt.  Since this is not an
1351     official specification yet and no extension type assignment by
1352     IANA exists, this extension (for now) will have to be explicitly
1353     enabled when building OpenSSL by providing the extension number
1354     to use.  For example, specify an option
1355
1356         -DTLSEXT_TYPE_opaque_prf_input=0x9527
1357
1358     to the "config" or "Configure" script to enable the extension,
1359     assuming extension number 0x9527 (which is a completely arbitrary
1360     and unofficial assignment based on the MD5 hash of the Internet
1361     Draft).  Note that by doing so, you potentially lose
1362     interoperability with other TLS implementations since these might
1363     be using the same extension number for other purposes.
1364
1365     SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
1366     opaque PRF input value to use in the handshake.  This will create
1367     an interal copy of the length-'len' string at 'src', and will
1368     return non-zero for success.
1369
1370     To get more control and flexibility, provide a callback function
1371     by using
1372
1373          SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
1374          SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
1375
1376     where
1377
1378          int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
1379          void *arg;
1380
1381     Callback function 'cb' will be called in handshakes, and is
1382     expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
1383     Argument 'arg' is for application purposes (the value as given to
1384     SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
1385     be provided to the callback function).  The callback function
1386     has to return non-zero to report success: usually 1 to use opaque
1387     PRF input just if possible, or 2 to enforce use of the opaque PRF
1388     input.  In the latter case, the library will abort the handshake
1389     if opaque PRF input is not successfully negotiated.
1390
1391     Arguments 'peerinput' and 'len' given to the callback function
1392     will always be NULL and 0 in the case of a client.  A server will
1393     see the client's opaque PRF input through these variables if
1394     available (NULL and 0 otherwise).  Note that if the server
1395     provides an opaque PRF input, the length must be the same as the
1396     length of the client's opaque PRF input.
1397
1398     Note that the callback function will only be called when creating
1399     a new session (session resumption can resume whatever was
1400     previously negotiated), and will not be called in SSL 2.0
1401     handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
1402     SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
1403     for applications that need to enforce opaque PRF input.
1404
1405     [Bodo Moeller]
1406
1407  *) Update ssl code to support digests other than SHA1+MD5 for handshake
1408     MAC. 
1409
1410     [Victor B. Wagner <vitus@cryptocom.ru>]
1411
1412  *) Add RFC4507 support to OpenSSL. This includes the corrections in
1413     RFC4507bis. The encrypted ticket format is an encrypted encoded
1414     SSL_SESSION structure, that way new session features are automatically
1415     supported.
1416
1417     If a client application caches session in an SSL_SESSION structure
1418     support is transparent because tickets are now stored in the encoded
1419     SSL_SESSION.
1420     
1421     The SSL_CTX structure automatically generates keys for ticket
1422     protection in servers so again support should be possible
1423     with no application modification.
1424
1425     If a client or server wishes to disable RFC4507 support then the option
1426     SSL_OP_NO_TICKET can be set.
1427
1428     Add a TLS extension debugging callback to allow the contents of any client
1429     or server extensions to be examined.
1430
1431     This work was sponsored by Google.
1432     [Steve Henson]
1433
1434  *) Final changes to avoid use of pointer pointer casts in OpenSSL.
1435     OpenSSL should now compile cleanly on gcc 4.2
1436     [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson]
1437
1438  *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
1439     support including streaming MAC support: this is required for GOST
1440     ciphersuite support.
1441     [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson]
1442
1443  *) Add option -stream to use PKCS#7 streaming in smime utility. New
1444     function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
1445     to output in BER and PEM format.
1446     [Steve Henson]
1447
1448  *) Experimental support for use of HMAC via EVP_PKEY interface. This
1449     allows HMAC to be handled via the EVP_DigestSign*() interface. The
1450     EVP_PKEY "key" in this case is the HMAC key, potentially allowing
1451     ENGINE support for HMAC keys which are unextractable. New -mac and
1452     -macopt options to dgst utility.
1453     [Steve Henson]
1454
1455  *) New option -sigopt to dgst utility. Update dgst to use
1456     EVP_Digest{Sign,Verify}*. These two changes make it possible to use
1457     alternative signing paramaters such as X9.31 or PSS in the dgst 
1458     utility.
1459     [Steve Henson]
1460
1461  *) Change ssl_cipher_apply_rule(), the internal function that does
1462     the work each time a ciphersuite string requests enabling
1463     ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
1464     removing ("!foo+bar") a class of ciphersuites: Now it maintains
1465     the order of disabled ciphersuites such that those ciphersuites
1466     that most recently went from enabled to disabled not only stay
1467     in order with respect to each other, but also have higher priority
1468     than other disabled ciphersuites the next time ciphersuites are
1469     enabled again.
1470
1471     This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
1472     the same ciphersuites as with "HIGH" alone, but in a specific
1473     order where the PSK ciphersuites come first (since they are the
1474     most recently disabled ciphersuites when "HIGH" is parsed).
1475
1476     Also, change ssl_create_cipher_list() (using this new
1477     funcionality) such that between otherwise identical
1478     cihpersuites, ephemeral ECDH is preferred over ephemeral DH in
1479     the default order.
1480     [Bodo Moeller]
1481
1482  *) Change ssl_create_cipher_list() so that it automatically
1483     arranges the ciphersuites in reasonable order before starting
1484     to process the rule string.  Thus, the definition for "DEFAULT"
1485     (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
1486     remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH".
1487     This makes it much easier to arrive at a reasonable default order
1488     in applications for which anonymous ciphers are OK (meaning
1489     that you can't actually use DEFAULT).
1490     [Bodo Moeller; suggested by Victor Duchovni]
1491
1492  *) Split the SSL/TLS algorithm mask (as used for ciphersuite string
1493     processing) into multiple integers instead of setting
1494     "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
1495     "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
1496     (These masks as well as the individual bit definitions are hidden
1497     away into the non-exported interface ssl/ssl_locl.h, so this
1498     change to the definition of the SSL_CIPHER structure shouldn't
1499     affect applications.)  This give us more bits for each of these
1500     categories, so there is no longer a need to coagulate AES128 and
1501     AES256 into a single algorithm bit, and to coagulate Camellia128
1502     and Camellia256 into a single algorithm bit, which has led to all
1503     kinds of kludges.
1504
1505     Thus, among other things, the kludge introduced in 0.9.7m and
1506     0.9.8e for masking out AES256 independently of AES128 or masking
1507     out Camellia256 independently of AES256 is not needed here in 0.9.9.
1508
1509     With the change, we also introduce new ciphersuite aliases that
1510     so far were missing: "AES128", "AES256", "CAMELLIA128", and
1511     "CAMELLIA256".
1512     [Bodo Moeller]
1513
1514  *) Add support for dsa-with-SHA224 and dsa-with-SHA256.
1515     Use the leftmost N bytes of the signature input if the input is
1516     larger than the prime q (with N being the size in bytes of q).
1517     [Nils Larsch]
1518
1519  *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
1520     it yet and it is largely untested.
1521     [Steve Henson]
1522
1523  *) Add support for the ecdsa-with-SHA224/256/384/512 signature types.
1524     [Nils Larsch]
1525
1526  *) Initial incomplete changes to avoid need for function casts in OpenSSL
1527     some compilers (gcc 4.2 and later) reject their use. Safestack is
1528     reimplemented.  Update ASN1 to avoid use of legacy functions. 
1529     [Steve Henson]
1530
1531  *) Win32/64 targets are linked with Winsock2.
1532     [Andy Polyakov]
1533
1534  *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
1535     to external functions. This can be used to increase CRL handling 
1536     efficiency especially when CRLs are very large by (for example) storing
1537     the CRL revoked certificates in a database.
1538     [Steve Henson]
1539
1540  *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so
1541     new CRLs added to a directory can be used. New command line option
1542     -verify_return_error to s_client and s_server. This causes real errors
1543     to be returned by the verify callback instead of carrying on no matter
1544     what. This reflects the way a "real world" verify callback would behave.
1545     [Steve Henson]
1546
1547  *) GOST engine, supporting several GOST algorithms and public key formats.
1548     Kindly donated by Cryptocom.
1549     [Cryptocom]
1550
1551  *) Partial support for Issuing Distribution Point CRL extension. CRLs
1552     partitioned by DP are handled but no indirect CRL or reason partitioning
1553     (yet). Complete overhaul of CRL handling: now the most suitable CRL is
1554     selected via a scoring technique which handles IDP and AKID in CRLs.
1555     [Steve Henson]
1556
1557  *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
1558     will ultimately be used for all verify operations: this will remove the
1559     X509_STORE dependency on certificate verification and allow alternative
1560     lookup methods.  X509_STORE based implementations of these two callbacks.
1561     [Steve Henson]
1562
1563  *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
1564     Modify get_crl() to find a valid (unexpired) CRL if possible.
1565     [Steve Henson]
1566
1567  *) New function X509_CRL_match() to check if two CRLs are identical. Normally
1568     this would be called X509_CRL_cmp() but that name is already used by
1569     a function that just compares CRL issuer names. Cache several CRL 
1570     extensions in X509_CRL structure and cache CRLDP in X509.
1571     [Steve Henson]
1572
1573  *) Store a "canonical" representation of X509_NAME structure (ASN1 Name)
1574     this maps equivalent X509_NAME structures into a consistent structure.
1575     Name comparison can then be performed rapidly using memcmp().
1576     [Steve Henson]
1577
1578  *) Non-blocking OCSP request processing. Add -timeout option to ocsp 
1579     utility.
1580     [Steve Henson]
1581
1582  *) Allow digests to supply their own micalg string for S/MIME type using
1583     the ctrl EVP_MD_CTRL_MICALG.
1584     [Steve Henson]
1585
1586  *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the
1587     EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
1588     ctrl. It can then customise the structure before and/or after signing
1589     if necessary.
1590     [Steve Henson]
1591
1592  *) New function OBJ_add_sigid() to allow application defined signature OIDs
1593     to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
1594     to free up any added signature OIDs.
1595     [Steve Henson]
1596
1597  *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
1598     EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
1599     digest and cipher tables. New options added to openssl utility:
1600     list-message-digest-algorithms and list-cipher-algorithms.
1601     [Steve Henson]
1602
1603  *) Change the array representation of binary polynomials: the list
1604     of degrees of non-zero coefficients is now terminated with -1.
1605     Previously it was terminated with 0, which was also part of the
1606     value; thus, the array representation was not applicable to
1607     polynomials where t^0 has coefficient zero.  This change makes
1608     the array representation useful in a more general context.
1609     [Douglas Stebila]
1610
1611  *) Various modifications and fixes to SSL/TLS cipher string
1612     handling.  For ECC, the code now distinguishes between fixed ECDH
1613     with RSA certificates on the one hand and with ECDSA certificates
1614     on the other hand, since these are separate ciphersuites.  The
1615     unused code for Fortezza ciphersuites has been removed.
1616
1617     For consistency with EDH, ephemeral ECDH is now called "EECDH"
1618     (not "ECDHE").  For consistency with the code for DH
1619     certificates, use of ECDH certificates is now considered ECDH
1620     authentication, not RSA or ECDSA authentication (the latter is
1621     merely the CA's signing algorithm and not actively used in the
1622     protocol).
1623
1624     The temporary ciphersuite alias "ECCdraft" is no longer
1625     available, and ECC ciphersuites are no longer excluded from "ALL"
1626     and "DEFAULT".  The following aliases now exist for RFC 4492
1627     ciphersuites, most of these by analogy with the DH case:
1628
1629         kECDHr   - ECDH cert, signed with RSA
1630         kECDHe   - ECDH cert, signed with ECDSA
1631         kECDH    - ECDH cert (signed with either RSA or ECDSA)
1632         kEECDH   - ephemeral ECDH
1633         ECDH     - ECDH cert or ephemeral ECDH
1634
1635         aECDH    - ECDH cert
1636         aECDSA   - ECDSA cert
1637         ECDSA    - ECDSA cert
1638
1639         AECDH    - anonymous ECDH
1640         EECDH    - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
1641
1642     [Bodo Moeller]
1643
1644  *) Add additional S/MIME capabilities for AES and GOST ciphers if supported.
1645     Use correct micalg parameters depending on digest(s) in signed message.
1646     [Steve Henson]
1647
1648  *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
1649     an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
1650     [Steve Henson]
1651
1652  *) Initial engine support for EVP_PKEY_METHOD. New functions to permit
1653     an engine to register a method. Add ENGINE lookups for methods and
1654     functional reference processing.
1655     [Steve Henson]
1656
1657  *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of
1658     EVP_{Sign,Verify}* which allow an application to customise the signature
1659     process.
1660     [Steve Henson]
1661
1662  *) New -resign option to smime utility. This adds one or more signers
1663     to an existing PKCS#7 signedData structure. Also -md option to use an
1664     alternative message digest algorithm for signing.
1665     [Steve Henson]
1666
1667  *) Tidy up PKCS#7 routines and add new functions to make it easier to
1668     create PKCS7 structures containing multiple signers. Update smime
1669     application to support multiple signers.
1670     [Steve Henson]
1671
1672  *) New -macalg option to pkcs12 utility to allow setting of an alternative
1673     digest MAC.
1674     [Steve Henson]
1675
1676  *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
1677     Reorganize PBE internals to lookup from a static table using NIDs,
1678     add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
1679     EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
1680     PRF which will be automatically used with PBES2.
1681     [Steve Henson]
1682
1683  *) Replace the algorithm specific calls to generate keys in "req" with the
1684     new API.
1685     [Steve Henson]
1686
1687  *) Update PKCS#7 enveloped data routines to use new API. This is now
1688     supported by any public key method supporting the encrypt operation. A
1689     ctrl is added to allow the public key algorithm to examine or modify
1690     the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
1691     a no op.
1692     [Steve Henson]
1693
1694  *) Add a ctrl to asn1 method to allow a public key algorithm to express
1695     a default digest type to use. In most cases this will be SHA1 but some
1696     algorithms (such as GOST) need to specify an alternative digest. The
1697     return value indicates how strong the prefernce is 1 means optional and
1698     2 is mandatory (that is it is the only supported type). Modify
1699     ASN1_item_sign() to accept a NULL digest argument to indicate it should
1700     use the default md. Update openssl utilities to use the default digest
1701     type for signing if it is not explicitly indicated.
1702     [Steve Henson]
1703
1704  *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 
1705     EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
1706     signing method from the key type. This effectively removes the link
1707     between digests and public key types.
1708     [Steve Henson]
1709
1710  *) Add an OID cross reference table and utility functions. Its purpose is to
1711     translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
1712     rsaEncryption. This will allow some of the algorithm specific hackery
1713     needed to use the correct OID to be removed. 
1714     [Steve Henson]
1715
1716  *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
1717     structures for PKCS7_sign(). They are now set up by the relevant public
1718     key ASN1 method.
1719     [Steve Henson]
1720
1721  *) Add provisional EC pkey method with support for ECDSA and ECDH.
1722     [Steve Henson]
1723
1724  *) Add support for key derivation (agreement) in the API, DH method and
1725     pkeyutl.
1726     [Steve Henson]
1727
1728  *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
1729     public and private key formats. As a side effect these add additional 
1730     command line functionality not previously available: DSA signatures can be
1731     generated and verified using pkeyutl and DH key support and generation in
1732     pkey, genpkey.
1733     [Steve Henson]
1734
1735  *) BeOS support.
1736     [Oliver Tappe <zooey@hirschkaefer.de>]
1737
1738  *) New make target "install_html_docs" installs HTML renditions of the
1739     manual pages.
1740     [Oliver Tappe <zooey@hirschkaefer.de>]
1741
1742  *) New utility "genpkey" this is analagous to "genrsa" etc except it can
1743     generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
1744     support key and parameter generation and add initial key generation
1745     functionality for RSA.
1746     [Steve Henson]
1747
1748  *) Add functions for main EVP_PKEY_method operations. The undocumented
1749     functions EVP_PKEY_{encrypt,decrypt} have been renamed to
1750     EVP_PKEY_{encrypt,decrypt}_old. 
1751     [Steve Henson]
1752
1753  *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public
1754     key API, doesn't do much yet.
1755     [Steve Henson]
1756
1757  *) New function EVP_PKEY_asn1_get0_info() to retrieve information about
1758     public key algorithms. New option to openssl utility:
1759     "list-public-key-algorithms" to print out info.
1760     [Steve Henson]
1761
1762  *) Implement the Supported Elliptic Curves Extension for
1763     ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
1764     [Douglas Stebila]
1765
1766  *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
1767     EVP_CIPHER structures to avoid later problems in EVP_cleanup().
1768     [Steve Henson]
1769
1770  *) New utilities pkey and pkeyparam. These are similar to algorithm specific
1771     utilities such as rsa, dsa, dsaparam etc except they process any key
1772     type.
1773     [Steve Henson]
1774
1775  *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 
1776     functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
1777     EVP_PKEY_print_param() to print public key data from an EVP_PKEY
1778     structure.
1779     [Steve Henson]
1780
1781  *) Initial support for pluggable public key ASN1.
1782     De-spaghettify the public key ASN1 handling. Move public and private
1783     key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
1784     algorithm specific handling to a single module within the relevant
1785     algorithm directory. Add functions to allow (near) opaque processing
1786     of public and private key structures.
1787     [Steve Henson]
1788
1789  *) Implement the Supported Point Formats Extension for
1790     ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
1791     [Douglas Stebila]
1792
1793  *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
1794     for the psk identity [hint] and the psk callback functions to the
1795     SSL_SESSION, SSL and SSL_CTX structure.
1796     
1797     New ciphersuites:
1798         PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
1799         PSK-AES256-CBC-SHA
1800 
1801     New functions:
1802         SSL_CTX_use_psk_identity_hint
1803         SSL_get_psk_identity_hint
1804         SSL_get_psk_identity
1805         SSL_use_psk_identity_hint
1806
1807     [Mika Kousa and Pasi Eronen of Nokia Corporation]
1808
1809  *) Add RFC 3161 compliant time stamp request creation, response generation
1810     and response verification functionality.
1811     [Zolt�n Gl�zik <zglozik@opentsa.org>, The OpenTSA Project]
1812
1813  *) Add initial support for TLS extensions, specifically for the server_name
1814     extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
1815     have new members for a host name.  The SSL data structure has an
1816     additional member SSL_CTX *initial_ctx so that new sessions can be
1817     stored in that context to allow for session resumption, even after the
1818     SSL has been switched to a new SSL_CTX in reaction to a client's
1819     server_name extension.
1820
1821     New functions (subject to change):
1822
1823         SSL_get_servername()
1824         SSL_get_servername_type()
1825         SSL_set_SSL_CTX()
1826
1827     New CTRL codes and macros (subject to change):
1828
1829         SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
1830                                 - SSL_CTX_set_tlsext_servername_callback()
1831         SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
1832                                      - SSL_CTX_set_tlsext_servername_arg()
1833         SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
1834
1835     openssl s_client has a new '-servername ...' option.
1836
1837     openssl s_server has new options '-servername_host ...', '-cert2 ...',
1838     '-key2 ...', '-servername_fatal' (subject to change).  This allows
1839     testing the HostName extension for a specific single host name ('-cert'
1840     and '-key' remain fallbacks for handshakes without HostName
1841     negotiation).  If the unrecogninzed_name alert has to be sent, this by
1842     default is a warning; it becomes fatal with the '-servername_fatal'
1843     option.
1844
1845     [Peter Sylvester,  Remy Allais, Christophe Renou]
1846
1847  *) Whirlpool hash implementation is added.
1848     [Andy Polyakov]
1849
1850  *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
1851     bn(64,32). Because of instruction set limitations it doesn't have
1852     any negative impact on performance. This was done mostly in order
1853     to make it possible to share assembler modules, such as bn_mul_mont
1854     implementations, between 32- and 64-bit builds without hassle.
1855     [Andy Polyakov]
1856
1857  *) Move code previously exiled into file crypto/ec/ec2_smpt.c
1858     to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
1859     macro.
1860     [Bodo Moeller]
1861
1862  *) New candidate for BIGNUM assembler implementation, bn_mul_mont,
1863     dedicated Montgomery multiplication procedure, is introduced.
1864     BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
1865     "64-bit" performance on certain 32-bit targets.
1866     [Andy Polyakov]
1867
1868  *) New option SSL_OP_NO_COMP to disable use of compression selectively
1869     in SSL structures. New SSL ctrl to set maximum send fragment size. 
1870     Save memory by seeting the I/O buffer sizes dynamically instead of
1871     using the maximum available value.
1872     [Steve Henson]
1873
1874  *) New option -V for 'openssl ciphers'. This prints the ciphersuite code
1875     in addition to the text details.
1876     [Bodo Moeller]
1877
1878  *) Very, very preliminary EXPERIMENTAL support for printing of general
1879     ASN1 structures. This currently produces rather ugly output and doesn't
1880     handle several customised structures at all.
1881     [Steve Henson]
1882
1883  *) Integrated support for PVK file format and some related formats such
1884     as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
1885     these in the 'rsa' and 'dsa' utilities.
1886     [Steve Henson]
1887
1888  *) Support for PKCS#1 RSAPublicKey format on rsa utility command line.
1889     [Steve Henson]
1890
1891  *) Remove the ancient ASN1_METHOD code. This was only ever used in one
1892     place for the (very old) "NETSCAPE" format certificates which are now
1893     handled using new ASN1 code equivalents.
1894     [Steve Henson]
1895
1896  *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
1897     pointer and make the SSL_METHOD parameter in SSL_CTX_new,
1898     SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
1899     [Nils Larsch]
1900
1901  *) Modify CRL distribution points extension code to print out previously
1902     unsupported fields. Enhance extension setting code to allow setting of
1903     all fields.
1904     [Steve Henson]
1905
1906  *) Add print and set support for Issuing Distribution Point CRL extension.
1907     [Steve Henson]
1908
1909  *) Change 'Configure' script to enable Camellia by default.
1910     [NTT]
1911
1912 Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
1913
1914  *) When rejecting SSL/TLS records due to an incorrect version number, never
1915     update s->server with a new major version number.  As of
1916     - OpenSSL 0.9.8m if 'short' is a 16-bit type,
1917     - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
1918     the previous behavior could result in a read attempt at NULL when
1919     receiving specific incorrect SSL/TLS records once record payload
1920     protection is active.  (CVE-2010-0740)
1921     [Bodo Moeller, Adam Langley <agl@chromium.org>]
1922
1923  *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 
1924     could be crashed if the relevant tables were not present (e.g. chrooted).
1925     [Tomas Hoger <thoger@redhat.com>]
1926
1927 Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
1928
1929  *) Always check bn_wexpend() return values for failure.  (CVE-2009-3245)
1930     [Martin Olsson, Neel Mehta]
1931
1932  *) Fix X509_STORE locking: Every 'objs' access requires a lock (to
1933     accommodate for stack sorting, always a write lock!).
1934     [Bodo Moeller]
1935
1936  *) On some versions of WIN32 Heap32Next is very slow. This can cause
1937     excessive delays in the RAND_poll(): over a minute. As a workaround
1938     include a time check in the inner Heap32Next loop too.
1939     [Steve Henson]
1940
1941  *) The code that handled flushing of data in SSL/TLS originally used the
1942     BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
1943     the problem outlined in PR#1949. The fix suggested there however can
1944     trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
1945     of Apache). So instead simplify the code to flush unconditionally.
1946     This should be fine since flushing with no data to flush is a no op.
1947     [Steve Henson]
1948
1949  *) Handle TLS versions 2.0 and later properly and correctly use the
1950     highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
1951     off ancient servers have a habit of sticking around for a while...
1952     [Steve Henson]
1953
1954  *) Modify compression code so it frees up structures without using the
1955     ex_data callbacks. This works around a problem where some applications
1956     call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
1957     restarting) then use compression (e.g. SSL with compression) later.
1958     This results in significant per-connection memory leaks and
1959     has caused some security issues including CVE-2008-1678 and
1960     CVE-2009-4355.
1961     [Steve Henson]
1962
1963  *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
1964     change when encrypting or decrypting.
1965     [Bodo Moeller]
1966
1967  *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
1968     connect and renegotiate with servers which do not support RI.
1969     Until RI is more widely deployed this option is enabled by default.
1970     [Steve Henson]
1971
1972  *) Add "missing" ssl ctrls to clear options and mode.
1973     [Steve Henson]
1974
1975  *) If client attempts to renegotiate and doesn't support RI respond with
1976     a no_renegotiation alert as required by RFC5746.  Some renegotiating
1977     TLS clients will continue a connection gracefully when they receive
1978     the alert. Unfortunately OpenSSL mishandled this alert and would hang
1979     waiting for a server hello which it will never receive. Now we treat a
1980     received no_renegotiation alert as a fatal error. This is because
1981     applications requesting a renegotiation might well expect it to succeed
1982     and would have no code in place to handle the server denying it so the
1983     only safe thing to do is to terminate the connection.
1984     [Steve Henson]
1985
1986  *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
1987     peer supports secure renegotiation and 0 otherwise. Print out peer
1988     renegotiation support in s_client/s_server.
1989     [Steve Henson]
1990
1991  *) Replace the highly broken and deprecated SPKAC certification method with
1992     the updated NID creation version. This should correctly handle UTF8.
1993     [Steve Henson]
1994
1995  *) Implement RFC5746. Re-enable renegotiation but require the extension
1996     as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
1997     turns out to be a bad idea. It has been replaced by
1998     SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
1999     SSL_CTX_set_options(). This is really not recommended unless you
2000     know what you are doing.
2001     [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson]
2002
2003  *) Fixes to stateless session resumption handling. Use initial_ctx when
2004     issuing and attempting to decrypt tickets in case it has changed during
2005     servername handling. Use a non-zero length session ID when attempting
2006     stateless session resumption: this makes it possible to determine if
2007     a resumption has occurred immediately after receiving server hello
2008     (several places in OpenSSL subtly assume this) instead of later in
2009     the handshake.
2010     [Steve Henson]
2011
2012  *) The functions ENGINE_ctrl(), OPENSSL_isservice(),
2013     CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
2014     fixes for a few places where the return code is not checked
2015     correctly.
2016     [Julia Lawall <julia@diku.dk>]
2017
2018  *) Add --strict-warnings option to Configure script to include devteam
2019     warnings in other configurations.
2020     [Steve Henson]
2021
2022  *) Add support for --libdir option and LIBDIR variable in makefiles. This
2023     makes it possible to install openssl libraries in locations which
2024     have names other than "lib", for example "/usr/lib64" which some
2025     systems need.
2026     [Steve Henson, based on patch from Jeremy Utley]
2027
2028  *) Don't allow the use of leading 0x80 in OIDs. This is a violation of
2029     X690 8.9.12 and can produce some misleading textual output of OIDs.
2030     [Steve Henson, reported by Dan Kaminsky]
2031
2032  *) Delete MD2 from algorithm tables. This follows the recommendation in
2033     several standards that it is not used in new applications due to
2034     several cryptographic weaknesses. For binary compatibility reasons
2035     the MD2 API is still compiled in by default.
2036     [Steve Henson]
2037
2038  *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
2039     and restored.
2040     [Steve Henson]
2041
2042  *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
2043     OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
2044     clash.
2045     [Guenter <lists@gknw.net>]
2046
2047  *) Fix the server certificate chain building code to use X509_verify_cert(),
2048     it used to have an ad-hoc builder which was unable to cope with anything
2049     other than a simple chain.
2050     [David Woodhouse <dwmw2@infradead.org>, Steve Henson]
2051
2052  *) Don't check self signed certificate signatures in X509_verify_cert()
2053     by default (a flag can override this): it just wastes time without
2054     adding any security. As a useful side effect self signed root CAs
2055     with non-FIPS digests are now usable in FIPS mode.
2056     [Steve Henson]
2057
2058  *) In dtls1_process_out_of_seq_message() the check if the current message
2059     is already buffered was missing. For every new message was memory
2060     allocated, allowing an attacker to perform an denial of service attack
2061     with sending out of seq handshake messages until there is no memory
2062     left. Additionally every future messege was buffered, even if the
2063     sequence number made no sense and would be part of another handshake.
2064     So only messages with sequence numbers less than 10 in advance will be
2065     buffered.  (CVE-2009-1378)
2066     [Robin Seggelmann, discovered by Daniel Mentz] 	
2067
2068  *) Records are buffered if they arrive with a future epoch to be
2069     processed after finishing the corresponding handshake. There is
2070     currently no limitation to this buffer allowing an attacker to perform
2071     a DOS attack with sending records with future epochs until there is no
2072     memory left. This patch adds the pqueue_size() function to detemine
2073     the size of a buffer and limits the record buffer to 100 entries.
2074     (CVE-2009-1377)
2075     [Robin Seggelmann, discovered by Daniel Mentz] 	
2076
2077  *) Keep a copy of frag->msg_header.frag_len so it can be used after the
2078     parent structure is freed.  (CVE-2009-1379)
2079     [Daniel Mentz] 	
2080
2081  *) Handle non-blocking I/O properly in SSL_shutdown() call.
2082     [Darryl Miles <darryl-mailinglists@netbauds.net>]
2083
2084  *) Add 2.5.4.* OIDs
2085     [Ilya O. <vrghost@gmail.com>]
2086
2087 Changes between 0.9.8k and 0.9.8l  [5 Nov 2009]
2088
2089  *) Disable renegotiation completely - this fixes a severe security
2090     problem (CVE-2009-3555) at the cost of breaking all
2091     renegotiation. Renegotiation can be re-enabled by setting
2092     SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
2093     run-time. This is really not recommended unless you know what
2094     you're doing.
2095     [Ben Laurie]
2096
2097 Changes between 0.9.8j and 0.9.8k  [25 Mar 2009]
2098
2099  *) Don't set val to NULL when freeing up structures, it is freed up by
2100     underlying code. If sizeof(void *) > sizeof(long) this can result in
2101     zeroing past the valid field. (CVE-2009-0789)
2102     [Paolo Ganci <Paolo.Ganci@AdNovum.CH>]
2103
2104  *) Fix bug where return value of CMS_SignerInfo_verify_content() was not
2105     checked correctly. This would allow some invalid signed attributes to
2106     appear to verify correctly. (CVE-2009-0591)
2107     [Ivan Nestlerode <inestlerode@us.ibm.com>]
2108
2109  *) Reject UniversalString and BMPString types with invalid lengths. This
2110     prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
2111     a legal length. (CVE-2009-0590)
2112     [Steve Henson]
2113
2114  *) Set S/MIME signing as the default purpose rather than setting it 
2115     unconditionally. This allows applications to override it at the store
2116     level.
2117     [Steve Henson]
2118
2119  *) Permit restricted recursion of ASN1 strings. This is needed in practice
2120     to handle some structures.
2121     [Steve Henson]
2122
2123  *) Improve efficiency of mem_gets: don't search whole buffer each time
2124     for a '\n'
2125     [Jeremy Shapiro <jnshapir@us.ibm.com>]
2126
2127  *) New -hex option for openssl rand.
2128     [Matthieu Herrb]
2129
2130  *) Print out UTF8String and NumericString when parsing ASN1.
2131     [Steve Henson]
2132
2133  *) Support NumericString type for name components.
2134     [Steve Henson]
2135
2136  *) Allow CC in the environment to override the automatically chosen
2137     compiler. Note that nothing is done to ensure flags work with the
2138     chosen compiler.
2139     [Ben Laurie]
2140
2141 Changes between 0.9.8i and 0.9.8j  [07 Jan 2009]
2142
2143  *) Properly check EVP_VerifyFinal() and similar return values
2144     (CVE-2008-5077).
2145     [Ben Laurie, Bodo Moeller, Google Security Team]
2146
2147  *) Enable TLS extensions by default.
2148     [Ben Laurie]
2149
2150  *) Allow the CHIL engine to be loaded, whether the application is
2151     multithreaded or not. (This does not release the developer from the
2152     obligation to set up the dynamic locking callbacks.)
2153     [Sander Temme <sander@temme.net>]
2154
2155  *) Use correct exit code if there is an error in dgst command.
2156     [Steve Henson; problem pointed out by Roland Dirlewanger]
2157
2158  *) Tweak Configure so that you need to say "experimental-jpake" to enable
2159     JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
2160     [Bodo Moeller]
2161
2162  *) Add experimental JPAKE support, including demo authentication in
2163     s_client and s_server.
2164     [Ben Laurie]
2165
2166  *) Set the comparison function in v3_addr_canonize().
2167     [Rob Austein <sra@hactrn.net>]
2168
2169  *) Add support for XMPP STARTTLS in s_client.
2170     [Philip Paeps <philip@freebsd.org>]
2171
2172  *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
2173     to ensure that even with this option, only ciphersuites in the
2174     server's preference list will be accepted.  (Note that the option
2175     applies only when resuming a session, so the earlier behavior was
2176     just about the algorithm choice for symmetric cryptography.)
2177     [Bodo Moeller]
2178
2179 Changes between 0.9.8h and 0.9.8i  [15 Sep 2008]
2180
2181  *) Fix NULL pointer dereference if a DTLS server received
2182     ChangeCipherSpec as first record (CVE-2009-1386).
2183     [PR #1679]
2184
2185  *) Fix a state transitition in s3_srvr.c and d1_srvr.c
2186     (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...).
2187     [Nagendra Modadugu]
2188
2189  *) The fix in 0.9.8c that supposedly got rid of unsafe
2190     double-checked locking was incomplete for RSA blinding,
2191     addressing just one layer of what turns out to have been
2192     doubly unsafe triple-checked locking.
2193
2194     So now fix this for real by retiring the MONT_HELPER macro
2195     in crypto/rsa/rsa_eay.c.
2196
2197     [Bodo Moeller; problem pointed out by Marius Schilder]
2198
2199  *) Various precautionary measures:
2200
2201     - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
2202
2203     - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
2204       (NB: This would require knowledge of the secret session ticket key
2205       to exploit, in which case you'd be SOL either way.)
2206
2207     - Change bn_nist.c so that it will properly handle input BIGNUMs
2208       outside the expected range.
2209
2210     - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
2211       builds.
2212
2213     [Neel Mehta, Bodo Moeller]
2214
2215  *) Allow engines to be "soft loaded" - i.e. optionally don't die if
2216     the load fails. Useful for distros.
2217     [Ben Laurie and the FreeBSD team]
2218
2219  *) Add support for Local Machine Keyset attribute in PKCS#12 files.
2220     [Steve Henson]
2221
2222  *) Fix BN_GF2m_mod_arr() top-bit cleanup code.
2223     [Huang Ying]
2224
2225  *) Expand ENGINE to support engine supplied SSL client certificate functions.
2226
2227     This work was sponsored by Logica.
2228     [Steve Henson]
2229
2230  *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
2231     keystores. Support for SSL/TLS client authentication too.
2232     Not compiled unless enable-capieng specified to Configure.
2233
2234     This work was sponsored by Logica.
2235     [Steve Henson]
2236
2237  *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using
2238     ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
2239     attribute creation routines such as certifcate requests and PKCS#12
2240     files.
2241     [Steve Henson]
2242
2243 Changes between 0.9.8g and 0.9.8h  [28 May 2008]
2244
2245  *) Fix flaw if 'Server Key exchange message' is omitted from a TLS
2246     handshake which could lead to a cilent crash as found using the
2247     Codenomicon TLS test suite (CVE-2008-1672) 
2248     [Steve Henson, Mark Cox]
2249
2250  *) Fix double free in TLS server name extensions which could lead to
2251     a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 
2252     [Joe Orton]
2253
2254  *) Clear error queue in SSL_CTX_use_certificate_chain_file()
2255
2256     Clear the error queue to ensure that error entries left from
2257     older function calls do not interfere with the correct operation.
2258     [Lutz Jaenicke, Erik de Castro Lopo]
2259
2260  *) Remove root CA certificates of commercial CAs:
2261
2262     The OpenSSL project does not recommend any specific CA and does not
2263     have any policy with respect to including or excluding any CA.
2264     Therefore it does not make any sense to ship an arbitrary selection
2265     of root CA certificates with the OpenSSL software.
2266     [Lutz Jaenicke]
2267
2268  *) RSA OAEP patches to fix two separate invalid memory reads.
2269     The first one involves inputs when 'lzero' is greater than
2270     'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
2271     before the beginning of from). The second one involves inputs where
2272     the 'db' section contains nothing but zeroes (there is a one-byte
2273     invalid read after the end of 'db').
2274     [Ivan Nestlerode <inestlerode@us.ibm.com>]
2275
2276  *) Partial backport from 0.9.9-dev:
2277
2278     Introduce bn_mul_mont (dedicated Montgomery multiplication
2279     procedure) as a candidate for BIGNUM assembler implementation.
2280     While 0.9.9-dev uses assembler for various architectures, only
2281     x86_64 is available by default here in the 0.9.8 branch, and
2282     32-bit x86 is available through a compile-time setting.
2283
2284     To try the 32-bit x86 assembler implementation, use Configure
2285     option "enable-montasm" (which exists only for this backport).
2286
2287     As "enable-montasm" for 32-bit x86 disclaims code stability
2288     anyway, in this constellation we activate additional code
2289     backported from 0.9.9-dev for further performance improvements,
2290     namely BN_from_montgomery_word.  (To enable this otherwise,
2291     e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".)
2292
2293     [Andy Polyakov (backport partially by Bodo Moeller)]
2294
2295  *) Add TLS session ticket callback. This allows an application to set
2296     TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
2297     values. This is useful for key rollover for example where several key
2298     sets may exist with different names.
2299     [Steve Henson]
2300
2301  *) Reverse ENGINE-internal logic for caching default ENGINE handles.
2302     This was broken until now in 0.9.8 releases, such that the only way
2303     a registered ENGINE could be used (assuming it initialises
2304     successfully on the host) was to explicitly set it as the default
2305     for the relevant algorithms. This is in contradiction with 0.9.7
2306     behaviour and the documentation. With this fix, when an ENGINE is
2307     registered into a given algorithm's table of implementations, the
2308     'uptodate' flag is reset so that auto-discovery will be used next
2309     time a new context for that algorithm attempts to select an
2310     implementation.
2311     [Ian Lister (tweaked by Geoff Thorpe)]
2312
2313  *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
2314     implemention in the following ways:
2315
2316     Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
2317     hard coded.
2318
2319     Lack of BER streaming support means one pass streaming processing is
2320     only supported if data is detached: setting the streaming flag is
2321     ignored for embedded content.
2322
2323     CMS support is disabled by default and must be explicitly enabled
2324     with the enable-cms configuration option.
2325     [Steve Henson]
2326
2327  *) Update the GMP engine glue to do direct copies between BIGNUM and
2328     mpz_t when openssl and GMP use the same limb size. Otherwise the
2329     existing "conversion via a text string export" trick is still used.
2330     [Paul Sheer <paulsheer@gmail.com>]
2331
2332  *) Zlib compression BIO. This is a filter BIO which compressed and
2333     uncompresses any data passed through it.
2334     [Steve Henson]
2335
2336  *) Add AES_wrap_key() and AES_unwrap_key() functions to implement
2337     RFC3394 compatible AES key wrapping.
2338     [Steve Henson]
2339
2340  *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
2341     sets string data without copying. X509_ALGOR_set0() and
2342     X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
2343     data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
2344     from an X509_ATTRIBUTE structure optionally checking it occurs only
2345     once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
2346     data.
2347     [Steve Henson]
2348
2349  *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
2350     to get the expected BN_FLG_CONSTTIME behavior.
2351     [Bodo Moeller (Google)]
2352  
2353  *) Netware support:
2354
2355     - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
2356     - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
2357     - added some more tests to do_tests.pl
2358     - fixed RunningProcess usage so that it works with newer LIBC NDKs too
2359     - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
2360     - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
2361       netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
2362     - various changes to netware.pl to enable gcc-cross builds on Win32
2363       platform
2364     - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
2365     - various changes to fix missing prototype warnings
2366     - fixed x86nasm.pl to create correct asm files for NASM COFF output
2367     - added AES, WHIRLPOOL and CPUID assembler code to build files
2368     - added missing AES assembler make rules to mk1mf.pl
2369     - fixed order of includes in apps/ocsp.c so that e_os.h settings apply
2370     [Guenter Knauf <eflash@gmx.net>]
2371
2372  *) Implement certificate status request TLS extension defined in RFC3546.
2373     A client can set the appropriate parameters and receive the encoded
2374     OCSP response via a callback. A server can query the supplied parameters
2375     and set the encoded OCSP response in the callback. Add simplified examples
2376     to s_client and s_server.
2377     [Steve Henson]
2378
2379 Changes between 0.9.8f and 0.9.8g  [19 Oct 2007]
2380
2381  *) Fix various bugs:
2382     + Binary incompatibility of ssl_ctx_st structure
2383     + DTLS interoperation with non-compliant servers
2384     + Don't call get_session_cb() without proposed session
2385     + Fix ia64 assembler code
2386     [Andy Polyakov, Steve Henson]
2387
2388 Changes between 0.9.8e and 0.9.8f  [11 Oct 2007]
2389
2390  *) DTLS Handshake overhaul. There were longstanding issues with
2391     OpenSSL DTLS implementation, which were making it impossible for
2392     RFC 4347 compliant client to communicate with OpenSSL server.
2393     Unfortunately just fixing these incompatibilities would "cut off"
2394     pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
2395     server keeps tolerating non RFC compliant syntax. The opposite is
2396     not true, 0.9.8f client can not communicate with earlier server.
2397     This update even addresses CVE-2007-4995.
2398     [Andy Polyakov]
2399
2400  *) Changes to avoid need for function casts in OpenSSL: some compilers
2401     (gcc 4.2 and later) reject their use.
2402     [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
2403      Steve Henson]
2404  
2405  *) Add RFC4507 support to OpenSSL. This includes the corrections in
2406     RFC4507bis. The encrypted ticket format is an encrypted encoded
2407     SSL_SESSION structure, that way new session features are automatically
2408     supported.
2409
2410     If a client application caches session in an SSL_SESSION structure
2411     support is transparent because tickets are now stored in the encoded
2412     SSL_SESSION.
2413     
2414     The SSL_CTX structure automatically generates keys for ticket
2415     protection in servers so again support should be possible
2416     with no application modification.
2417
2418     If a client or server wishes to disable RFC4507 support then the option
2419     SSL_OP_NO_TICKET can be set.
2420
2421     Add a TLS extension debugging callback to allow the contents of any client
2422     or server extensions to be examined.
2423
2424     This work was sponsored by Google.
2425     [Steve Henson]
2426
2427  *) Add initial support for TLS extensions, specifically for the server_name
2428     extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
2429     have new members for a host name.  The SSL data structure has an
2430     additional member SSL_CTX *initial_ctx so that new sessions can be
2431     stored in that context to allow for session resumption, even after the
2432     SSL has been switched to a new SSL_CTX in reaction to a client's
2433     server_name extension.
2434
2435     New functions (subject to change):
2436
2437         SSL_get_servername()
2438         SSL_get_servername_type()
2439         SSL_set_SSL_CTX()
2440
2441     New CTRL codes and macros (subject to change):
2442
2443         SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
2444                                 - SSL_CTX_set_tlsext_servername_callback()
2445         SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
2446                                      - SSL_CTX_set_tlsext_servername_arg()
2447         SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
2448
2449     openssl s_client has a new '-servername ...' option.
2450
2451     openssl s_server has new options '-servername_host ...', '-cert2 ...',
2452     '-key2 ...', '-servername_fatal' (subject to change).  This allows
2453     testing the HostName extension for a specific single host name ('-cert'
2454     and '-key' remain fallbacks for handshakes without HostName
2455     negotiation).  If the unrecogninzed_name alert has to be sent, this by
2456     default is a warning; it becomes fatal with the '-servername_fatal'
2457     option.
2458
2459     [Peter Sylvester,  Remy Allais, Christophe Renou, Steve Henson]
2460
2461  *) Add AES and SSE2 assembly language support to VC++ build.
2462     [Steve Henson]
2463
2464  *) Mitigate attack on final subtraction in Montgomery reduction.
2465     [Andy Polyakov]
2466
2467  *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
2468     (which previously caused an internal error).
2469     [Bodo Moeller]
2470
2471  *) Squeeze another 10% out of IGE mode when in != out.
2472     [Ben Laurie]
2473
2474  *) AES IGE mode speedup.
2475     [Dean Gaudet (Google)]
2476
2477  *) Add the Korean symmetric 128-bit cipher SEED (see
2478     http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and
2479     add SEED ciphersuites from RFC 4162:
2480
2481        TLS_RSA_WITH_SEED_CBC_SHA      =  "SEED-SHA"
2482        TLS_DHE_DSS_WITH_SEED_CBC_SHA  =  "DHE-DSS-SEED-SHA"
2483        TLS_DHE_RSA_WITH_SEED_CBC_SHA  =  "DHE-RSA-SEED-SHA"
2484        TLS_DH_anon_WITH_SEED_CBC_SHA  =  "ADH-SEED-SHA"
2485
2486     To minimize changes between patchlevels in the OpenSSL 0.9.8
2487     series, SEED remains excluded from compilation unless OpenSSL
2488     is configured with 'enable-seed'.
2489     [KISA, Bodo Moeller]
2490
2491  *) Mitigate branch prediction attacks, which can be practical if a
2492     single processor is shared, allowing a spy process to extract
2493     information.  For detailed background information, see
2494     http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron,
2495     J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
2496     and Necessary Software Countermeasures").  The core of the change
2497     are new versions BN_div_no_branch() and
2498     BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
2499     respectively, which are slower, but avoid the security-relevant
2500     conditional branches.  These are automatically called by BN_div()
2501     and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
2502     of the input BIGNUMs.  Also, BN_is_bit_set() has been changed to
2503     remove a conditional branch.
2504
2505     BN_FLG_CONSTTIME is the new name for the previous
2506     BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
2507     modular exponentiation.  (Since OpenSSL 0.9.7h, setting this flag
2508     in the exponent causes BN_mod_exp_mont() to use the alternative
2509     implementation in BN_mod_exp_mont_consttime().)  The old name
2510     remains as a deprecated alias.
2511
2512     Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
2513     RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
2514     constant-time implementations for more than just exponentiation.
2515     Here too the old name is kept as a deprecated alias.
2516
2517     BN_BLINDING_new() will now use BN_dup() for the modulus so that
2518     the BN_BLINDING structure gets an independent copy of the
2519     modulus.  This means that the previous "BIGNUM *m" argument to
2520     BN_BLINDING_new() and to BN_BLINDING_create_param() now
2521     essentially becomes "const BIGNUM *m", although we can't actually
2522     change this in the header file before 0.9.9.  It allows
2523     RSA_setup_blinding() to use BN_with_flags() on the modulus to
2524     enable BN_FLG_CONSTTIME.
2525
2526     [Matthew D Wood (Intel Corp)]
2527
2528  *) In the SSL/TLS server implementation, be strict about session ID
2529     context matching (which matters if an application uses a single
2530     external cache for different purposes).  Previously,
2531     out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
2532     set.  This did ensure strict client verification, but meant that,
2533     with applications using a single external cache for quite
2534     different requirements, clients could circumvent ciphersuite
2535     restrictions for a given session ID context by starting a session
2536     in a different context.
2537     [Bodo Moeller]
2538
2539  *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
2540     a ciphersuite string such as "DEFAULT:RSA" cannot enable
2541     authentication-only ciphersuites.
2542     [Bodo Moeller]
2543
2544  *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
2545     not complete and could lead to a possible single byte overflow
2546     (CVE-2007-5135) [Ben Laurie]
2547
2548 Changes between 0.9.8d and 0.9.8e  [23 Feb 2007]
2549
2550  *) Since AES128 and AES256 (and similarly Camellia128 and
2551     Camellia256) share a single mask bit in the logic of
2552     ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
2553     kludge to work properly if AES128 is available and AES256 isn't
2554     (or if Camellia128 is available and Camellia256 isn't).
2555     [Victor Duchovni]
2556
2557  *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
2558     (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
2559     When a point or a seed is encoded in a BIT STRING, we need to
2560     prevent the removal of trailing zero bits to get the proper DER
2561     encoding.  (By default, crypto/asn1/a_bitstr.c assumes the case
2562     of a NamedBitList, for which trailing 0 bits need to be removed.)
2563     [Bodo Moeller]
2564
2565  *) Have SSL/TLS server implementation tolerate "mismatched" record
2566     protocol version while receiving ClientHello even if the
2567     ClientHello is fragmented.  (The server can't insist on the
2568     particular protocol version it has chosen before the ServerHello
2569     message has informed the client about his choice.)
2570     [Bodo Moeller]
2571
2572  *) Add RFC 3779 support.
2573     [Rob Austein for ARIN, Ben Laurie]
2574
2575  *) Load error codes if they are not already present instead of using a
2576     static variable. This allows them to be cleanly unloaded and reloaded.
2577     Improve header file function name parsing.
2578     [Steve Henson]
2579
2580  *) extend SMTP and IMAP protocol emulation in s_client to use EHLO
2581     or CAPABILITY handshake as required by RFCs.
2582     [Goetz Babin-Ebell]
2583
2584 Changes between 0.9.8c and 0.9.8d  [28 Sep 2006]
2585
2586  *) Introduce limits to prevent malicious keys being able to
2587     cause a denial of service.  (CVE-2006-2940)
2588     [Steve Henson, Bodo Moeller]
2589
2590  *) Fix ASN.1 parsing of certain invalid structures that can result
2591     in a denial of service.  (CVE-2006-2937)  [Steve Henson]
2592
2593  *) Fix buffer overflow in SSL_get_shared_ciphers() function. 
2594     (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
2595
2596  *) Fix SSL client code which could crash if connecting to a
2597     malicious SSLv2 server.  (CVE-2006-4343)
2598     [Tavis Ormandy and Will Drewry, Google Security Team]
2599
2600  *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
2601     match only those.  Before that, "AES256-SHA" would be interpreted
2602     as a pattern and match "AES128-SHA" too (since AES128-SHA got
2603     the same strength classification in 0.9.7h) as we currently only
2604     have a single AES bit in the ciphersuite description bitmap.
2605     That change, however, also applied to ciphersuite strings such as
2606     "RC4-MD5" that intentionally matched multiple ciphersuites --
2607     namely, SSL 2.0 ciphersuites in addition to the more common ones
2608     from SSL 3.0/TLS 1.0.
2609
2610     So we change the selection algorithm again: Naming an explicit
2611     ciphersuite selects this one ciphersuite, and any other similar
2612     ciphersuite (same bitmap) from *other* protocol versions.
2613     Thus, "RC4-MD5" again will properly select both the SSL 2.0
2614     ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
2615
2616     Since SSL 2.0 does not have any ciphersuites for which the
2617     128/256 bit distinction would be relevant, this works for now.
2618     The proper fix will be to use different bits for AES128 and
2619     AES256, which would have avoided the problems from the beginning;
2620     however, bits are scarce, so we can only do this in a new release
2621     (not just a patchlevel) when we can change the SSL_CIPHER
2622     definition to split the single 'unsigned long mask' bitmap into
2623     multiple values to extend the available space.
2624
2625     [Bodo Moeller]
2626
2627 Changes between 0.9.8b and 0.9.8c  [05 Sep 2006]
2628
2629  *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
2630     (CVE-2006-4339)  [Ben Laurie and Google Security Team]
2631
2632  *) Add AES IGE and biIGE modes.
2633     [Ben Laurie]
2634
2635  *) Change the Unix randomness entropy gathering to use poll() when
2636     possible instead of select(), since the latter has some
2637     undesirable limitations.
2638     [Darryl Miles via Richard Levitte and Bodo Moeller]
2639
2640  *) Disable "ECCdraft" ciphersuites more thoroughly.  Now special
2641     treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
2642     cannot be implicitly activated as part of, e.g., the "AES" alias.
2643     However, please upgrade to OpenSSL 0.9.9[-dev] for
2644     non-experimental use of the ECC ciphersuites to get TLS extension
2645     support, which is required for curve and point format negotiation
2646     to avoid potential handshake problems.
2647     [Bodo Moeller]
2648
2649  *) Disable rogue ciphersuites:
2650
2651      - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
2652      - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
2653      - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
2654
2655     The latter two were purportedly from
2656     draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
2657     appear there.
2658
2659     Also deactivate the remaining ciphersuites from
2660     draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
2661     unofficial, and the ID has long expired.
2662     [Bodo Moeller]
2663
2664  *) Fix RSA blinding Heisenbug (problems sometimes occured on
2665     dual-core machines) and other potential thread-safety issues.
2666     [Bodo Moeller]
2667
2668  *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
2669     versions), which is now available for royalty-free use
2670     (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html).
2671     Also, add Camellia TLS ciphersuites from RFC 4132.
2672
2673     To minimize changes between patchlevels in the OpenSSL 0.9.8
2674     series, Camellia remains excluded from compilation unless OpenSSL
2675     is configured with 'enable-camellia'.
2676     [NTT]
2677
2678  *) Disable the padding bug check when compression is in use. The padding
2679     bug check assumes the first packet is of even length, this is not
2680     necessarily true if compresssion is enabled and can result in false
2681     positives causing handshake failure. The actual bug test is ancient
2682     code so it is hoped that implementations will either have fixed it by
2683     now or any which still have the bug do not support compression.
2684     [Steve Henson]
2685
2686 Changes between 0.9.8a and 0.9.8b  [04 May 2006]
2687
2688  *) When applying a cipher rule check to see if string match is an explicit
2689     cipher suite and only match that one cipher suite if it is.
2690     [Steve Henson]
2691
2692  *) Link in manifests for VC++ if needed.
2693     [Austin Ziegler <halostatue@gmail.com>]
2694
2695  *) Update support for ECC-based TLS ciphersuites according to
2696     draft-ietf-tls-ecc-12.txt with proposed changes (but without
2697     TLS extensions, which are supported starting with the 0.9.9
2698     branch, not in the OpenSSL 0.9.8 branch).
2699     [Douglas Stebila]
2700
2701  *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
2702     opaque EVP_CIPHER_CTX handling.
2703     [Steve Henson]
2704
2705  *) Fixes and enhancements to zlib compression code. We now only use
2706     "zlib1.dll" and use the default __cdecl calling convention on Win32
2707     to conform with the standards mentioned here:
2708           http://www.zlib.net/DLL_FAQ.txt
2709     Static zlib linking now works on Windows and the new --with-zlib-include
2710     --with-zlib-lib options to Configure can be used to supply the location
2711     of the headers and library. Gracefully handle case where zlib library
2712     can't be loaded.
2713     [Steve Henson]
2714
2715  *) Several fixes and enhancements to the OID generation code. The old code
2716     sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
2717     handle numbers larger than ULONG_MAX, truncated printing and had a
2718     non standard OBJ_obj2txt() behaviour.
2719     [Steve Henson]
2720
2721  *) Add support for building of engines under engine/ as shared libraries
2722     under VC++ build system.
2723     [Steve Henson]
2724
2725  *) Corrected the numerous bugs in the Win32 path splitter in DSO.
2726     Hopefully, we will not see any false combination of paths any more.
2727     [Richard Levitte]
2728
2729 Changes between 0.9.8 and 0.9.8a  [11 Oct 2005]
2730
2731  *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
2732     (part of SSL_OP_ALL).  This option used to disable the
2733     countermeasure against man-in-the-middle protocol-version
2734     rollback in the SSL 2.0 server implementation, which is a bad
2735     idea.  (CVE-2005-2969)
2736
2737     [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
2738     for Information Security, National Institute of Advanced Industrial
2739     Science and Technology [AIST], Japan)]
2740
2741  *) Add two function to clear and return the verify parameter flags.
2742     [Steve Henson]
2743
2744  *) Keep cipherlists sorted in the source instead of sorting them at
2745     runtime, thus removing the need for a lock.
2746     [Nils Larsch]
2747
2748  *) Avoid some small subgroup attacks in Diffie-Hellman.
2749     [Nick Mathewson and Ben Laurie]
2750
2751  *) Add functions for well-known primes.
2752     [Nick Mathewson]
2753
2754  *) Extended Windows CE support.
2755     [Satoshi Nakamura and Andy Polyakov]
2756
2757  *) Initialize SSL_METHOD structures at compile time instead of during
2758     runtime, thus removing the need for a lock.
2759     [Steve Henson]
2760
2761  *) Make PKCS7_decrypt() work even if no certificate is supplied by
2762     attempting to decrypt each encrypted key in turn. Add support to
2763     smime utility.
2764     [Steve Henson]
2765
2766 Changes between 0.9.7h and 0.9.8  [05 Jul 2005]
2767
2768  [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
2769  OpenSSL 0.9.8.]
2770
2771  *) Add libcrypto.pc and libssl.pc for those who feel they need them.
2772     [Richard Levitte]
2773
2774  *) Change CA.sh and CA.pl so they don't bundle the CSR and the private
2775     key into the same file any more.
2776     [Richard Levitte]
2777
2778  *) Add initial support for Win64, both IA64 and AMD64/x64 flavors.
2779     [Andy Polyakov]
2780
2781  *) Add -utf8 command line and config file option to 'ca'.
2782     [Stefan <stf@udoma.org]
2783
2784  *) Removed the macro des_crypt(), as it seems to conflict with some
2785     libraries.  Use DES_crypt().
2786     [Richard Levitte]
2787
2788  *) Correct naming of the 'chil' and '4758cca' ENGINEs. This
2789     involves renaming the source and generated shared-libs for
2790     both. The engines will accept the corrected or legacy ids
2791     ('ncipher' and '4758_cca' respectively) when binding. NB,
2792     this only applies when building 'shared'.
2793     [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe]
2794
2795  *) Add attribute functions to EVP_PKEY structure. Modify
2796     PKCS12_create() to recognize a CSP name attribute and
2797     use it. Make -CSP option work again in pkcs12 utility.
2798     [Steve Henson]
2799
2800  *) Add new functionality to the bn blinding code:
2801     - automatic re-creation of the BN_BLINDING parameters after
2802       a fixed number of uses (currently 32)
2803     - add new function for parameter creation
2804     - introduce flags to control the update behaviour of the
2805       BN_BLINDING parameters
2806     - hide BN_BLINDING structure
2807     Add a second BN_BLINDING slot to the RSA structure to improve
2808     performance when a single RSA object is shared among several
2809     threads.
2810     [Nils Larsch]
2811
2812  *) Add support for DTLS.
2813     [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie]
2814
2815  *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
2816     to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
2817     [Walter Goulet]
2818
2819  *) Remove buggy and incompletet DH cert support from
2820     ssl/ssl_rsa.c and ssl/s3_both.c
2821     [Nils Larsch]
2822
2823  *) Use SHA-1 instead of MD5 as the default digest algorithm for
2824     the apps/openssl applications.
2825     [Nils Larsch]
2826
2827  *) Compile clean with "-Wall -Wmissing-prototypes
2828     -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
2829     DEBUG_SAFESTACK must also be set.
2830     [Ben Laurie]
2831
2832  *) Change ./Configure so that certain algorithms can be disabled by default.
2833     The new counterpiece to "no-xxx" is "enable-xxx".
2834
2835     The patented RC5 and MDC2 algorithms will now be disabled unless
2836     "enable-rc5" and "enable-mdc2", respectively, are specified.
2837
2838     (IDEA remains enabled despite being patented.  This is because IDEA
2839     is frequently required for interoperability, and there is no license
2840     fee for non-commercial use.  As before, "no-idea" can be used to
2841     avoid this algorithm.)
2842
2843     [Bodo Moeller]
2844
2845  *) Add processing of proxy certificates (see RFC 3820).  This work was
2846     sponsored by KTH (The Royal Institute of Technology in Stockholm) and
2847     EGEE (Enabling Grids for E-science in Europe).
2848     [Richard Levitte]
2849
2850  *) RC4 performance overhaul on modern architectures/implementations, such
2851     as Intel P4, IA-64 and AMD64.
2852     [Andy Polyakov]
2853
2854  *) New utility extract-section.pl. This can be used specify an alternative
2855     section number in a pod file instead of having to treat each file as
2856     a separate case in Makefile. This can be done by adding two lines to the
2857     pod file:
2858
2859     =for comment openssl_section:XXX
2860
2861     The blank line is mandatory.
2862
2863     [Steve Henson]
2864
2865  *) New arguments -certform, -keyform and -pass for s_client and s_server
2866     to allow alternative format key and certificate files and passphrase
2867     sources.
2868     [Steve Henson]
2869
2870  *) New structure X509_VERIFY_PARAM which combines current verify parameters,
2871     update associated structures and add various utility functions.
2872
2873     Add new policy related verify parameters, include policy checking in 
2874     standard verify code. Enhance 'smime' application with extra parameters
2875     to support policy checking and print out.
2876     [Steve Henson]
2877
2878  *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3
2879     Nehemiah processors. These extensions support AES encryption in hardware
2880     as well as RNG (though RNG support is currently disabled).
2881     [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov]
2882
2883  *) Deprecate BN_[get|set]_params() functions (they were ignored internally).
2884     [Geoff Thorpe]
2885
2886  *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
2887     [Andy Polyakov and a number of other people]
2888
2889  *) Improved PowerPC platform support. Most notably BIGNUM assembler
2890     implementation contributed by IBM.
2891     [Suresh Chari, Peter Waltenberg, Andy Polyakov]
2892
2893  *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
2894     exponent rather than 'unsigned long'. There is a corresponding change to
2895     the new 'rsa_keygen' element of the RSA_METHOD structure.
2896     [Jelte Jansen, Geoff Thorpe]
2897
2898  *) Functionality for creating the initial serial number file is now
2899     moved from CA.pl to the 'ca' utility with a new option -create_serial.
2900
2901     (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
2902     number file to 1, which is bound to cause problems.  To avoid
2903     the problems while respecting compatibility between different 0.9.7
2904     patchlevels, 0.9.7e  employed 'openssl x509 -next_serial' in
2905     CA.pl for serial number initialization.  With the new release 0.9.8,
2906     we can fix the problem directly in the 'ca' utility.)
2907     [Steve Henson]
2908
2909  *) Reduced header interdepencies by declaring more opaque objects in
2910     ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
2911     give fewer recursive includes, which could break lazy source code - so
2912     this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
2913     developers should define this symbol when building and using openssl to
2914     ensure they track the recommended behaviour, interfaces, [etc], but
2915     backwards-compatible behaviour prevails when this isn't defined.
2916     [Geoff Thorpe]
2917
2918  *) New function X509_POLICY_NODE_print() which prints out policy nodes.
2919     [Steve Henson]
2920
2921  *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
2922     This will generate a random key of the appropriate length based on the 
2923     cipher context. The EVP_CIPHER can provide its own random key generation
2924     routine to support keys of a specific form. This is used in the des and 
2925     3des routines to generate a key of the correct parity. Update S/MIME
2926     code to use new functions and hence generate correct parity DES keys.
2927     Add EVP_CHECK_DES_KEY #define to return an error if the key is not 
2928     valid (weak or incorrect parity).
2929     [Steve Henson]
2930
2931  *) Add a local set of CRLs that can be used by X509_verify_cert() as well
2932     as looking them up. This is useful when the verified structure may contain
2933     CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
2934     present unless the new PKCS7_NO_CRL flag is asserted.
2935     [Steve Henson]
2936
2937  *) Extend ASN1 oid configuration module. It now additionally accepts the
2938     syntax:
2939
2940     shortName = some long name, 1.2.3.4
2941     [Steve Henson]
2942
2943  *) Reimplemented the BN_CTX implementation. There is now no more static
2944     limitation on the number of variables it can handle nor the depth of the
2945     "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
2946     information can now expand as required, and rather than having a single
2947     static array of bignums, BN_CTX now uses a linked-list of such arrays
2948     allowing it to expand on demand whilst maintaining the usefulness of
2949     BN_CTX's "bundling".
2950     [Geoff Thorpe]
2951
2952  *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
2953     to allow all RSA operations to function using a single BN_CTX.
2954     [Geoff Thorpe]
2955
2956  *) Preliminary support for certificate policy evaluation and checking. This
2957     is initially intended to pass the tests outlined in "Conformance Testing
2958     of Relying Party Client Certificate Path Processing Logic" v1.07.
2959     [Steve Henson]
2960
2961  *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
2962     remained unused and not that useful. A variety of other little bignum
2963     tweaks and fixes have also been made continuing on from the audit (see
2964     below).
2965     [Geoff Thorpe]
2966
2967  *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
2968     associated ASN1, EVP and SSL functions and old ASN1 macros.
2969     [Richard Levitte]
2970
2971  *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
2972     and this should never fail. So the return value from the use of
2973     BN_set_word() (which can fail due to needless expansion) is now deprecated;
2974     if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
2975     [Geoff Thorpe]
2976
2977  *) BN_CTX_get() should return zero-valued bignums, providing the same
2978     initialised value as BN_new().
2979     [Geoff Thorpe, suggested by Ulf M�ller]
2980
2981  *) Support for inhibitAnyPolicy certificate extension.
2982     [Steve Henson]
2983
2984  *) An audit of the BIGNUM code is underway, for which debugging code is
2985     enabled when BN_DEBUG is defined. This makes stricter enforcements on what
2986     is considered valid when processing BIGNUMs, and causes execution to
2987     assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
2988     further steps are taken to deliberately pollute unused data in BIGNUM
2989     structures to try and expose faulty code further on. For now, openssl will
2990     (in its default mode of operation) continue to tolerate the inconsistent
2991     forms that it has tolerated in the past, but authors and packagers should
2992     consider trying openssl and their own applications when compiled with
2993     these debugging symbols defined. It will help highlight potential bugs in
2994     their own code, and will improve the test coverage for OpenSSL itself. At
2995     some point, these tighter rules will become openssl's default to improve
2996     maintainability, though the assert()s and other overheads will remain only
2997     in debugging configurations. See bn.h for more details.
2998     [Geoff Thorpe, Nils Larsch, Ulf M�ller]
2999
3000  *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
3001     that can only be obtained through BN_CTX_new() (which implicitly
3002     initialises it). The presence of this function only made it possible
3003     to overwrite an existing structure (and cause memory leaks).
3004     [Geoff Thorpe]
3005
3006  *) Because of the callback-based approach for implementing LHASH as a
3007     template type, lh_insert() adds opaque objects to hash-tables and
3008     lh_doall() or lh_doall_arg() are typically used with a destructor callback
3009     to clean up those corresponding objects before destroying the hash table
3010     (and losing the object pointers). So some over-zealous constifications in
3011     LHASH have been relaxed so that lh_insert() does not take (nor store) the
3012     objects as "const" and the lh_doall[_arg] callback wrappers are not
3013     prototyped to have "const" restrictions on the object pointers they are
3014     given (and so aren't required to cast them away any more).
3015     [Geoff Thorpe]
3016
3017  *) The tmdiff.h API was so ugly and minimal that our own timing utility
3018     (speed) prefers to use its own implementation. The two implementations
3019     haven't been consolidated as yet (volunteers?) but the tmdiff API has had
3020     its object type properly exposed (MS_TM) instead of casting to/from "char
3021     *". This may still change yet if someone realises MS_TM and "ms_time_***"
3022     aren't necessarily the greatest nomenclatures - but this is what was used
3023     internally to the implementation so I've used that for now.
3024     [Geoff Thorpe]
3025
3026  *) Ensure that deprecated functions do not get compiled when
3027     OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
3028     the self-tests were still using deprecated key-generation functions so
3029     these have been updated also.
3030     [Geoff Thorpe]
3031
3032  *) Reorganise PKCS#7 code to separate the digest location functionality
3033     into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest().
3034     New function PKCS7_set_digest() to set the digest type for PKCS#7
3035     digestedData type. Add additional code to correctly generate the
3036     digestedData type and add support for this type in PKCS7 initialization
3037     functions.
3038     [Steve Henson]
3039
3040  *) New function PKCS7_set0_type_other() this initializes a PKCS7 
3041     structure of type "other".
3042     [Steve Henson]
3043
3044  *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
3045     sure the loop does correctly stop and breaking ("division by zero")
3046     modulus operations are not performed. The (pre-generated) prime
3047     table crypto/bn/bn_prime.h was already correct, but it could not be
3048     re-generated on some platforms because of the "division by zero"
3049     situation in the script.
3050     [Ralf S. Engelschall]
3051
3052  *) Update support for ECC-based TLS ciphersuites according to
3053     draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
3054     SHA-1 now is only used for "small" curves (where the
3055     representation of a field element takes up to 24 bytes); for
3056     larger curves, the field element resulting from ECDH is directly
3057     used as premaster secret.
3058     [Douglas Stebila (Sun Microsystems Laboratories)]
3059
3060  *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
3061     curve secp160r1 to the tests.
3062     [Douglas Stebila (Sun Microsystems Laboratories)]
3063
3064  *) Add the possibility to load symbols globally with DSO.
3065     [G�tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
3066
3067  *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
3068     control of the error stack.
3069     [Richard Levitte]
3070
3071  *) Add support for STORE in ENGINE.
3072     [Richard Levitte]
3073
3074  *) Add the STORE type.  The intention is to provide a common interface
3075     to certificate and key stores, be they simple file-based stores, or
3076     HSM-type store, or LDAP stores, or...
3077     NOTE: The code is currently UNTESTED and isn't really used anywhere.
3078     [Richard Levitte]
3079
3080  *) Add a generic structure called OPENSSL_ITEM.  This can be used to
3081     pass a list of arguments to any function as well as provide a way
3082     for a function to pass data back to the caller.
3083     [Richard Levitte]
3084
3085  *) Add the functions BUF_strndup() and BUF_memdup().  BUF_strndup()
3086     works like BUF_strdup() but can be used to duplicate a portion of
3087     a string.  The copy gets NUL-terminated.  BUF_memdup() duplicates
3088     a memory area.
3089     [Richard Levitte]
3090
3091  *) Add the function sk_find_ex() which works like sk_find(), but will
3092     return an index to an element even if an exact match couldn't be
3093     found.  The index is guaranteed to point at the element where the
3094     searched-for key would be inserted to preserve sorting order.
3095     [Richard Levitte]
3096
3097  *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
3098     takes an extra flags argument for optional functionality.  Currently,
3099     the following flags are defined:
3100
3101	OBJ_BSEARCH_VALUE_ON_NOMATCH
3102	This one gets OBJ_bsearch_ex() to return a pointer to the first
3103	element where the comparing function returns a negative or zero
3104	number.
3105
3106	OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
3107	This one gets OBJ_bsearch_ex() to return a pointer to the first
3108	element where the comparing function returns zero.  This is useful
3109	if there are more than one element where the comparing function
3110	returns zero.
3111     [Richard Levitte]
3112
3113  *) Make it possible to create self-signed certificates with 'openssl ca'
3114     in such a way that the self-signed certificate becomes part of the
3115     CA database and uses the same mechanisms for serial number generation
3116     as all other certificate signing.  The new flag '-selfsign' enables
3117     this functionality.  Adapt CA.sh and CA.pl.in.
3118     [Richard Levitte]
3119
3120  *) Add functionality to check the public key of a certificate request
3121     against a given private.  This is useful to check that a certificate
3122     request can be signed by that key (self-signing).
3123     [Richard Levitte]
3124
3125  *) Make it possible to have multiple active certificates with the same
3126     subject in the CA index file.  This is done only if the keyword
3127     'unique_subject' is set to 'no' in the main CA section (default
3128     if 'CA_default') of the configuration file.  The value is saved
3129     with the database itself in a separate index attribute file,
3130     named like the index file with '.attr' appended to the name.
3131     [Richard Levitte]
3132
3133  *) Generate muti valued AVAs using '+' notation in config files for
3134     req and dirName.
3135     [Steve Henson]
3136
3137  *) Support for nameConstraints certificate extension.
3138     [Steve Henson]
3139
3140  *) Support for policyConstraints certificate extension.
3141     [Steve Henson]
3142
3143  *) Support for policyMappings certificate extension.
3144     [Steve Henson]
3145
3146  *) Make sure the default DSA_METHOD implementation only uses its
3147     dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
3148     and change its own handlers to be NULL so as to remove unnecessary
3149     indirection. This lets alternative implementations fallback to the
3150     default implementation more easily.
3151     [Geoff Thorpe]
3152
3153  *) Support for directoryName in GeneralName related extensions
3154     in config files.
3155     [Steve Henson]
3156
3157  *) Make it possible to link applications using Makefile.shared.
3158     Make that possible even when linking against static libraries!
3159     [Richard Levitte]
3160
3161  *) Support for single pass processing for S/MIME signing. This now
3162     means that S/MIME signing can be done from a pipe, in addition
3163     cleartext signing (multipart/signed type) is effectively streaming
3164     and the signed data does not need to be all held in memory.
3165
3166     This is done with a new flag PKCS7_STREAM. When this flag is set
3167     PKCS7_sign() only initializes the PKCS7 structure and the actual signing
3168     is done after the data is output (and digests calculated) in
3169     SMIME_write_PKCS7().
3170     [Steve Henson]
3171
3172  *) Add full support for -rpath/-R, both in shared libraries and
3173     applications, at least on the platforms where it's known how
3174     to do it.
3175     [Richard Levitte]
3176
3177  *) In crypto/ec/ec_mult.c, implement fast point multiplication with
3178     precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
3179     will now compute a table of multiples of the generator that
3180     makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
3181     faster (notably in the case of a single point multiplication,
3182     scalar * generator).
3183     [Nils Larsch, Bodo Moeller]
3184
3185  *) IPv6 support for certificate extensions. The various extensions
3186     which use the IP:a.b.c.d can now take IPv6 addresses using the
3187     formats of RFC1884 2.2 . IPv6 addresses are now also displayed
3188     correctly.
3189     [Steve Henson]
3190
3191  *) Added an ENGINE that implements RSA by performing private key
3192     exponentiations with the GMP library. The conversions to and from
3193     GMP's mpz_t format aren't optimised nor are any montgomery forms
3194     cached, and on x86 it appears OpenSSL's own performance has caught up.
3195     However there are likely to be other architectures where GMP could
3196     provide a boost. This ENGINE is not built in by default, but it can be
3197     specified at Configure time and should be accompanied by the necessary
3198     linker additions, eg;
3199         ./config -DOPENSSL_USE_GMP -lgmp
3200     [Geoff Thorpe]
3201
3202  *) "openssl engine" will not display ENGINE/DSO load failure errors when
3203     testing availability of engines with "-t" - the old behaviour is
3204     produced by increasing the feature's verbosity with "-tt".
3205     [Geoff Thorpe]
3206
3207  *) ECDSA routines: under certain error conditions uninitialized BN objects
3208     could be freed. Solution: make sure initialization is performed early
3209     enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
3210     via PR#459)
3211     [Lutz Jaenicke]
3212
3213  *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
3214     and DH_METHOD (eg. by ENGINE implementations) to override the normal
3215     software implementations. For DSA and DH, parameter generation can
3216     also be overriden by providing the appropriate method callbacks.
3217     [Geoff Thorpe]
3218
3219  *) Change the "progress" mechanism used in key-generation and
3220     primality testing to functions that take a new BN_GENCB pointer in
3221     place of callback/argument pairs. The new API functions have "_ex"
3222     postfixes and the older functions are reimplemented as wrappers for
3223     the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
3224     declarations of the old functions to help (graceful) attempts to
3225     migrate to the new functions. Also, the new key-generation API
3226     functions operate on a caller-supplied key-structure and return
3227     success/failure rather than returning a key or NULL - this is to
3228     help make "keygen" another member function of RSA_METHOD etc.
3229
3230     Example for using the new callback interface:
3231
3232          int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
3233          void *my_arg = ...;
3234          BN_GENCB my_cb;
3235
3236          BN_GENCB_set(&my_cb, my_callback, my_arg);
3237
3238          return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
3239          /* For the meaning of a, b in calls to my_callback(), see the
3240           * documentation of the function that calls the callback.
3241           * cb will point to my_cb; my_arg can be retrieved as cb->arg.
3242           * my_callback should return 1 if it wants BN_is_prime_ex()
3243           * to continue, or 0 to stop.
3244           */
3245
3246     [Geoff Thorpe]
3247
3248  *) Change the ZLIB compression method to be stateful, and make it
3249     available to TLS with the number defined in 
3250     draft-ietf-tls-compression-04.txt.
3251     [Richard Levitte]
3252
3253  *) Add the ASN.1 structures and functions for CertificatePair, which
3254     is defined as follows (according to X.509_4thEditionDraftV6.pdf):
3255
3256     CertificatePair ::= SEQUENCE {
3257        forward		[0]	Certificate OPTIONAL,
3258        reverse		[1]	Certificate OPTIONAL,
3259        -- at least one of the pair shall be present -- }
3260
3261     Also implement the PEM functions to read and write certificate
3262     pairs, and defined the PEM tag as "CERTIFICATE PAIR".
3263
3264     This needed to be defined, mostly for the sake of the LDAP
3265     attribute crossCertificatePair, but may prove useful elsewhere as
3266     well.
3267     [Richard Levitte]
3268
3269  *) Make it possible to inhibit symlinking of shared libraries in
3270     Makefile.shared, for Cygwin's sake.
3271     [Richard Levitte]
3272
3273  *) Extend the BIGNUM API by creating a function 
3274          void BN_set_negative(BIGNUM *a, int neg);
3275     and a macro that behave like
3276          int  BN_is_negative(const BIGNUM *a);
3277
3278     to avoid the need to access 'a->neg' directly in applications.
3279     [Nils Larsch]
3280
3281  *) Implement fast modular reduction for pseudo-Mersenne primes
3282     used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
3283     EC_GROUP_new_curve_GFp() will now automatically use this
3284     if applicable.
3285     [Nils Larsch <nla@trustcenter.de>]
3286
3287  *) Add new lock type (CRYPTO_LOCK_BN).
3288     [Bodo Moeller]
3289
3290  *) Change the ENGINE framework to automatically load engines
3291     dynamically from specific directories unless they could be
3292     found to already be built in or loaded.  Move all the
3293     current engines except for the cryptodev one to a new
3294     directory engines/.
3295     The engines in engines/ are built as shared libraries if
3296     the "shared" options was given to ./Configure or ./config.
3297     Otherwise, they are inserted in libcrypto.a.
3298     /usr/local/ssl/engines is the default directory for dynamic
3299     engines, but that can be overriden at configure time through
3300     the usual use of --prefix and/or --openssldir, and at run
3301     time with the environment variable OPENSSL_ENGINES.
3302     [Geoff Thorpe and Richard Levitte]
3303
3304  *) Add Makefile.shared, a helper makefile to build shared
3305     libraries.  Addapt Makefile.org.
3306     [Richard Levitte]
3307
3308  *) Add version info to Win32 DLLs.
3309     [Peter 'Luna' Runestig" <peter@runestig.com>]
3310
3311  *) Add new 'medium level' PKCS#12 API. Certificates and keys
3312     can be added using this API to created arbitrary PKCS#12
3313     files while avoiding the low level API.
3314
3315     New options to PKCS12_create(), key or cert can be NULL and
3316     will then be omitted from the output file. The encryption
3317     algorithm NIDs can be set to -1 for no encryption, the mac
3318     iteration count can be set to 0 to omit the mac.
3319
3320     Enhance pkcs12 utility by making the -nokeys and -nocerts
3321     options work when creating a PKCS#12 file. New option -nomac
3322     to omit the mac, NONE can be set for an encryption algorithm.
3323     New code is modified to use the enhanced PKCS12_create()
3324     instead of the low level API.
3325     [Steve Henson]
3326
3327  *) Extend ASN1 encoder to support indefinite length constructed
3328     encoding. This can output sequences tags and octet strings in
3329     this form. Modify pk7_asn1.c to support indefinite length
3330     encoding. This is experimental and needs additional code to
3331     be useful, such as an ASN1 bio and some enhanced streaming
3332     PKCS#7 code.
3333
3334     Extend template encode functionality so that tagging is passed
3335     down to the template encoder.
3336     [Steve Henson]
3337
3338  *) Let 'openssl req' fail if an argument to '-newkey' is not
3339     recognized instead of using RSA as a default.
3340     [Bodo Moeller]
3341
3342  *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
3343     As these are not official, they are not included in "ALL";
3344     the "ECCdraft" ciphersuite group alias can be used to select them.
3345     [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
3346
3347  *) Add ECDH engine support.
3348     [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
3349
3350  *) Add ECDH in new directory crypto/ecdh/.
3351     [Douglas Stebila (Sun Microsystems Laboratories)]
3352
3353  *) Let BN_rand_range() abort with an error after 100 iterations
3354     without success (which indicates a broken PRNG).
3355     [Bodo Moeller]
3356
3357  *) Change BN_mod_sqrt() so that it verifies that the input value
3358     is really the square of the return value.  (Previously,
3359     BN_mod_sqrt would show GIGO behaviour.)
3360     [Bodo Moeller]
3361
3362  *) Add named elliptic curves over binary fields from X9.62, SECG,
3363     and WAP/WTLS; add OIDs that were still missing.
3364
3365     [Sheueling Chang Shantz and Douglas Stebila
3366     (Sun Microsystems Laboratories)]
3367
3368  *) Extend the EC library for elliptic curves over binary fields
3369     (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
3370     New EC_METHOD:
3371
3372          EC_GF2m_simple_method
3373
3374     New API functions:
3375
3376          EC_GROUP_new_curve_GF2m
3377          EC_GROUP_set_curve_GF2m
3378          EC_GROUP_get_curve_GF2m
3379          EC_POINT_set_affine_coordinates_GF2m
3380          EC_POINT_get_affine_coordinates_GF2m
3381          EC_POINT_set_compressed_coordinates_GF2m
3382
3383     Point compression for binary fields is disabled by default for
3384     patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
3385     enable it).
3386
3387     As binary polynomials are represented as BIGNUMs, various members
3388     of the EC_GROUP and EC_POINT data structures can be shared
3389     between the implementations for prime fields and binary fields;
3390     the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
3391     are essentially identical to their ..._GFp counterparts.
3392     (For simplicity, the '..._GFp' prefix has been dropped from
3393     various internal method names.)
3394
3395     An internal 'field_div' method (similar to 'field_mul' and
3396     'field_sqr') has been added; this is used only for binary fields.
3397
3398     [Sheueling Chang Shantz and Douglas Stebila
3399     (Sun Microsystems Laboratories)]
3400
3401  *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
3402     through methods ('mul', 'precompute_mult').
3403
3404     The generic implementations (now internally called 'ec_wNAF_mul'
3405     and 'ec_wNAF_precomputed_mult') remain the default if these
3406     methods are undefined.
3407
3408     [Sheueling Chang Shantz and Douglas Stebila
3409     (Sun Microsystems Laboratories)]
3410
3411  *) New function EC_GROUP_get_degree, which is defined through
3412     EC_METHOD.  For curves over prime fields, this returns the bit
3413     length of the modulus.
3414
3415     [Sheueling Chang Shantz and Douglas Stebila
3416     (Sun Microsystems Laboratories)]
3417
3418  *) New functions EC_GROUP_dup, EC_POINT_dup.
3419     (These simply call ..._new  and ..._copy).
3420
3421     [Sheueling Chang Shantz and Douglas Stebila
3422     (Sun Microsystems Laboratories)]
3423
3424  *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
3425     Polynomials are represented as BIGNUMs (where the sign bit is not
3426     used) in the following functions [macros]:  
3427
3428          BN_GF2m_add
3429          BN_GF2m_sub             [= BN_GF2m_add]
3430          BN_GF2m_mod             [wrapper for BN_GF2m_mod_arr]
3431          BN_GF2m_mod_mul         [wrapper for BN_GF2m_mod_mul_arr]
3432          BN_GF2m_mod_sqr         [wrapper for BN_GF2m_mod_sqr_arr]
3433          BN_GF2m_mod_inv
3434          BN_GF2m_mod_exp         [wrapper for BN_GF2m_mod_exp_arr]
3435          BN_GF2m_mod_sqrt        [wrapper for BN_GF2m_mod_sqrt_arr]
3436          BN_GF2m_mod_solve_quad  [wrapper for BN_GF2m_mod_solve_quad_arr]
3437          BN_GF2m_cmp             [= BN_ucmp]
3438
3439     (Note that only the 'mod' functions are actually for fields GF(2^m).
3440     BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
3441
3442     For some functions, an the irreducible polynomial defining a
3443     field can be given as an 'unsigned int[]' with strictly
3444     decreasing elements giving the indices of those bits that are set;
3445     i.e., p[] represents the polynomial
3446          f(t) = t^p[0] + t^p[1] + ... + t^p[k]
3447     where
3448          p[0] > p[1] > ... > p[k] = 0.
3449     This applies to the following functions:
3450
3451          BN_GF2m_mod_arr
3452          BN_GF2m_mod_mul_arr
3453          BN_GF2m_mod_sqr_arr
3454          BN_GF2m_mod_inv_arr        [wrapper for BN_GF2m_mod_inv]
3455          BN_GF2m_mod_div_arr        [wrapper for BN_GF2m_mod_div]
3456          BN_GF2m_mod_exp_arr
3457          BN_GF2m_mod_sqrt_arr
3458          BN_GF2m_mod_solve_quad_arr
3459          BN_GF2m_poly2arr
3460          BN_GF2m_arr2poly
3461
3462     Conversion can be performed by the following functions:
3463
3464          BN_GF2m_poly2arr
3465          BN_GF2m_arr2poly
3466
3467     bntest.c has additional tests for binary polynomial arithmetic.
3468
3469     Two implementations for BN_GF2m_mod_div() are available.
3470     The default algorithm simply uses BN_GF2m_mod_inv() and
3471     BN_GF2m_mod_mul().  The alternative algorithm is compiled in only
3472     if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
3473     copyright notice in crypto/bn/bn_gf2m.c before enabling it).
3474
3475     [Sheueling Chang Shantz and Douglas Stebila
3476     (Sun Microsystems Laboratories)]
3477
3478  *) Add new error code 'ERR_R_DISABLED' that can be used when some
3479     functionality is disabled at compile-time.
3480     [Douglas Stebila <douglas.stebila@sun.com>]
3481
3482  *) Change default behaviour of 'openssl asn1parse' so that more
3483     information is visible when viewing, e.g., a certificate:
3484
3485     Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
3486     mode the content of non-printable OCTET STRINGs is output in a
3487     style similar to INTEGERs, but with '[HEX DUMP]' prepended to
3488     avoid the appearance of a printable string.
3489     [Nils Larsch <nla@trustcenter.de>]
3490
3491  *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
3492     functions
3493          EC_GROUP_set_asn1_flag()
3494          EC_GROUP_get_asn1_flag()
3495          EC_GROUP_set_point_conversion_form()
3496          EC_GROUP_get_point_conversion_form()
3497     These control ASN1 encoding details:
3498     - Curves (i.e., groups) are encoded explicitly unless asn1_flag
3499       has been set to OPENSSL_EC_NAMED_CURVE.
3500     - Points are encoded in uncompressed form by default; options for
3501       asn1_for are as for point2oct, namely
3502          POINT_CONVERSION_COMPRESSED
3503          POINT_CONVERSION_UNCOMPRESSED
3504          POINT_CONVERSION_HYBRID
3505
3506     Also add 'seed' and 'seed_len' members to EC_GROUP with access
3507     functions
3508          EC_GROUP_set_seed()
3509          EC_GROUP_get0_seed()
3510          EC_GROUP_get_seed_len()
3511     This is used only for ASN1 purposes (so far).
3512     [Nils Larsch <nla@trustcenter.de>]
3513
3514  *) Add 'field_type' member to EC_METHOD, which holds the NID
3515     of the appropriate field type OID.  The new function
3516     EC_METHOD_get_field_type() returns this value.
3517     [Nils Larsch <nla@trustcenter.de>]
3518
3519  *) Add functions 
3520          EC_POINT_point2bn()
3521          EC_POINT_bn2point()
3522          EC_POINT_point2hex()
3523          EC_POINT_hex2point()
3524     providing useful interfaces to EC_POINT_point2oct() and
3525     EC_POINT_oct2point().
3526     [Nils Larsch <nla@trustcenter.de>]
3527
3528  *) Change internals of the EC library so that the functions
3529          EC_GROUP_set_generator()
3530          EC_GROUP_get_generator()
3531          EC_GROUP_get_order()
3532          EC_GROUP_get_cofactor()
3533     are implemented directly in crypto/ec/ec_lib.c and not dispatched
3534     to methods, which would lead to unnecessary code duplication when
3535     adding different types of curves.
3536     [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
3537
3538  *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
3539     arithmetic, and such that modified wNAFs are generated
3540     (which avoid length expansion in many cases).
3541     [Bodo Moeller]
3542
3543  *) Add a function EC_GROUP_check_discriminant() (defined via
3544     EC_METHOD) that verifies that the curve discriminant is non-zero.
3545
3546     Add a function EC_GROUP_check() that makes some sanity tests
3547     on a EC_GROUP, its generator and order.  This includes
3548     EC_GROUP_check_discriminant().
3549     [Nils Larsch <nla@trustcenter.de>]
3550
3551  *) Add ECDSA in new directory crypto/ecdsa/.
3552
3553     Add applications 'openssl ecparam' and 'openssl ecdsa'
3554     (these are based on 'openssl dsaparam' and 'openssl dsa').
3555
3556     ECDSA support is also included in various other files across the
3557     library.  Most notably,
3558     - 'openssl req' now has a '-newkey ecdsa:file' option;
3559     - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
3560     - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
3561       d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
3562       them suitable for ECDSA where domain parameters must be
3563       extracted before the specific public key;
3564     - ECDSA engine support has been added.
3565     [Nils Larsch <nla@trustcenter.de>]
3566
3567  *) Include some named elliptic curves, and add OIDs from X9.62,
3568     SECG, and WAP/WTLS.  Each curve can be obtained from the new
3569     function
3570          EC_GROUP_new_by_curve_name(),
3571     and the list of available named curves can be obtained with
3572          EC_get_builtin_curves().
3573     Also add a 'curve_name' member to EC_GROUP objects, which can be
3574     accessed via
3575         EC_GROUP_set_curve_name()
3576         EC_GROUP_get_curve_name()
3577     [Nils Larsch <larsch@trustcenter.de, Bodo Moeller]
3578 
3579  *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
3580     was actually never needed) and in BN_mul().  The removal in BN_mul()
3581     required a small change in bn_mul_part_recursive() and the addition
3582     of the functions bn_cmp_part_words(), bn_sub_part_words() and
3583     bn_add_part_words(), which do the same thing as bn_cmp_words(),
3584     bn_sub_words() and bn_add_words() except they take arrays with
3585     differing sizes.
3586     [Richard Levitte]
3587
3588 Changes between 0.9.7l and 0.9.7m  [23 Feb 2007]
3589
3590  *) Cleanse PEM buffers before freeing them since they may contain 
3591     sensitive data.
3592     [Benjamin Bennett <ben@psc.edu>]
3593
3594  *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
3595     a ciphersuite string such as "DEFAULT:RSA" cannot enable
3596     authentication-only ciphersuites.
3597     [Bodo Moeller]
3598
3599  *) Since AES128 and AES256 share a single mask bit in the logic of
3600     ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
3601     kludge to work properly if AES128 is available and AES256 isn't.
3602     [Victor Duchovni]
3603
3604  *) Expand security boundary to match 1.1.1 module.
3605     [Steve Henson]
3606
3607  *) Remove redundant features: hash file source, editing of test vectors
3608     modify fipsld to use external fips_premain.c signature.
3609     [Steve Henson]
3610
3611  *) New perl script mkfipsscr.pl to create shell scripts or batch files to
3612     run algorithm test programs.
3613     [Steve Henson]
3614
3615  *) Make algorithm test programs more tolerant of whitespace.
3616     [Steve Henson]
3617
3618  *) Have SSL/TLS server implementation tolerate "mismatched" record
3619     protocol version while receiving ClientHello even if the
3620     ClientHello is fragmented.  (The server can't insist on the
3621     particular protocol version it has chosen before the ServerHello
3622     message has informed the client about his choice.)
3623     [Bodo Moeller]
3624
3625  *) Load error codes if they are not already present instead of using a
3626     static variable. This allows them to be cleanly unloaded and reloaded.
3627     [Steve Henson]
3628
3629 Changes between 0.9.7k and 0.9.7l  [28 Sep 2006]
3630
3631  *) Introduce limits to prevent malicious keys being able to
3632     cause a denial of service.  (CVE-2006-2940)
3633     [Steve Henson, Bodo Moeller]
3634
3635  *) Fix ASN.1 parsing of certain invalid structures that can result
3636     in a denial of service.  (CVE-2006-2937)  [Steve Henson]
3637
3638  *) Fix buffer overflow in SSL_get_shared_ciphers() function. 
3639     (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
3640
3641  *) Fix SSL client code which could crash if connecting to a
3642     malicious SSLv2 server.  (CVE-2006-4343)
3643     [Tavis Ormandy and Will Drewry, Google Security Team]
3644
3645  *) Change ciphersuite string processing so that an explicit
3646     ciphersuite selects this one ciphersuite (so that "AES256-SHA"
3647     will no longer include "AES128-SHA"), and any other similar
3648     ciphersuite (same bitmap) from *other* protocol versions (so that
3649     "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
3650     SSL 3.0/TLS 1.0 ciphersuite).  This is a backport combining
3651     changes from 0.9.8b and 0.9.8d.
3652     [Bodo Moeller]
3653
3654 Changes between 0.9.7j and 0.9.7k  [05 Sep 2006]
3655
3656  *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
3657     (CVE-2006-4339)  [Ben Laurie and Google Security Team]
3658
3659  *) Change the Unix randomness entropy gathering to use poll() when
3660     possible instead of select(), since the latter has some
3661     undesirable limitations.
3662     [Darryl Miles via Richard Levitte and Bodo Moeller]
3663
3664  *) Disable rogue ciphersuites:
3665
3666      - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
3667      - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
3668      - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
3669
3670     The latter two were purportedly from
3671     draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
3672     appear there.
3673
3674     Also deactive the remaining ciphersuites from
3675     draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
3676     unofficial, and the ID has long expired.
3677     [Bodo Moeller]
3678
3679  *) Fix RSA blinding Heisenbug (problems sometimes occured on
3680     dual-core machines) and other potential thread-safety issues.
3681     [Bodo Moeller]
3682
3683 Changes between 0.9.7i and 0.9.7j  [04 May 2006]
3684
3685  *) Adapt fipsld and the build system to link against the validated FIPS
3686     module in FIPS mode.
3687     [Steve Henson]
3688
3689  *) Fixes for VC++ 2005 build under Windows.
3690     [Steve Henson]
3691
3692  *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 
3693     from a Windows bash shell such as MSYS. It is autodetected from the
3694     "config" script when run from a VC++ environment. Modify standard VC++
3695     build to use fipscanister.o from the GNU make build. 
3696     [Steve Henson]
3697
3698 Changes between 0.9.7h and 0.9.7i  [14 Oct 2005]
3699
3700  *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
3701     The value now differs depending on if you build for FIPS or not.
3702     BEWARE!  A program linked with a shared FIPSed libcrypto can't be
3703     safely run with a non-FIPSed libcrypto, as it may crash because of
3704     the difference induced by this change.
3705     [Andy Polyakov]
3706
3707 Changes between 0.9.7g and 0.9.7h  [11 Oct 2005]
3708
3709  *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
3710     (part of SSL_OP_ALL).  This option used to disable the
3711     countermeasure against man-in-the-middle protocol-version
3712     rollback in the SSL 2.0 server implementation, which is a bad
3713     idea.  (CVE-2005-2969)
3714
3715     [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
3716     for Information Security, National Institute of Advanced Industrial
3717     Science and Technology [AIST], Japan)]
3718
3719  *) Minimal support for X9.31 signatures and PSS padding modes. This is
3720     mainly for FIPS compliance and not fully integrated at this stage.
3721     [Steve Henson]
3722
3723  *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
3724     the exponentiation using a fixed-length exponent.  (Otherwise,
3725     the information leaked through timing could expose the secret key
3726     after many signatures; cf. Bleichenbacher's attack on DSA with
3727     biased k.)
3728     [Bodo Moeller]
3729
3730  *) Make a new fixed-window mod_exp implementation the default for
3731     RSA, DSA, and DH private-key operations so that the sequence of
3732     squares and multiplies and the memory access pattern are
3733     independent of the particular secret key.  This will mitigate
3734     cache-timing and potential related attacks.
3735
3736     BN_mod_exp_mont_consttime() is the new exponentiation implementation,
3737     and this is automatically used by BN_mod_exp_mont() if the new flag
3738     BN_FLG_EXP_CONSTTIME is set for the exponent.  RSA, DSA, and DH
3739     will use this BN flag for private exponents unless the flag
3740     RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
3741     DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
3742
3743     [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller]
3744
3745  *) Change the client implementation for SSLv23_method() and
3746     SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
3747     Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
3748     (Previously, the SSL 2.0 backwards compatible Client Hello
3749     message format would be used even with SSL_OP_NO_SSLv2.)
3750     [Bodo Moeller]
3751
3752  *) Add support for smime-type MIME parameter in S/MIME messages which some
3753     clients need.
3754     [Steve Henson]
3755
3756  *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in
3757     a threadsafe manner. Modify rsa code to use new function and add calls
3758     to dsa and dh code (which had race conditions before).
3759     [Steve Henson]
3760
3761  *) Include the fixed error library code in the C error file definitions
3762     instead of fixing them up at runtime. This keeps the error code
3763     structures constant.
3764     [Steve Henson]
3765
3766 Changes between 0.9.7f and 0.9.7g  [11 Apr 2005]
3767
3768  [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
3769  OpenSSL 0.9.8.]
3770
3771  *) Fixes for newer kerberos headers. NB: the casts are needed because
3772     the 'length' field is signed on one version and unsigned on another
3773     with no (?) obvious way to tell the difference, without these VC++
3774     complains. Also the "definition" of FAR (blank) is no longer included
3775     nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
3776     some needed definitions.
3777     [Steve Henson]
3778
3779  *) Undo Cygwin change.
3780     [Ulf M�ller]
3781
3782  *) Added support for proxy certificates according to RFC 3820.
3783     Because they may be a security thread to unaware applications,
3784     they must be explicitely allowed in run-time.  See
3785     docs/HOWTO/proxy_certificates.txt for further information.
3786     [Richard Levitte]
3787
3788 Changes between 0.9.7e and 0.9.7f  [22 Mar 2005]
3789
3790  *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
3791     server and client random values. Previously
3792     (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
3793     less random data when sizeof(time_t) > 4 (some 64 bit platforms).
3794
3795     This change has negligible security impact because:
3796
3797     1. Server and client random values still have 24 bytes of pseudo random
3798        data.
3799
3800     2. Server and client random values are sent in the clear in the initial
3801        handshake.
3802
3803     3. The master secret is derived using the premaster secret (48 bytes in
3804        size for static RSA ciphersuites) as well as client server and random
3805        values.
3806
3807     The OpenSSL team would like to thank the UK NISCC for bringing this issue
3808     to our attention. 
3809
3810     [Stephen Henson, reported by UK NISCC]
3811
3812  *) Use Windows randomness collection on Cygwin.
3813     [Ulf M�ller]
3814
3815  *) Fix hang in EGD/PRNGD query when communication socket is closed
3816     prematurely by EGD/PRNGD.
3817     [Darren Tucker <dtucker@zip.com.au> via Lutz J�nicke, resolves #1014]
3818
3819  *) Prompt for pass phrases when appropriate for PKCS12 input format.
3820     [Steve Henson]
3821
3822  *) Back-port of selected performance improvements from development
3823     branch, as well as improved support for PowerPC platforms.
3824     [Andy Polyakov]
3825
3826  *) Add lots of checks for memory allocation failure, error codes to indicate
3827     failure and freeing up memory if a failure occurs.
3828     [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson]
3829
3830  *) Add new -passin argument to dgst.
3831     [Steve Henson]
3832
3833  *) Perform some character comparisons of different types in X509_NAME_cmp:
3834     this is needed for some certificates that reencode DNs into UTF8Strings
3835     (in violation of RFC3280) and can't or wont issue name rollover
3836     certificates.
3837     [Steve Henson]
3838
3839  *) Make an explicit check during certificate validation to see that
3840     the CA setting in each certificate on the chain is correct.  As a
3841     side effect always do the following basic checks on extensions,
3842     not just when there's an associated purpose to the check:
3843
3844      - if there is an unhandled critical extension (unless the user
3845        has chosen to ignore this fault)
3846      - if the path length has been exceeded (if one is set at all)
3847      - that certain extensions fit the associated purpose (if one has
3848        been given)
3849     [Richard Levitte]
3850
3851 Changes between 0.9.7d and 0.9.7e  [25 Oct 2004]
3852
3853  *) Avoid a race condition when CRLs are checked in a multi threaded 
3854     environment. This would happen due to the reordering of the revoked
3855     entries during signature checking and serial number lookup. Now the
3856     encoding is cached and the serial number sort performed under a lock.
3857     Add new STACK function sk_is_sorted().
3858     [Steve Henson]
3859
3860  *) Add Delta CRL to the extension code.
3861     [Steve Henson]
3862
3863  *) Various fixes to s3_pkt.c so alerts are sent properly.
3864     [David Holmes <d.holmes@f5.com>]
3865
3866  *) Reduce the chances of duplicate issuer name and serial numbers (in
3867     violation of RFC3280) using the OpenSSL certificate creation utilities.
3868     This is done by creating a random 64 bit value for the initial serial
3869     number when a serial number file is created or when a self signed
3870     certificate is created using 'openssl req -x509'. The initial serial
3871     number file is created using 'openssl x509 -next_serial' in CA.pl
3872     rather than being initialized to 1.
3873     [Steve Henson]
3874
3875 Changes between 0.9.7c and 0.9.7d  [17 Mar 2004]
3876
3877  *) Fix null-pointer assignment in do_change_cipher_spec() revealed           
3878     by using the Codenomicon TLS Test Tool (CVE-2004-0079)                    
3879     [Joe Orton, Steve Henson]   
3880
3881  *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
3882     (CVE-2004-0112)
3883     [Joe Orton, Steve Henson]   
3884
3885  *) Make it possible to have multiple active certificates with the same
3886     subject in the CA index file.  This is done only if the keyword
3887     'unique_subject' is set to 'no' in the main CA section (default
3888     if 'CA_default') of the configuration file.  The value is saved
3889     with the database itself in a separate index attribute file,
3890     named like the index file with '.attr' appended to the name.
3891     [Richard Levitte]
3892
3893  *) X509 verify fixes. Disable broken certificate workarounds when 
3894     X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
3895     keyUsage extension present. Don't accept CRLs with unhandled critical
3896     extensions: since verify currently doesn't process CRL extensions this
3897     rejects a CRL with *any* critical extensions. Add new verify error codes
3898     for these cases.
3899     [Steve Henson]
3900
3901  *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
3902     A clarification of RFC2560 will require the use of OCTET STRINGs and 
3903     some implementations cannot handle the current raw format. Since OpenSSL
3904     copies and compares OCSP nonces as opaque blobs without any attempt at
3905     parsing them this should not create any compatibility issues.
3906     [Steve Henson]
3907
3908  *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
3909     calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
3910     this HMAC (and other) operations are several times slower than OpenSSL
3911     < 0.9.7.
3912     [Steve Henson]
3913
3914  *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
3915     [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
3916
3917  *) Use the correct content when signing type "other".
3918     [Steve Henson]
3919
3920 Changes between 0.9.7b and 0.9.7c  [30 Sep 2003]
3921
3922  *) Fix various bugs revealed by running the NISCC test suite:
3923
3924     Stop out of bounds reads in the ASN1 code when presented with
3925     invalid tags (CVE-2003-0543 and CVE-2003-0544).
3926     
3927     Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545).
3928
3929     If verify callback ignores invalid public key errors don't try to check
3930     certificate signature with the NULL public key.
3931
3932     [Steve Henson]
3933
3934  *) New -ignore_err option in ocsp application to stop the server
3935     exiting on the first error in a request.
3936     [Steve Henson]
3937
3938  *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
3939     if the server requested one: as stated in TLS 1.0 and SSL 3.0
3940     specifications.
3941     [Steve Henson]
3942
3943  *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
3944     extra data after the compression methods not only for TLS 1.0
3945     but also for SSL 3.0 (as required by the specification).
3946     [Bodo Moeller; problem pointed out by Matthias Loepfe]
3947
3948  *) Change X509_certificate_type() to mark the key as exported/exportable
3949     when it's 512 *bits* long, not 512 bytes.
3950     [Richard Levitte]
3951
3952  *) Change AES_cbc_encrypt() so it outputs exact multiple of
3953     blocks during encryption.
3954     [Richard Levitte]
3955
3956  *) Various fixes to base64 BIO and non blocking I/O. On write 
3957     flushes were not handled properly if the BIO retried. On read
3958     data was not being buffered properly and had various logic bugs.
3959     This also affects blocking I/O when the data being decoded is a
3960     certain size.
3961     [Steve Henson]
3962
3963  *) Various S/MIME bugfixes and compatibility changes:
3964     output correct application/pkcs7 MIME type if
3965     PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
3966     Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
3967     of files as .eml work). Correctly handle very long lines in MIME
3968     parser.
3969     [Steve Henson]
3970
3971 Changes between 0.9.7a and 0.9.7b  [10 Apr 2003]
3972
3973  *) Countermeasure against the Klima-Pokorny-Rosa extension of
3974     Bleichbacher's attack on PKCS #1 v1.5 padding: treat
3975     a protocol version number mismatch like a decryption error
3976     in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
3977     [Bodo Moeller]
3978
3979  *) Turn on RSA blinding by default in the default implementation
3980     to avoid a timing attack. Applications that don't want it can call
3981     RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
3982     They would be ill-advised to do so in most cases.
3983     [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
3984
3985  *) Change RSA blinding code so that it works when the PRNG is not
3986     seeded (in this case, the secret RSA exponent is abused as
3987     an unpredictable seed -- if it is not unpredictable, there
3988     is no point in blinding anyway).  Make RSA blinding thread-safe
3989     by remembering the creator's thread ID in rsa->blinding and
3990     having all other threads use local one-time blinding factors
3991     (this requires more computation than sharing rsa->blinding, but
3992     avoids excessive locking; and if an RSA object is not shared
3993     between threads, blinding will still be very fast).
3994     [Bodo Moeller]
3995
3996  *) Fixed a typo bug that would cause ENGINE_set_default() to set an
3997     ENGINE as defaults for all supported algorithms irrespective of
3998     the 'flags' parameter. 'flags' is now honoured, so applications
3999     should make sure they are passing it correctly.
4000     [Geoff Thorpe]
4001
4002  *) Target "mingw" now allows native Windows code to be generated in
4003     the Cygwin environment as well as with the MinGW compiler.
4004     [Ulf Moeller] 
4005
4006 Changes between 0.9.7 and 0.9.7a  [19 Feb 2003]
4007
4008  *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
4009     via timing by performing a MAC computation even if incorrrect
4010     block cipher padding has been found.  This is a countermeasure
4011     against active attacks where the attacker has to distinguish
4012     between bad padding and a MAC verification error. (CVE-2003-0078)
4013
4014     [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
4015     Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
4016     Martin Vuagnoux (EPFL, Ilion)]
4017
4018  *) Make the no-err option work as intended.  The intention with no-err
4019     is not to have the whole error stack handling routines removed from
4020     libcrypto, it's only intended to remove all the function name and
4021     reason texts, thereby removing some of the footprint that may not
4022     be interesting if those errors aren't displayed anyway.
4023
4024     NOTE: it's still possible for any application or module to have it's
4025     own set of error texts inserted.  The routines are there, just not
4026     used by default when no-err is given.
4027     [Richard Levitte]
4028
4029  *) Add support for FreeBSD on IA64.
4030     [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
4031
4032  *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
4033     Kerberos function mit_des_cbc_cksum().  Before this change,
4034     the value returned by DES_cbc_cksum() was like the one from
4035     mit_des_cbc_cksum(), except the bytes were swapped.
4036     [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
4037
4038  *) Allow an application to disable the automatic SSL chain building.
4039     Before this a rather primitive chain build was always performed in
4040     ssl3_output_cert_chain(): an application had no way to send the 
4041     correct chain if the automatic operation produced an incorrect result.
4042
4043     Now the chain builder is disabled if either:
4044
4045     1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
4046
4047     2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
4048
4049     The reasoning behind this is that an application would not want the
4050     auto chain building to take place if extra chain certificates are
4051     present and it might also want a means of sending no additional
4052     certificates (for example the chain has two certificates and the
4053     root is omitted).
4054     [Steve Henson]
4055
4056  *) Add the possibility to build without the ENGINE framework.
4057     [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
4058
4059  *) Under Win32 gmtime() can return NULL: check return value in
4060     OPENSSL_gmtime(). Add error code for case where gmtime() fails.
4061     [Steve Henson]
4062
4063  *) DSA routines: under certain error conditions uninitialized BN objects
4064     could be freed. Solution: make sure initialization is performed early
4065     enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
4066     Nils Larsch <nla@trustcenter.de> via PR#459)
4067     [Lutz Jaenicke]
4068
4069  *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
4070     checked on reconnect on the client side, therefore session resumption
4071     could still fail with a "ssl session id is different" error. This
4072     behaviour is masked when SSL_OP_ALL is used due to
4073     SSL_OP_MICROSOFT_SESS_ID_BUG being set.
4074     Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
4075     followup to PR #377.
4076     [Lutz Jaenicke]
4077
4078  *) IA-32 assembler support enhancements: unified ELF targets, support
4079     for SCO/Caldera platforms, fix for Cygwin shared build.
4080     [Andy Polyakov]
4081
4082  *) Add support for FreeBSD on sparc64.  As a consequence, support for
4083     FreeBSD on non-x86 processors is separate from x86 processors on
4084     the config script, much like the NetBSD support.
4085     [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
4086
4087 Changes between 0.9.6h and 0.9.7  [31 Dec 2002]
4088
4089  [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
4090  OpenSSL 0.9.7.]
4091
4092  *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
4093     code (06) was taken as the first octet of the session ID and the last
4094     octet was ignored consequently. As a result SSLv2 client side session
4095     caching could not have worked due to the session ID mismatch between
4096     client and server.
4097     Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
4098     PR #377.
4099     [Lutz Jaenicke]
4100
4101  *) Change the declaration of needed Kerberos libraries to use EX_LIBS
4102     instead of the special (and badly supported) LIBKRB5.  LIBKRB5 is
4103     removed entirely.
4104     [Richard Levitte]
4105
4106  *) The hw_ncipher.c engine requires dynamic locks.  Unfortunately, it
4107     seems that in spite of existing for more than a year, many application
4108     author have done nothing to provide the necessary callbacks, which
4109     means that this particular engine will not work properly anywhere.
4110     This is a very unfortunate situation which forces us, in the name
4111     of usability, to give the hw_ncipher.c a static lock, which is part
4112     of libcrypto.
4113     NOTE: This is for the 0.9.7 series ONLY.  This hack will never
4114     appear in 0.9.8 or later.  We EXPECT application authors to have
4115     dealt properly with this when 0.9.8 is released (unless we actually
4116     make such changes in the libcrypto locking code that changes will
4117     have to be made anyway).
4118     [Richard Levitte]
4119
4120  *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
4121     octets have been read, EOF or an error occurs. Without this change
4122     some truncated ASN1 structures will not produce an error.
4123     [Steve Henson]
4124
4125  *) Disable Heimdal support, since it hasn't been fully implemented.
4126     Still give the possibility to force the use of Heimdal, but with
4127     warnings and a request that patches get sent to openssl-dev.
4128     [Richard Levitte]
4129
4130  *) Add the VC-CE target, introduce the WINCE sysname, and add
4131     INSTALL.WCE and appropriate conditionals to make it build.
4132     [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
4133
4134  *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
4135     cygssl-x.y.z.dll, where x, y and z are the major, minor and
4136     edit numbers of the version.
4137     [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
4138
4139  *) Introduce safe string copy and catenation functions
4140     (BUF_strlcpy() and BUF_strlcat()).
4141     [Ben Laurie (CHATS) and Richard Levitte]
4142
4143  *) Avoid using fixed-size buffers for one-line DNs.
4144     [Ben Laurie (CHATS)]
4145
4146  *) Add BUF_MEM_grow_clean() to avoid information leakage when
4147     resizing buffers containing secrets, and use where appropriate.
4148     [Ben Laurie (CHATS)]
4149
4150  *) Avoid using fixed size buffers for configuration file location.
4151     [Ben Laurie (CHATS)]
4152
4153  *) Avoid filename truncation for various CA files.
4154     [Ben Laurie (CHATS)]
4155
4156  *) Use sizeof in preference to magic numbers.
4157     [Ben Laurie (CHATS)]
4158
4159  *) Avoid filename truncation in cert requests.
4160     [Ben Laurie (CHATS)]
4161
4162  *) Add assertions to check for (supposedly impossible) buffer
4163     overflows.
4164     [Ben Laurie (CHATS)]
4165
4166  *) Don't cache truncated DNS entries in the local cache (this could
4167     potentially lead to a spoofing attack).
4168     [Ben Laurie (CHATS)]
4169
4170  *) Fix various buffers to be large enough for hex/decimal
4171     representations in a platform independent manner.
4172     [Ben Laurie (CHATS)]
4173
4174  *) Add CRYPTO_realloc_clean() to avoid information leakage when
4175     resizing buffers containing secrets, and use where appropriate.
4176     [Ben Laurie (CHATS)]
4177
4178  *) Add BIO_indent() to avoid much slightly worrying code to do
4179     indents.
4180     [Ben Laurie (CHATS)]
4181
4182  *) Convert sprintf()/BIO_puts() to BIO_printf().
4183     [Ben Laurie (CHATS)]
4184
4185  *) buffer_gets() could terminate with the buffer only half
4186     full. Fixed.
4187     [Ben Laurie (CHATS)]
4188
4189  *) Add assertions to prevent user-supplied crypto functions from
4190     overflowing internal buffers by having large block sizes, etc.
4191     [Ben Laurie (CHATS)]
4192
4193  *) New OPENSSL_assert() macro (similar to assert(), but enabled
4194     unconditionally).
4195     [Ben Laurie (CHATS)]
4196
4197  *) Eliminate unused copy of key in RC4.
4198     [Ben Laurie (CHATS)]
4199
4200  *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
4201     [Ben Laurie (CHATS)]
4202
4203  *) Fix off-by-one error in EGD path.
4204     [Ben Laurie (CHATS)]
4205
4206  *) If RANDFILE path is too long, ignore instead of truncating.
4207     [Ben Laurie (CHATS)]
4208
4209  *) Eliminate unused and incorrectly sized X.509 structure
4210     CBCParameter.
4211     [Ben Laurie (CHATS)]
4212
4213  *) Eliminate unused and dangerous function knumber().
4214     [Ben Laurie (CHATS)]
4215
4216  *) Eliminate unused and dangerous structure, KSSL_ERR.
4217     [Ben Laurie (CHATS)]
4218
4219  *) Protect against overlong session ID context length in an encoded
4220     session object. Since these are local, this does not appear to be
4221     exploitable.
4222     [Ben Laurie (CHATS)]
4223
4224  *) Change from security patch (see 0.9.6e below) that did not affect
4225     the 0.9.6 release series:
4226
4227     Remote buffer overflow in SSL3 protocol - an attacker could
4228     supply an oversized master key in Kerberos-enabled versions.
4229     (CVE-2002-0657)
4230     [Ben Laurie (CHATS)]
4231
4232  *) Change the SSL kerb5 codes to match RFC 2712.
4233     [Richard Levitte]
4234
4235  *) Make -nameopt work fully for req and add -reqopt switch.
4236     [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
4237
4238  *) The "block size" for block ciphers in CFB and OFB mode should be 1.
4239     [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
4240
4241  *) Make sure tests can be performed even if the corresponding algorithms
4242     have been removed entirely.  This was also the last step to make
4243     OpenSSL compilable with DJGPP under all reasonable conditions.
4244     [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
4245
4246  *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
4247     to allow version independent disabling of normally unselected ciphers,
4248     which may be activated as a side-effect of selecting a single cipher.
4249
4250     (E.g., cipher list string "RSA" enables ciphersuites that are left
4251     out of "ALL" because they do not provide symmetric encryption.
4252     "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
4253     [Lutz Jaenicke, Bodo Moeller]
4254
4255  *) Add appropriate support for separate platform-dependent build
4256     directories.  The recommended way to make a platform-dependent
4257     build directory is the following (tested on Linux), maybe with
4258     some local tweaks:
4259
4260	# Place yourself outside of the OpenSSL source tree.  In
4261	# this example, the environment variable OPENSSL_SOURCE
4262	# is assumed to contain the absolute OpenSSL source directory.
4263	mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
4264	cd objtree/"`uname -s`-`uname -r`-`uname -m`"
4265	(cd $OPENSSL_SOURCE; find . -type f) | while read F; do
4266		mkdir -p `dirname $F`
4267		ln -s $OPENSSL_SOURCE/$F $F
4268	done
4269
4270     To be absolutely sure not to disturb the source tree, a "make clean"
4271     is a good thing.  If it isn't successfull, don't worry about it,
4272     it probably means the source directory is very clean.
4273     [Richard Levitte]
4274
4275  *) Make sure any ENGINE control commands make local copies of string
4276     pointers passed to them whenever necessary. Otherwise it is possible
4277     the caller may have overwritten (or deallocated) the original string
4278     data when a later ENGINE operation tries to use the stored values.
4279     [G�tz Babin-Ebell <babinebell@trustcenter.de>]
4280
4281  *) Improve diagnostics in file reading and command-line digests.
4282     [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
4283
4284  *) Add AES modes CFB and OFB to the object database.  Correct an
4285     error in AES-CFB decryption.
4286     [Richard Levitte]
4287
4288  *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 
4289     allows existing EVP_CIPHER_CTX structures to be reused after
4290     calling EVP_*Final(). This behaviour is used by encryption
4291     BIOs and some applications. This has the side effect that
4292     applications must explicitly clean up cipher contexts with
4293     EVP_CIPHER_CTX_cleanup() or they will leak memory.
4294     [Steve Henson]
4295
4296  *) Check the values of dna and dnb in bn_mul_recursive before calling
4297     bn_mul_comba (a non zero value means the a or b arrays do not contain
4298     n2 elements) and fallback to bn_mul_normal if either is not zero.
4299     [Steve Henson]
4300
4301  *) Fix escaping of non-ASCII characters when using the -subj option
4302     of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
4303     [Lutz Jaenicke]
4304
4305  *) Make object definitions compliant to LDAP (RFC2256): SN is the short
4306     form for "surname", serialNumber has no short form.
4307     Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
4308     therefore remove "mail" short name for "internet 7".
4309     The OID for unique identifiers in X509 certificates is
4310     x500UniqueIdentifier, not uniqueIdentifier.
4311     Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
4312     [Lutz Jaenicke]
4313
4314  *) Add an "init" command to the ENGINE config module and auto initialize
4315     ENGINEs. Without any "init" command the ENGINE will be initialized 
4316     after all ctrl commands have been executed on it. If init=1 the 
4317     ENGINE is initailized at that point (ctrls before that point are run
4318     on the uninitialized ENGINE and after on the initialized one). If
4319     init=0 then the ENGINE will not be iniatialized at all.
4320     [Steve Henson]
4321
4322  *) Fix the 'app_verify_callback' interface so that the user-defined
4323     argument is actually passed to the callback: In the
4324     SSL_CTX_set_cert_verify_callback() prototype, the callback
4325     declaration has been changed from
4326          int (*cb)()
4327     into
4328          int (*cb)(X509_STORE_CTX *,void *);
4329     in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
4330          i=s->ctx->app_verify_callback(&ctx)
4331     has been changed into
4332          i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
4333
4334     To update applications using SSL_CTX_set_cert_verify_callback(),
4335     a dummy argument can be added to their callback functions.
4336     [D. K. Smetters <smetters@parc.xerox.com>]
4337
4338  *) Added the '4758cca' ENGINE to support IBM 4758 cards.
4339     [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
4340
4341  *) Add and OPENSSL_LOAD_CONF define which will cause
4342     OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
4343     This allows older applications to transparently support certain
4344     OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
4345     Two new functions OPENSSL_add_all_algorithms_noconf() which will never
4346     load the config file and OPENSSL_add_all_algorithms_conf() which will
4347     always load it have also been added.
4348     [Steve Henson]
4349
4350  *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
4351     Adjust NIDs and EVP layer.
4352     [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
4353
4354  *) Config modules support in openssl utility.
4355
4356     Most commands now load modules from the config file,
4357     though in a few (such as version) this isn't done 
4358     because it couldn't be used for anything.
4359
4360     In the case of ca and req the config file used is
4361     the same as the utility itself: that is the -config
4362     command line option can be used to specify an
4363     alternative file.
4364     [Steve Henson]
4365
4366  *) Move default behaviour from OPENSSL_config(). If appname is NULL
4367     use "openssl_conf" if filename is NULL use default openssl config file.
4368     [Steve Henson]
4369
4370  *) Add an argument to OPENSSL_config() to allow the use of an alternative
4371     config section name. Add a new flag to tolerate a missing config file
4372     and move code to CONF_modules_load_file().
4373     [Steve Henson]
4374
4375  *) Support for crypto accelerator cards from Accelerated Encryption
4376     Processing, www.aep.ie.  (Use engine 'aep')
4377     The support was copied from 0.9.6c [engine] and adapted/corrected
4378     to work with the new engine framework.
4379     [AEP Inc. and Richard Levitte]
4380
4381  *) Support for SureWare crypto accelerator cards from Baltimore
4382     Technologies.  (Use engine 'sureware')
4383     The support was copied from 0.9.6c [engine] and adapted
4384     to work with the new engine framework.
4385     [Richard Levitte]
4386
4387  *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
4388     make the newer ENGINE framework commands for the CHIL engine work.
4389     [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
4390
4391  *) Make it possible to produce shared libraries on ReliantUNIX.
4392     [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
4393
4394  *) Add the configuration target debug-linux-ppro.
4395     Make 'openssl rsa' use the general key loading routines
4396     implemented in apps.c, and make those routines able to
4397     handle the key format FORMAT_NETSCAPE and the variant
4398     FORMAT_IISSGC.
4399     [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
4400
4401 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
4402     [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
4403
4404  *) Add -keyform to rsautl, and document -engine.
4405     [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
4406
4407  *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
4408     BIO_R_NO_SUCH_FILE error code rather than the generic
4409     ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
4410     [Ben Laurie]
4411
4412  *) Add new functions
4413          ERR_peek_last_error
4414          ERR_peek_last_error_line
4415          ERR_peek_last_error_line_data.
4416     These are similar to
4417          ERR_peek_error
4418          ERR_peek_error_line
4419          ERR_peek_error_line_data,
4420     but report on the latest error recorded rather than the first one
4421     still in the error queue.
4422     [Ben Laurie, Bodo Moeller]
4423        
4424  *) default_algorithms option in ENGINE config module. This allows things
4425     like:
4426     default_algorithms = ALL
4427     default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
4428     [Steve Henson]
4429
4430  *) Prelminary ENGINE config module.
4431     [Steve Henson]
4432
4433  *) New experimental application configuration code.
4434     [Steve Henson]
4435
4436  *) Change the AES code to follow the same name structure as all other
4437     symmetric ciphers, and behave the same way.  Move everything to
4438     the directory crypto/aes, thereby obsoleting crypto/rijndael.
4439     [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
4440
4441  *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
4442     [Ben Laurie and Theo de Raadt]
4443
4444  *) Add option to output public keys in req command.
4445     [Massimiliano Pala madwolf@openca.org]
4446
4447  *) Use wNAFs in EC_POINTs_mul() for improved efficiency
4448     (up to about 10% better than before for P-192 and P-224).
4449     [Bodo Moeller]
4450
4451  *) New functions/macros
4452
4453          SSL_CTX_set_msg_callback(ctx, cb)
4454          SSL_CTX_set_msg_callback_arg(ctx, arg)
4455          SSL_set_msg_callback(ssl, cb)
4456          SSL_set_msg_callback_arg(ssl, arg)
4457
4458     to request calling a callback function
4459
4460          void cb(int write_p, int version, int content_type,
4461                  const void *buf, size_t len, SSL *ssl, void *arg)
4462
4463     whenever a protocol message has been completely received
4464     (write_p == 0) or sent (write_p == 1).  Here 'version' is the
4465     protocol version  according to which the SSL library interprets
4466     the current protocol message (SSL2_VERSION, SSL3_VERSION, or
4467     TLS1_VERSION).  'content_type' is 0 in the case of SSL 2.0, or
4468     the content type as defined in the SSL 3.0/TLS 1.0 protocol
4469     specification (change_cipher_spec(20), alert(21), handshake(22)).
4470     'buf' and 'len' point to the actual message, 'ssl' to the
4471     SSL object, and 'arg' is the application-defined value set by
4472     SSL[_CTX]_set_msg_callback_arg().
4473
4474     'openssl s_client' and 'openssl s_server' have new '-msg' options
4475     to enable a callback that displays all protocol messages.
4476     [Bodo Moeller]
4477
4478  *) Change the shared library support so shared libraries are built as
4479     soon as the corresponding static library is finished, and thereby get
4480     openssl and the test programs linked against the shared library.
4481     This still only happens when the keyword "shard" has been given to
4482     the configuration scripts.
4483
4484     NOTE: shared library support is still an experimental thing, and
4485     backward binary compatibility is still not guaranteed.
4486     ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
4487
4488  *) Add support for Subject Information Access extension.
4489     [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
4490
4491  *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
4492     additional bytes when new memory had to be allocated, not just
4493     when reusing an existing buffer.
4494     [Bodo Moeller]
4495
4496  *) New command line and configuration option 'utf8' for the req command.
4497     This allows field values to be specified as UTF8 strings.
4498     [Steve Henson]
4499
4500  *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
4501     runs for the former and machine-readable output for the latter.
4502     [Ben Laurie]
4503
4504  *) Add '-noemailDN' option to 'openssl ca'.  This prevents inclusion
4505     of the e-mail address in the DN (i.e., it will go into a certificate
4506     extension only).  The new configuration file option 'email_in_dn = no'
4507     has the same effect.
4508     [Massimiliano Pala madwolf@openca.org]
4509
4510  *) Change all functions with names starting with des_ to be starting
4511     with DES_ instead.  Add wrappers that are compatible with libdes,
4512     but are named _ossl_old_des_*.  Finally, add macros that map the
4513     des_* symbols to the corresponding _ossl_old_des_* if libdes
4514     compatibility is desired.  If OpenSSL 0.9.6c compatibility is
4515     desired, the des_* symbols will be mapped to DES_*, with one
4516     exception.
4517
4518     Since we provide two compatibility mappings, the user needs to
4519     define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
4520     compatibility is desired.  The default (i.e., when that macro
4521     isn't defined) is OpenSSL 0.9.6c compatibility.
4522
4523     There are also macros that enable and disable the support of old
4524     des functions altogether.  Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
4525     and OPENSSL_DISABLE_OLD_DES_SUPPORT.  If none or both of those
4526     are defined, the default will apply: to support the old des routines.
4527
4528     In either case, one must include openssl/des.h to get the correct
4529     definitions.  Do not try to just include openssl/des_old.h, that
4530     won't work.
4531
4532     NOTE: This is a major break of an old API into a new one.  Software
4533     authors are encouraged to switch to the DES_ style functions.  Some
4534     time in the future, des_old.h and the libdes compatibility functions
4535     will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
4536     default), and then completely removed.
4537     [Richard Levitte]
4538
4539  *) Test for certificates which contain unsupported critical extensions.
4540     If such a certificate is found during a verify operation it is 
4541     rejected by default: this behaviour can be overridden by either
4542     handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
4543     by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
4544     X509_supported_extension() has also been added which returns 1 if a
4545     particular extension is supported.
4546     [Steve Henson]
4547
4548  *) Modify the behaviour of EVP cipher functions in similar way to digests
4549     to retain compatibility with existing code.
4550     [Steve Henson]
4551
4552  *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
4553     compatibility with existing code. In particular the 'ctx' parameter does
4554     not have to be to be initialized before the call to EVP_DigestInit() and
4555     it is tidied up after a call to EVP_DigestFinal(). New function
4556     EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
4557     EVP_MD_CTX_copy() changed to not require the destination to be
4558     initialized valid and new function EVP_MD_CTX_copy_ex() added which
4559     requires the destination to be valid.
4560
4561     Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
4562     EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
4563     [Steve Henson]
4564
4565  *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
4566     so that complete 'Handshake' protocol structures are kept in memory
4567     instead of overwriting 'msg_type' and 'length' with 'body' data.
4568     [Bodo Moeller]
4569
4570  *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
4571     [Massimo Santin via Richard Levitte]
4572
4573  *) Major restructuring to the underlying ENGINE code. This includes
4574     reduction of linker bloat, separation of pure "ENGINE" manipulation
4575     (initialisation, etc) from functionality dealing with implementations
4576     of specific crypto iterfaces. This change also introduces integrated
4577     support for symmetric ciphers and digest implementations - so ENGINEs
4578     can now accelerate these by providing EVP_CIPHER and EVP_MD
4579     implementations of their own. This is detailed in crypto/engine/README
4580     as it couldn't be adequately described here. However, there are a few
4581     API changes worth noting - some RSA, DSA, DH, and RAND functions that
4582     were changed in the original introduction of ENGINE code have now
4583     reverted back - the hooking from this code to ENGINE is now a good
4584     deal more passive and at run-time, operations deal directly with
4585     RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
4586     dereferencing through an ENGINE pointer any more. Also, the ENGINE
4587     functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
4588     they were not being used by the framework as there is no concept of a
4589     BIGNUM_METHOD and they could not be generalised to the new
4590     'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
4591     ENGINE_cpy() has been removed as it cannot be consistently defined in
4592     the new code.
4593     [Geoff Thorpe]
4594
4595  *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
4596     [Steve Henson]
4597
4598  *) Change mkdef.pl to sort symbols that get the same entry number,
4599     and make sure the automatically generated functions ERR_load_*
4600     become part of libeay.num as well.
4601     [Richard Levitte]
4602
4603  *) New function SSL_renegotiate_pending().  This returns true once
4604     renegotiation has been requested (either SSL_renegotiate() call
4605     or HelloRequest/ClientHello receveived from the peer) and becomes
4606     false once a handshake has been completed.
4607     (For servers, SSL_renegotiate() followed by SSL_do_handshake()
4608     sends a HelloRequest, but does not ensure that a handshake takes
4609     place.  SSL_renegotiate_pending() is useful for checking if the
4610     client has followed the request.)
4611     [Bodo Moeller]
4612
4613  *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
4614     By default, clients may request session resumption even during
4615     renegotiation (if session ID contexts permit); with this option,
4616     session resumption is possible only in the first handshake.
4617
4618     SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL.  This makes
4619     more bits available for options that should not be part of
4620     SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
4621     [Bodo Moeller]
4622
4623  *) Add some demos for certificate and certificate request creation.
4624     [Steve Henson]
4625
4626  *) Make maximum certificate chain size accepted from the peer application
4627     settable (SSL*_get/set_max_cert_list()), as proposed by
4628     "Douglas E. Engert" <deengert@anl.gov>.
4629     [Lutz Jaenicke]
4630
4631  *) Add support for shared libraries for Unixware-7
4632     (Boyd Lynn Gerber <gerberb@zenez.com>).
4633     [Lutz Jaenicke]
4634
4635  *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
4636     be done prior to destruction. Use this to unload error strings from
4637     ENGINEs that load their own error strings. NB: This adds two new API
4638     functions to "get" and "set" this destroy handler in an ENGINE.
4639     [Geoff Thorpe]
4640
4641  *) Alter all existing ENGINE implementations (except "openssl" and
4642     "openbsd") to dynamically instantiate their own error strings. This
4643     makes them more flexible to be built both as statically-linked ENGINEs
4644     and self-contained shared-libraries loadable via the "dynamic" ENGINE.
4645     Also, add stub code to each that makes building them as self-contained
4646     shared-libraries easier (see README.ENGINE).
4647     [Geoff Thorpe]
4648
4649  *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
4650     implementations into applications that are completely implemented in
4651     self-contained shared-libraries. The "dynamic" ENGINE exposes control
4652     commands that can be used to configure what shared-library to load and
4653     to control aspects of the way it is handled. Also, made an update to
4654     the README.ENGINE file that brings its information up-to-date and
4655     provides some information and instructions on the "dynamic" ENGINE
4656     (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
4657     [Geoff Thorpe]
4658
4659  *) Make it possible to unload ranges of ERR strings with a new
4660     "ERR_unload_strings" function.
4661     [Geoff Thorpe]
4662
4663  *) Add a copy() function to EVP_MD.
4664     [Ben Laurie]
4665
4666  *) Make EVP_MD routines take a context pointer instead of just the
4667     md_data void pointer.
4668     [Ben Laurie]
4669
4670  *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
4671     that the digest can only process a single chunk of data
4672     (typically because it is provided by a piece of
4673     hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
4674     is only going to provide a single chunk of data, and hence the
4675     framework needn't accumulate the data for oneshot drivers.
4676     [Ben Laurie]
4677
4678  *) As with "ERR", make it possible to replace the underlying "ex_data"
4679     functions. This change also alters the storage and management of global
4680     ex_data state - it's now all inside ex_data.c and all "class" code (eg.
4681     RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
4682     index counters. The API functions that use this state have been changed
4683     to take a "class_index" rather than pointers to the class's local STACK
4684     and counter, and there is now an API function to dynamically create new
4685     classes. This centralisation allows us to (a) plug a lot of the
4686     thread-safety problems that existed, and (b) makes it possible to clean
4687     up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
4688     such data would previously have always leaked in application code and
4689     workarounds were in place to make the memory debugging turn a blind eye
4690     to it. Application code that doesn't use this new function will still
4691     leak as before, but their memory debugging output will announce it now
4692     rather than letting it slide.
4693
4694     Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
4695     induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
4696     has a return value to indicate success or failure.
4697     [Geoff Thorpe]
4698
4699  *) Make it possible to replace the underlying "ERR" functions such that the
4700     global state (2 LHASH tables and 2 locks) is only used by the "default"
4701     implementation. This change also adds two functions to "get" and "set"
4702     the implementation prior to it being automatically set the first time
4703     any other ERR function takes place. Ie. an application can call "get",
4704     pass the return value to a module it has just loaded, and that module
4705     can call its own "set" function using that value. This means the
4706     module's "ERR" operations will use (and modify) the error state in the
4707     application and not in its own statically linked copy of OpenSSL code.
4708     [Geoff Thorpe]
4709
4710  *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
4711     reference counts. This performs normal REF_PRINT/REF_CHECK macros on
4712     the operation, and provides a more encapsulated way for external code
4713     (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
4714     to use these functions rather than manually incrementing the counts.
4715
4716     Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
4717     [Geoff Thorpe]
4718
4719  *) Add EVP test program.
4720     [Ben Laurie]
4721
4722  *) Add symmetric cipher support to ENGINE. Expect the API to change!
4723     [Ben Laurie]
4724
4725  *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
4726     X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
4727     X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
4728     These allow a CRL to be built without having to access X509_CRL fields
4729     directly. Modify 'ca' application to use new functions.
4730     [Steve Henson]
4731
4732  *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
4733     bug workarounds. Rollback attack detection is a security feature.
4734     The problem will only arise on OpenSSL servers when TLSv1 is not
4735     available (sslv3_server_method() or SSL_OP_NO_TLSv1).
4736     Software authors not wanting to support TLSv1 will have special reasons
4737     for their choice and can explicitly enable this option.
4738     [Bodo Moeller, Lutz Jaenicke]
4739
4740  *) Rationalise EVP so it can be extended: don't include a union of
4741     cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
4742     (similar to those existing for EVP_CIPHER_CTX).
4743     Usage example:
4744
4745         EVP_MD_CTX md;
4746
4747         EVP_MD_CTX_init(&md);             /* new function call */
4748         EVP_DigestInit(&md, EVP_sha1());
4749         EVP_DigestUpdate(&md, in, len);
4750         EVP_DigestFinal(&md, out, NULL);
4751         EVP_MD_CTX_cleanup(&md);          /* new function call */
4752
4753     [Ben Laurie]
4754
4755  *) Make DES key schedule conform to the usual scheme, as well as
4756     correcting its structure. This means that calls to DES functions
4757     now have to pass a pointer to a des_key_schedule instead of a
4758     plain des_key_schedule (which was actually always a pointer
4759     anyway): E.g.,
4760
4761         des_key_schedule ks;
4762
4763	 des_set_key_checked(..., &ks);
4764	 des_ncbc_encrypt(..., &ks, ...);
4765
4766     (Note that a later change renames 'des_...' into 'DES_...'.)
4767     [Ben Laurie]
4768
4769  *) Initial reduction of linker bloat: the use of some functions, such as
4770     PEM causes large amounts of unused functions to be linked in due to
4771     poor organisation. For example pem_all.c contains every PEM function
4772     which has a knock on effect of linking in large amounts of (unused)
4773     ASN1 code. Grouping together similar functions and splitting unrelated
4774     functions prevents this.
4775     [Steve Henson]
4776
4777  *) Cleanup of EVP macros.
4778     [Ben Laurie]
4779
4780  *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
4781     correct _ecb suffix.
4782     [Ben Laurie]
4783
4784  *) Add initial OCSP responder support to ocsp application. The
4785     revocation information is handled using the text based index
4786     use by the ca application. The responder can either handle
4787     requests generated internally, supplied in files (for example
4788     via a CGI script) or using an internal minimal server.
4789     [Steve Henson]
4790
4791  *) Add configuration choices to get zlib compression for TLS.
4792     [Richard Levitte]
4793
4794  *) Changes to Kerberos SSL for RFC 2712 compliance:
4795     1.  Implemented real KerberosWrapper, instead of just using
4796         KRB5 AP_REQ message.  [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
4797     2.  Implemented optional authenticator field of KerberosWrapper.
4798
4799     Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
4800     and authenticator structs; see crypto/krb5/.
4801
4802     Generalized Kerberos calls to support multiple Kerberos libraries.
4803     [Vern Staats <staatsvr@asc.hpc.mil>,
4804      Jeffrey Altman <jaltman@columbia.edu>
4805      via Richard Levitte]
4806
4807  *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
4808     already does with RSA. testdsa.h now has 'priv_key/pub_key'
4809     values for each of the key sizes rather than having just
4810     parameters (and 'speed' generating keys each time).
4811     [Geoff Thorpe]
4812
4813  *) Speed up EVP routines.
4814     Before:
4815encrypt
4816type              8 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
4817des-cbc           4408.85k     5560.51k     5778.46k     5862.20k     5825.16k
4818des-cbc           4389.55k     5571.17k     5792.23k     5846.91k     5832.11k
4819des-cbc           4394.32k     5575.92k     5807.44k     5848.37k     5841.30k
4820decrypt
4821des-cbc           3482.66k     5069.49k     5496.39k     5614.16k     5639.28k
4822des-cbc           3480.74k     5068.76k     5510.34k     5609.87k     5635.52k
4823des-cbc           3483.72k     5067.62k     5504.60k     5708.01k     5724.80k
4824     After:
4825encrypt
4826des-cbc           4660.16k     5650.19k     5807.19k     5827.13k     5783.32k
4827decrypt
4828des-cbc           3624.96k     5258.21k     5530.91k     5624.30k     5628.26k
4829     [Ben Laurie]
4830
4831  *) Added the OS2-EMX target.
4832     ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
4833
4834  *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
4835     to support NCONF routines in extension code. New function CONF_set_nconf()
4836     to allow functions which take an NCONF to also handle the old LHASH
4837     structure: this means that the old CONF compatible routines can be
4838     retained (in particular wrt extensions) without having to duplicate the
4839     code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
4840     [Steve Henson]
4841
4842  *) Enhance the general user interface with mechanisms for inner control
4843     and with possibilities to have yes/no kind of prompts.
4844     [Richard Levitte]
4845
4846  *) Change all calls to low level digest routines in the library and
4847     applications to use EVP. Add missing calls to HMAC_cleanup() and
4848     don't assume HMAC_CTX can be copied using memcpy().
4849     [Verdon Walker <VWalker@novell.com>, Steve Henson]
4850
4851  *) Add the possibility to control engines through control names but with
4852     arbitrary arguments instead of just a string.
4853     Change the key loaders to take a UI_METHOD instead of a callback
4854     function pointer.  NOTE: this breaks binary compatibility with earlier
4855     versions of OpenSSL [engine].
4856     Adapt the nCipher code for these new conditions and add a card insertion
4857     callback.
4858     [Richard Levitte]
4859
4860  *) Enhance the general user interface with mechanisms to better support
4861     dialog box interfaces, application-defined prompts, the possibility
4862     to use defaults (for example default passwords from somewhere else)
4863     and interrupts/cancellations.
4864     [Richard Levitte]
4865
4866  *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
4867     attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
4868     [Steve Henson]
4869
4870  *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
4871     tidy up some unnecessarily weird code in 'sk_new()').
4872     [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
4873
4874  *) Change the key loading routines for ENGINEs to use the same kind
4875     callback (pem_password_cb) as all other routines that need this
4876     kind of callback.
4877     [Richard Levitte]
4878
4879  *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
4880     256 bit (=32 byte) keys. Of course seeding with more entropy bytes
4881     than this minimum value is recommended.
4882     [Lutz Jaenicke]
4883
4884  *) New random seeder for OpenVMS, using the system process statistics
4885     that are easily reachable.
4886     [Richard Levitte]
4887
4888  *) Windows apparently can't transparently handle global
4889     variables defined in DLLs. Initialisations such as:
4890
4891        const ASN1_ITEM *it = &ASN1_INTEGER_it;
4892
4893     wont compile. This is used by the any applications that need to
4894     declare their own ASN1 modules. This was fixed by adding the option
4895     EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
4896     needed for static libraries under Win32.
4897     [Steve Henson]
4898
4899  *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
4900     setting of purpose and trust fields. New X509_STORE trust and
4901     purpose functions and tidy up setting in other SSL functions.
4902     [Steve Henson]
4903
4904  *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
4905     structure. These are inherited by X509_STORE_CTX when it is 
4906     initialised. This allows various defaults to be set in the
4907     X509_STORE structure (such as flags for CRL checking and custom
4908     purpose or trust settings) for functions which only use X509_STORE_CTX
4909     internally such as S/MIME.
4910
4911     Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
4912     trust settings if they are not set in X509_STORE. This allows X509_STORE
4913     purposes and trust (in S/MIME for example) to override any set by default.
4914
4915     Add command line options for CRL checking to smime, s_client and s_server
4916     applications.
4917     [Steve Henson]
4918
4919  *) Initial CRL based revocation checking. If the CRL checking flag(s)
4920     are set then the CRL is looked up in the X509_STORE structure and
4921     its validity and signature checked, then if the certificate is found
4922     in the CRL the verify fails with a revoked error.
4923
4924     Various new CRL related callbacks added to X509_STORE_CTX structure.
4925
4926     Command line options added to 'verify' application to support this.
4927
4928     This needs some additional work, such as being able to handle multiple
4929     CRLs with different times, extension based lookup (rather than just
4930     by subject name) and ultimately more complete V2 CRL extension
4931     handling.
4932     [Steve Henson]
4933
4934  *) Add a general user interface API (crypto/ui/).  This is designed
4935     to replace things like des_read_password and friends (backward
4936     compatibility functions using this new API are provided).
4937     The purpose is to remove prompting functions from the DES code
4938     section as well as provide for prompting through dialog boxes in
4939     a window system and the like.
4940     [Richard Levitte]
4941
4942  *) Add "ex_data" support to ENGINE so implementations can add state at a
4943     per-structure level rather than having to store it globally.
4944     [Geoff]
4945
4946  *) Make it possible for ENGINE structures to be copied when retrieved by
4947     ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
4948     This causes the "original" ENGINE structure to act like a template,
4949     analogous to the RSA vs. RSA_METHOD type of separation. Because of this
4950     operational state can be localised to each ENGINE structure, despite the
4951     fact they all share the same "methods". New ENGINE structures returned in
4952     this case have no functional references and the return value is the single
4953     structural reference. This matches the single structural reference returned
4954     by ENGINE_by_id() normally, when it is incremented on the pre-existing
4955     ENGINE structure.
4956     [Geoff]
4957
4958  *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
4959     needs to match any other type at all we need to manually clear the
4960     tag cache.
4961     [Steve Henson]
4962
4963  *) Changes to the "openssl engine" utility to include;
4964     - verbosity levels ('-v', '-vv', and '-vvv') that provide information
4965       about an ENGINE's available control commands.
4966     - executing control commands from command line arguments using the
4967       '-pre' and '-post' switches. '-post' is only used if '-t' is
4968       specified and the ENGINE is successfully initialised. The syntax for
4969       the individual commands are colon-separated, for example;
4970	 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
4971     [Geoff]
4972
4973  *) New dynamic control command support for ENGINEs. ENGINEs can now
4974     declare their own commands (numbers), names (strings), descriptions,
4975     and input types for run-time discovery by calling applications. A
4976     subset of these commands are implicitly classed as "executable"
4977     depending on their input type, and only these can be invoked through
4978     the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
4979     can be based on user input, config files, etc). The distinction is
4980     that "executable" commands cannot return anything other than a boolean
4981     result and can only support numeric or string input, whereas some
4982     discoverable commands may only be for direct use through
4983     ENGINE_ctrl(), eg. supporting the exchange of binary data, function
4984     pointers, or other custom uses. The "executable" commands are to
4985     support parameterisations of ENGINE behaviour that can be
4986     unambiguously defined by ENGINEs and used consistently across any
4987     OpenSSL-based application. Commands have been added to all the
4988     existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
4989     control over shared-library paths without source code alterations.
4990     [Geoff]
4991
4992  *) Changed all ENGINE implementations to dynamically allocate their
4993     ENGINEs rather than declaring them statically. Apart from this being
4994     necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
4995     this also allows the implementations to compile without using the
4996     internal engine_int.h header.
4997     [Geoff]
4998
4999  *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
5000     'const' value. Any code that should be able to modify a RAND_METHOD
5001     should already have non-const pointers to it (ie. they should only
5002     modify their own ones).
5003     [Geoff]
5004
5005  *) Made a variety of little tweaks to the ENGINE code.
5006     - "atalla" and "ubsec" string definitions were moved from header files
5007       to C code. "nuron" string definitions were placed in variables
5008       rather than hard-coded - allowing parameterisation of these values
5009       later on via ctrl() commands.
5010     - Removed unused "#if 0"'d code.
5011     - Fixed engine list iteration code so it uses ENGINE_free() to release
5012       structural references.
5013     - Constified the RAND_METHOD element of ENGINE structures.
5014     - Constified various get/set functions as appropriate and added
5015       missing functions (including a catch-all ENGINE_cpy that duplicates
5016       all ENGINE values onto a new ENGINE except reference counts/state).
5017     - Removed NULL parameter checks in get/set functions. Setting a method
5018       or function to NULL is a way of cancelling out a previously set
5019       value.  Passing a NULL ENGINE parameter is just plain stupid anyway
5020       and doesn't justify the extra error symbols and code.
5021     - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
5022       flags from engine_int.h to engine.h.
5023     - Changed prototypes for ENGINE handler functions (init(), finish(),
5024       ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
5025     [Geoff]
5026
5027  *) Implement binary inversion algorithm for BN_mod_inverse in addition
5028     to the algorithm using long division.  The binary algorithm can be
5029     used only if the modulus is odd.  On 32-bit systems, it is faster
5030     only for relatively small moduli (roughly 20-30% for 128-bit moduli,
5031     roughly 5-15% for 256-bit moduli), so we use it only for moduli
5032     up to 450 bits.  In 64-bit environments, the binary algorithm
5033     appears to be advantageous for much longer moduli; here we use it
5034     for moduli up to 2048 bits.
5035     [Bodo Moeller]
5036
5037  *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
5038     could not support the combine flag in choice fields.
5039     [Steve Henson]
5040
5041  *) Add a 'copy_extensions' option to the 'ca' utility. This copies
5042     extensions from a certificate request to the certificate.
5043     [Steve Henson]
5044
5045  *) Allow multiple 'certopt' and 'nameopt' options to be separated
5046     by commas. Add 'namopt' and 'certopt' options to the 'ca' config
5047     file: this allows the display of the certificate about to be
5048     signed to be customised, to allow certain fields to be included
5049     or excluded and extension details. The old system didn't display
5050     multicharacter strings properly, omitted fields not in the policy
5051     and couldn't display additional details such as extensions.
5052     [Steve Henson]
5053
5054  *) Function EC_POINTs_mul for multiple scalar multiplication
5055     of an arbitrary number of elliptic curve points
5056          \sum scalars[i]*points[i],
5057     optionally including the generator defined for the EC_GROUP:
5058          scalar*generator +  \sum scalars[i]*points[i].
5059
5060     EC_POINT_mul is a simple wrapper function for the typical case
5061     that the point list has just one item (besides the optional
5062     generator).
5063     [Bodo Moeller]
5064
5065  *) First EC_METHODs for curves over GF(p):
5066
5067     EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
5068     operations and provides various method functions that can also
5069     operate with faster implementations of modular arithmetic.     
5070
5071     EC_GFp_mont_method() reuses most functions that are part of
5072     EC_GFp_simple_method, but uses Montgomery arithmetic.
5073
5074     [Bodo Moeller; point addition and point doubling
5075     implementation directly derived from source code provided by
5076     Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
5077
5078  *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
5079     crypto/ec/ec_lib.c):
5080
5081     Curves are EC_GROUP objects (with an optional group generator)
5082     based on EC_METHODs that are built into the library.
5083
5084     Points are EC_POINT objects based on EC_GROUP objects.
5085
5086     Most of the framework would be able to handle curves over arbitrary
5087     finite fields, but as there are no obvious types for fields other
5088     than GF(p), some functions are limited to that for now.
5089     [Bodo Moeller]
5090
5091  *) Add the -HTTP option to s_server.  It is similar to -WWW, but requires
5092     that the file contains a complete HTTP response.
5093     [Richard Levitte]
5094
5095  *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
5096     change the def and num file printf format specifier from "%-40sXXX"
5097     to "%-39s XXX". The latter will always guarantee a space after the
5098     field while the former will cause them to run together if the field
5099     is 40 of more characters long.
5100     [Steve Henson]
5101
5102  *) Constify the cipher and digest 'method' functions and structures
5103     and modify related functions to take constant EVP_MD and EVP_CIPHER
5104     pointers.
5105     [Steve Henson]
5106
5107  *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
5108     in <openssl/bn.h>.  Also further increase BN_CTX_NUM to 32.
5109     [Bodo Moeller]
5110
5111  *) Modify EVP_Digest*() routines so they now return values. Although the
5112     internal software routines can never fail additional hardware versions
5113     might.
5114     [Steve Henson]
5115
5116  *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
5117
5118     Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
5119     (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
5120
5121     ASN1 error codes
5122          ERR_R_NESTED_ASN1_ERROR
5123          ...
5124          ERR_R_MISSING_ASN1_EOS
5125     were 4 .. 9, conflicting with
5126          ERR_LIB_RSA (= ERR_R_RSA_LIB)
5127          ...
5128          ERR_LIB_PEM (= ERR_R_PEM_LIB).
5129     They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
5130
5131     Add new error code 'ERR_R_INTERNAL_ERROR'.
5132     [Bodo Moeller]
5133
5134  *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
5135     suffices.
5136     [Bodo Moeller]
5137
5138  *) New option '-subj arg' for 'openssl req' and 'openssl ca'.  This
5139     sets the subject name for a new request or supersedes the
5140     subject name in a given request. Formats that can be parsed are
5141          'CN=Some Name, OU=myOU, C=IT'
5142     and
5143          'CN=Some Name/OU=myOU/C=IT'.
5144
5145     Add options '-batch' and '-verbose' to 'openssl req'.
5146     [Massimiliano Pala <madwolf@hackmasters.net>]
5147
5148  *) Introduce the possibility to access global variables through
5149     functions on platform were that's the best way to handle exporting
5150     global variables in shared libraries.  To enable this functionality,
5151     one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
5152     "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
5153     is normally done by Configure or something similar).
5154
5155     To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
5156     in the source file (foo.c) like this:
5157
5158	OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
5159	OPENSSL_IMPLEMENT_GLOBAL(double,bar);
5160
5161     To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
5162     and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
5163
5164	OPENSSL_DECLARE_GLOBAL(int,foo);
5165	#define foo OPENSSL_GLOBAL_REF(foo)
5166	OPENSSL_DECLARE_GLOBAL(double,bar);
5167	#define bar OPENSSL_GLOBAL_REF(bar)
5168
5169     The #defines are very important, and therefore so is including the
5170     header file everywhere where the defined globals are used.
5171
5172     The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
5173     of ASN.1 items, but that structure is a bit different.
5174
5175     The largest change is in util/mkdef.pl which has been enhanced with
5176     better and easier to understand logic to choose which symbols should
5177     go into the Windows .def files as well as a number of fixes and code
5178     cleanup (among others, algorithm keywords are now sorted
5179     lexicographically to avoid constant rewrites).
5180     [Richard Levitte]
5181
5182  *) In BN_div() keep a copy of the sign of 'num' before writing the
5183     result to 'rm' because if rm==num the value will be overwritten
5184     and produce the wrong result if 'num' is negative: this caused
5185     problems with BN_mod() and BN_nnmod().
5186     [Steve Henson]
5187
5188  *) Function OCSP_request_verify(). This checks the signature on an
5189     OCSP request and verifies the signer certificate. The signer
5190     certificate is just checked for a generic purpose and OCSP request
5191     trust settings.
5192     [Steve Henson]
5193
5194  *) Add OCSP_check_validity() function to check the validity of OCSP
5195     responses. OCSP responses are prepared in real time and may only
5196     be a few seconds old. Simply checking that the current time lies
5197     between thisUpdate and nextUpdate max reject otherwise valid responses
5198     caused by either OCSP responder or client clock inaccuracy. Instead
5199     we allow thisUpdate and nextUpdate to fall within a certain period of
5200     the current time. The age of the response can also optionally be
5201     checked. Two new options -validity_period and -status_age added to
5202     ocsp utility.
5203     [Steve Henson]
5204
5205  *) If signature or public key algorithm is unrecognized print out its
5206     OID rather that just UNKNOWN.
5207     [Steve Henson]
5208
5209  *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
5210     OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
5211     ID to be generated from the issuer certificate alone which can then be
5212     passed to OCSP_id_issuer_cmp().
5213     [Steve Henson]
5214
5215  *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
5216     ASN1 modules to export functions returning ASN1_ITEM pointers
5217     instead of the ASN1_ITEM structures themselves. This adds several
5218     new macros which allow the underlying ASN1 function/structure to
5219     be accessed transparently. As a result code should not use ASN1_ITEM
5220     references directly (such as &X509_it) but instead use the relevant
5221     macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
5222     use of the new ASN1 code on platforms where exporting structures
5223     is problematical (for example in shared libraries) but exporting
5224     functions returning pointers to structures is not.
5225     [Steve Henson]
5226
5227  *) Add support for overriding the generation of SSL/TLS session IDs.
5228     These callbacks can be registered either in an SSL_CTX or per SSL.
5229     The purpose of this is to allow applications to control, if they wish,
5230     the arbitrary values chosen for use as session IDs, particularly as it
5231     can be useful for session caching in multiple-server environments. A
5232     command-line switch for testing this (and any client code that wishes
5233     to use such a feature) has been added to "s_server".
5234     [Geoff Thorpe, Lutz Jaenicke]
5235
5236  *) Modify mkdef.pl to recognise and parse preprocessor conditionals
5237     of the form '#if defined(...) || defined(...) || ...' and
5238     '#if !defined(...) && !defined(...) && ...'.  This also avoids
5239     the growing number of special cases it was previously handling.
5240     [Richard Levitte]
5241
5242  *) Make all configuration macros available for application by making
5243     sure they are available in opensslconf.h, by giving them names starting
5244     with "OPENSSL_" to avoid conflicts with other packages and by making
5245     sure e_os2.h will cover all platform-specific cases together with
5246     opensslconf.h.
5247     Additionally, it is now possible to define configuration/platform-
5248     specific names (called "system identities").  In the C code, these
5249     are prefixed with "OPENSSL_SYSNAME_".  e_os2.h will create another
5250     macro with the name beginning with "OPENSSL_SYS_", which is determined
5251     from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
5252     what is available.
5253     [Richard Levitte]
5254
5255  *) New option -set_serial to 'req' and 'x509' this allows the serial
5256     number to use to be specified on the command line. Previously self
5257     signed certificates were hard coded with serial number 0 and the 
5258     CA options of 'x509' had to use a serial number in a file which was
5259     auto incremented.
5260     [Steve Henson]
5261
5262  *) New options to 'ca' utility to support V2 CRL entry extensions.
5263     Currently CRL reason, invalidity date and hold instruction are
5264     supported. Add new CRL extensions to V3 code and some new objects.
5265     [Steve Henson]
5266
5267  *) New function EVP_CIPHER_CTX_set_padding() this is used to
5268     disable standard block padding (aka PKCS#5 padding) in the EVP
5269     API, which was previously mandatory. This means that the data is
5270     not padded in any way and so the total length much be a multiple
5271     of the block size, otherwise an error occurs.
5272     [Steve Henson]
5273
5274  *) Initial (incomplete) OCSP SSL support.
5275     [Steve Henson]
5276
5277  *) New function OCSP_parse_url(). This splits up a URL into its host,
5278     port and path components: primarily to parse OCSP URLs. New -url
5279     option to ocsp utility.
5280     [Steve Henson]
5281
5282  *) New nonce behavior. The return value of OCSP_check_nonce() now 
5283     reflects the various checks performed. Applications can decide
5284     whether to tolerate certain situations such as an absent nonce
5285     in a response when one was present in a request: the ocsp application
5286     just prints out a warning. New function OCSP_add1_basic_nonce()
5287     this is to allow responders to include a nonce in a response even if
5288     the request is nonce-less.
5289     [Steve Henson]
5290
5291  *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
5292     skipped when using openssl x509 multiple times on a single input file,
5293     e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
5294     [Bodo Moeller]
5295
5296  *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
5297     set string type: to handle setting ASN1_TIME structures. Fix ca
5298     utility to correctly initialize revocation date of CRLs.
5299     [Steve Henson]
5300
5301  *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
5302     the clients preferred ciphersuites and rather use its own preferences.
5303     Should help to work around M$ SGC (Server Gated Cryptography) bug in
5304     Internet Explorer by ensuring unchanged hash method during stepup.
5305     (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
5306     [Lutz Jaenicke]
5307
5308  *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
5309     to aes and add a new 'exist' option to print out symbols that don't
5310     appear to exist.
5311     [Steve Henson]
5312
5313  *) Additional options to ocsp utility to allow flags to be set and
5314     additional certificates supplied.
5315     [Steve Henson]
5316
5317  *) Add the option -VAfile to 'openssl ocsp', so the user can give the
5318     OCSP client a number of certificate to only verify the response
5319     signature against.
5320     [Richard Levitte]
5321
5322  *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
5323     handle the new API. Currently only ECB, CBC modes supported. Add new
5324     AES OIDs.
5325
5326     Add TLS AES ciphersuites as described in RFC3268, "Advanced
5327     Encryption Standard (AES) Ciphersuites for Transport Layer
5328     Security (TLS)".  (In beta versions of OpenSSL 0.9.7, these were
5329     not enabled by default and were not part of the "ALL" ciphersuite
5330     alias because they were not yet official; they could be
5331     explicitly requested by specifying the "AESdraft" ciphersuite
5332     group alias.  In the final release of OpenSSL 0.9.7, the group
5333     alias is called "AES" and is part of "ALL".)
5334     [Ben Laurie, Steve  Henson, Bodo Moeller]
5335
5336  *) New function OCSP_copy_nonce() to copy nonce value (if present) from
5337     request to response.
5338     [Steve Henson]
5339
5340  *) Functions for OCSP responders. OCSP_request_onereq_count(),
5341     OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
5342     extract information from a certificate request. OCSP_response_create()
5343     creates a response and optionally adds a basic response structure.
5344     OCSP_basic_add1_status() adds a complete single response to a basic
5345     response and returns the OCSP_SINGLERESP structure just added (to allow
5346     extensions to be included for example). OCSP_basic_add1_cert() adds a
5347     certificate to a basic response and OCSP_basic_sign() signs a basic
5348     response with various flags. New helper functions ASN1_TIME_check()
5349     (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
5350     (converts ASN1_TIME to GeneralizedTime).
5351     [Steve Henson]
5352
5353  *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
5354     in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
5355     structure from a certificate. X509_pubkey_digest() digests the public_key
5356     contents: this is used in various key identifiers. 
5357     [Steve Henson]
5358
5359  *) Make sk_sort() tolerate a NULL argument.
5360     [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
5361
5362  *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
5363     passed by the function are trusted implicitly. If any of them signed the
5364     response then it is assumed to be valid and is not verified.
5365     [Steve Henson]
5366
5367  *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
5368     to data. This was previously part of the PKCS7 ASN1 code. This
5369     was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
5370     [Steve Henson, reported by Kenneth R. Robinette
5371				<support@securenetterm.com>]
5372
5373  *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
5374     routines: without these tracing memory leaks is very painful.
5375     Fix leaks in PKCS12 and PKCS7 routines.
5376     [Steve Henson]
5377
5378  *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
5379     Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
5380     effectively meant GeneralizedTime would never be used. Now it
5381     is initialised to -1 but X509_time_adj() now has to check the value
5382     and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
5383     V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
5384     [Steve Henson, reported by Kenneth R. Robinette
5385				<support@securenetterm.com>]
5386
5387  *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
5388     result in a zero length in the ASN1_INTEGER structure which was
5389     not consistent with the structure when d2i_ASN1_INTEGER() was used
5390     and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
5391     to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
5392     where it did not print out a minus for negative ASN1_INTEGER.
5393     [Steve Henson]
5394
5395  *) Add summary printout to ocsp utility. The various functions which
5396     convert status values to strings have been renamed to:
5397     OCSP_response_status_str(), OCSP_cert_status_str() and
5398     OCSP_crl_reason_str() and are no longer static. New options
5399     to verify nonce values and to disable verification. OCSP response
5400     printout format cleaned up.
5401     [Steve Henson]
5402
5403  *) Add additional OCSP certificate checks. These are those specified
5404     in RFC2560. This consists of two separate checks: the CA of the
5405     certificate being checked must either be the OCSP signer certificate
5406     or the issuer of the OCSP signer certificate. In the latter case the
5407     OCSP signer certificate must contain the OCSP signing extended key
5408     usage. This check is performed by attempting to match the OCSP
5409     signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
5410     in the OCSP_CERTID structures of the response.
5411     [Steve Henson]
5412
5413  *) Initial OCSP certificate verification added to OCSP_basic_verify()
5414     and related routines. This uses the standard OpenSSL certificate
5415     verify routines to perform initial checks (just CA validity) and
5416     to obtain the certificate chain. Then additional checks will be
5417     performed on the chain. Currently the root CA is checked to see
5418     if it is explicitly trusted for OCSP signing. This is used to set
5419     a root CA as a global signing root: that is any certificate that
5420     chains to that CA is an acceptable OCSP signing certificate.
5421     [Steve Henson]
5422
5423  *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
5424     extensions from a separate configuration file.
5425     As when reading extensions from the main configuration file,
5426     the '-extensions ...' option may be used for specifying the
5427     section to use.
5428     [Massimiliano Pala <madwolf@comune.modena.it>]
5429
5430  *) New OCSP utility. Allows OCSP requests to be generated or
5431     read. The request can be sent to a responder and the output
5432     parsed, outputed or printed in text form. Not complete yet:
5433     still needs to check the OCSP response validity.
5434     [Steve Henson]
5435
5436  *) New subcommands for 'openssl ca':
5437     'openssl ca -status <serial>' prints the status of the cert with
5438     the given serial number (according to the index file).
5439     'openssl ca -updatedb' updates the expiry status of certificates
5440     in the index file.
5441     [Massimiliano Pala <madwolf@comune.modena.it>]
5442
5443  *) New '-newreq-nodes' command option to CA.pl.  This is like
5444     '-newreq', but calls 'openssl req' with the '-nodes' option
5445     so that the resulting key is not encrypted.
5446     [Damien Miller <djm@mindrot.org>]
5447
5448  *) New configuration for the GNU Hurd.
5449     [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
5450
5451  *) Initial code to implement OCSP basic response verify. This
5452     is currently incomplete. Currently just finds the signer's
5453     certificate and verifies the signature on the response.
5454     [Steve Henson]
5455
5456  *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
5457     value of OPENSSLDIR.  This is available via the new '-d' option
5458     to 'openssl version', and is also included in 'openssl version -a'.
5459     [Bodo Moeller]
5460
5461  *) Allowing defining memory allocation callbacks that will be given
5462     file name and line number information in additional arguments
5463     (a const char* and an int).  The basic functionality remains, as
5464     well as the original possibility to just replace malloc(),
5465     realloc() and free() by functions that do not know about these
5466     additional arguments.  To register and find out the current
5467     settings for extended allocation functions, the following
5468     functions are provided:
5469
5470	CRYPTO_set_mem_ex_functions
5471	CRYPTO_set_locked_mem_ex_functions
5472	CRYPTO_get_mem_ex_functions
5473	CRYPTO_get_locked_mem_ex_functions
5474
5475     These work the same way as CRYPTO_set_mem_functions and friends.
5476     CRYPTO_get_[locked_]mem_functions now writes 0 where such an
5477     extended allocation function is enabled.
5478     Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
5479     a conventional allocation function is enabled.
5480     [Richard Levitte, Bodo Moeller]
5481
5482  *) Finish off removing the remaining LHASH function pointer casts.
5483     There should no longer be any prototype-casting required when using
5484     the LHASH abstraction, and any casts that remain are "bugs". See
5485     the callback types and macros at the head of lhash.h for details
5486     (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
5487     [Geoff Thorpe]
5488
5489  *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
5490     If /dev/[u]random devices are not available or do not return enough
5491     entropy, EGD style sockets (served by EGD or PRNGD) will automatically
5492     be queried.
5493     The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
5494     /etc/entropy will be queried once each in this sequence, quering stops
5495     when enough entropy was collected without querying more sockets.
5496     [Lutz Jaenicke]
5497
5498  *) Change the Unix RAND_poll() variant to be able to poll several
5499     random devices, as specified by DEVRANDOM, until a sufficient amount
5500     of data has been collected.   We spend at most 10 ms on each file
5501     (select timeout) and read in non-blocking mode.  DEVRANDOM now
5502     defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
5503     (previously it was just the string "/dev/urandom"), so on typical
5504     platforms the 10 ms delay will never occur.
5505     Also separate out the Unix variant to its own file, rand_unix.c.
5506     For VMS, there's a currently-empty rand_vms.c.
5507     [Richard Levitte]
5508
5509  *) Move OCSP client related routines to ocsp_cl.c. These
5510     provide utility functions which an application needing
5511     to issue a request to an OCSP responder and analyse the
5512     response will typically need: as opposed to those which an
5513     OCSP responder itself would need which will be added later.
5514
5515     OCSP_request_sign() signs an OCSP request with an API similar
5516     to PKCS7_sign(). OCSP_response_status() returns status of OCSP
5517     response. OCSP_response_get1_basic() extracts basic response
5518     from response. OCSP_resp_find_status(): finds and extracts status
5519     information from an OCSP_CERTID structure (which will be created
5520     when the request structure is built). These are built from lower
5521     level functions which work on OCSP_SINGLERESP structures but
5522     wont normally be used unless the application wishes to examine
5523     extensions in the OCSP response for example.
5524
5525     Replace nonce routines with a pair of functions.
5526     OCSP_request_add1_nonce() adds a nonce value and optionally
5527     generates a random value. OCSP_check_nonce() checks the
5528     validity of the nonce in an OCSP response.
5529     [Steve Henson]
5530
5531  *) Change function OCSP_request_add() to OCSP_request_add0_id().
5532     This doesn't copy the supplied OCSP_CERTID and avoids the
5533     need to free up the newly created id. Change return type
5534     to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
5535     This can then be used to add extensions to the request.
5536     Deleted OCSP_request_new(), since most of its functionality
5537     is now in OCSP_REQUEST_new() (and the case insensitive name
5538     clash) apart from the ability to set the request name which
5539     will be added elsewhere.
5540     [Steve Henson]
5541
5542  *) Update OCSP API. Remove obsolete extensions argument from
5543     various functions. Extensions are now handled using the new
5544     OCSP extension code. New simple OCSP HTTP function which 
5545     can be used to send requests and parse the response.
5546     [Steve Henson]
5547
5548  *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
5549     ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
5550     uses the special reorder version of SET OF to sort the attributes
5551     and reorder them to match the encoded order. This resolves a long
5552     standing problem: a verify on a PKCS7 structure just after signing
5553     it used to fail because the attribute order did not match the
5554     encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
5555     it uses the received order. This is necessary to tolerate some broken
5556     software that does not order SET OF. This is handled by encoding
5557     as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
5558     to produce the required SET OF.
5559     [Steve Henson]
5560
5561  *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
5562     OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
5563     files to get correct declarations of the ASN.1 item variables.
5564     [Richard Levitte]
5565
5566  *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
5567     PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
5568     asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
5569     NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
5570     New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
5571     ASN1_ITEM and no wrapper functions.
5572     [Steve Henson]
5573
5574  *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
5575     replace the old function pointer based I/O routines. Change most of
5576     the *_d2i_bio() and *_d2i_fp() functions to use these.
5577     [Steve Henson]
5578
5579  *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
5580     lines, recognice more "algorithms" that can be deselected, and make
5581     it complain about algorithm deselection that isn't recognised.
5582     [Richard Levitte]
5583
5584  *) New ASN1 functions to handle dup, sign, verify, digest, pack and
5585     unpack operations in terms of ASN1_ITEM. Modify existing wrappers
5586     to use new functions. Add NO_ASN1_OLD which can be set to remove
5587     some old style ASN1 functions: this can be used to determine if old
5588     code will still work when these eventually go away.
5589     [Steve Henson]
5590
5591  *) New extension functions for OCSP structures, these follow the
5592     same conventions as certificates and CRLs.
5593     [Steve Henson]
5594
5595  *) New function X509V3_add1_i2d(). This automatically encodes and
5596     adds an extension. Its behaviour can be customised with various
5597     flags to append, replace or delete. Various wrappers added for
5598     certifcates and CRLs.
5599     [Steve Henson]
5600
5601  *) Fix to avoid calling the underlying ASN1 print routine when
5602     an extension cannot be parsed. Correct a typo in the
5603     OCSP_SERVICELOC extension. Tidy up print OCSP format.
5604     [Steve Henson]
5605
5606  *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
5607     entries for variables.
5608     [Steve Henson]
5609
5610  *) Add functionality to apps/openssl.c for detecting locking
5611     problems: As the program is single-threaded, all we have
5612     to do is register a locking callback using an array for
5613     storing which locks are currently held by the program.
5614     [Bodo Moeller]
5615
5616  *) Use a lock around the call to CRYPTO_get_ex_new_index() in
5617     SSL_get_ex_data_X509_STORE_idx(), which is used in
5618     ssl_verify_cert_chain() and thus can be called at any time
5619     during TLS/SSL handshakes so that thread-safety is essential.
5620     Unfortunately, the ex_data design is not at all suited
5621     for multi-threaded use, so it probably should be abolished.
5622     [Bodo Moeller]
5623
5624  *) Added Broadcom "ubsec" ENGINE to OpenSSL.
5625     [Broadcom, tweaked and integrated by Geoff Thorpe]
5626
5627  *) Move common extension printing code to new function
5628     X509V3_print_extensions(). Reorganise OCSP print routines and
5629     implement some needed OCSP ASN1 functions. Add OCSP extensions.
5630     [Steve Henson]
5631
5632  *) New function X509_signature_print() to remove duplication in some
5633     print routines.
5634     [Steve Henson]
5635
5636  *) Add a special meaning when SET OF and SEQUENCE OF flags are both
5637     set (this was treated exactly the same as SET OF previously). This
5638     is used to reorder the STACK representing the structure to match the
5639     encoding. This will be used to get round a problem where a PKCS7
5640     structure which was signed could not be verified because the STACK
5641     order did not reflect the encoded order.
5642     [Steve Henson]
5643
5644  *) Reimplement the OCSP ASN1 module using the new code.
5645     [Steve Henson]
5646
5647  *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
5648     for its ASN1 operations. The old style function pointers still exist
5649     for now but they will eventually go away.
5650     [Steve Henson]
5651
5652  *) Merge in replacement ASN1 code from the ASN1 branch. This almost
5653     completely replaces the old ASN1 functionality with a table driven
5654     encoder and decoder which interprets an ASN1_ITEM structure describing
5655     the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
5656     largely maintained. Almost all of the old asn1_mac.h macro based ASN1
5657     has also been converted to the new form.
5658     [Steve Henson]
5659
5660  *) Change BN_mod_exp_recp so that negative moduli are tolerated
5661     (the sign is ignored).  Similarly, ignore the sign in BN_MONT_CTX_set
5662     so that BN_mod_exp_mont and BN_mod_exp_mont_word work
5663     for negative moduli.
5664     [Bodo Moeller]
5665
5666  *) Fix BN_uadd and BN_usub: Always return non-negative results instead
5667     of not touching the result's sign bit.
5668     [Bodo Moeller]
5669
5670  *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
5671     set.
5672     [Bodo Moeller]
5673
5674  *) Changed the LHASH code to use prototypes for callbacks, and created
5675     macros to declare and implement thin (optionally static) functions
5676     that provide type-safety and avoid function pointer casting for the
5677     type-specific callbacks.
5678     [Geoff Thorpe]
5679
5680  *) Added Kerberos Cipher Suites to be used with TLS, as written in
5681     RFC 2712.
5682     [Veers Staats <staatsvr@asc.hpc.mil>,
5683      Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
5684
5685  *) Reformat the FAQ so the different questions and answers can be divided
5686     in sections depending on the subject.
5687     [Richard Levitte]
5688
5689  *) Have the zlib compression code load ZLIB.DLL dynamically under
5690     Windows.
5691     [Richard Levitte]
5692
5693  *) New function BN_mod_sqrt for computing square roots modulo a prime
5694     (using the probabilistic Tonelli-Shanks algorithm unless
5695     p == 3 (mod 4)  or  p == 5 (mod 8),  which are cases that can
5696     be handled deterministically).
5697     [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
5698
5699  *) Make BN_mod_inverse faster by explicitly handling small quotients
5700     in the Euclid loop. (Speed gain about 20% for small moduli [256 or
5701     512 bits], about 30% for larger ones [1024 or 2048 bits].)
5702     [Bodo Moeller]
5703
5704  *) New function BN_kronecker.
5705     [Bodo Moeller]
5706
5707  *) Fix BN_gcd so that it works on negative inputs; the result is
5708     positive unless both parameters are zero.
5709     Previously something reasonably close to an infinite loop was
5710     possible because numbers could be growing instead of shrinking
5711     in the implementation of Euclid's algorithm.
5712     [Bodo Moeller]
5713
5714  *) Fix BN_is_word() and BN_is_one() macros to take into account the
5715     sign of the number in question.
5716
5717     Fix BN_is_word(a,w) to work correctly for w == 0.
5718
5719     The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
5720     because its test if the absolute value of 'a' equals 'w'.
5721     Note that BN_abs_is_word does *not* handle w == 0 reliably;
5722     it exists mostly for use in the implementations of BN_is_zero(),
5723     BN_is_one(), and BN_is_word().
5724     [Bodo Moeller]
5725
5726  *) New function BN_swap.
5727     [Bodo Moeller]
5728
5729  *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
5730     the exponentiation functions are more likely to produce reasonable
5731     results on negative inputs.
5732     [Bodo Moeller]
5733
5734  *) Change BN_mod_mul so that the result is always non-negative.
5735     Previously, it could be negative if one of the factors was negative;
5736     I don't think anyone really wanted that behaviour.
5737     [Bodo Moeller]
5738
5739  *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
5740     (except for exponentiation, which stays in crypto/bn/bn_exp.c,
5741     and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
5742     and add new functions:
5743
5744          BN_nnmod
5745          BN_mod_sqr
5746          BN_mod_add
5747          BN_mod_add_quick
5748          BN_mod_sub
5749          BN_mod_sub_quick
5750          BN_mod_lshift1
5751          BN_mod_lshift1_quick
5752          BN_mod_lshift
5753          BN_mod_lshift_quick
5754
5755     These functions always generate non-negative results.
5756
5757     BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder  r
5758     such that  |m| < r < 0,  BN_nnmod will output  rem + |m|  instead).
5759
5760     BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
5761     BN_mod_XXX(r, a, [b,] m, ctx), but requires that  a  [and  b]
5762     be reduced modulo  m.
5763     [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
5764
5765#if 0
5766     The following entry accidentily appeared in the CHANGES file
5767     distributed with OpenSSL 0.9.7.  The modifications described in
5768     it do *not* apply to OpenSSL 0.9.7.
5769
5770  *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
5771     was actually never needed) and in BN_mul().  The removal in BN_mul()
5772     required a small change in bn_mul_part_recursive() and the addition
5773     of the functions bn_cmp_part_words(), bn_sub_part_words() and
5774     bn_add_part_words(), which do the same thing as bn_cmp_words(),
5775     bn_sub_words() and bn_add_words() except they take arrays with
5776     differing sizes.
5777     [Richard Levitte]
5778#endif
5779
5780  *) In 'openssl passwd', verify passwords read from the terminal
5781     unless the '-salt' option is used (which usually means that
5782     verification would just waste user's time since the resulting
5783     hash is going to be compared with some given password hash)
5784     or the new '-noverify' option is used.
5785
5786     This is an incompatible change, but it does not affect
5787     non-interactive use of 'openssl passwd' (passwords on the command
5788     line, '-stdin' option, '-in ...' option) and thus should not
5789     cause any problems.
5790     [Bodo Moeller]
5791
5792  *) Remove all references to RSAref, since there's no more need for it.
5793     [Richard Levitte]
5794
5795  *) Make DSO load along a path given through an environment variable
5796     (SHLIB_PATH) with shl_load().
5797     [Richard Levitte]
5798
5799  *) Constify the ENGINE code as a result of BIGNUM constification.
5800     Also constify the RSA code and most things related to it.  In a
5801     few places, most notable in the depth of the ASN.1 code, ugly
5802     casts back to non-const were required (to be solved at a later
5803     time)
5804     [Richard Levitte]
5805
5806  *) Make it so the openssl application has all engines loaded by default.
5807     [Richard Levitte]
5808
5809  *) Constify the BIGNUM routines a little more.
5810     [Richard Levitte]
5811
5812  *) Add the following functions:
5813
5814	ENGINE_load_cswift()
5815	ENGINE_load_chil()
5816	ENGINE_load_atalla()
5817	ENGINE_load_nuron()
5818	ENGINE_load_builtin_engines()
5819
5820     That way, an application can itself choose if external engines that
5821     are built-in in OpenSSL shall ever be used or not.  The benefit is
5822     that applications won't have to be linked with libdl or other dso
5823     libraries unless it's really needed.
5824
5825     Changed 'openssl engine' to load all engines on demand.
5826     Changed the engine header files to avoid the duplication of some
5827     declarations (they differed!).
5828     [Richard Levitte]
5829
5830  *) 'openssl engine' can now list capabilities.
5831     [Richard Levitte]
5832
5833  *) Better error reporting in 'openssl engine'.
5834     [Richard Levitte]
5835
5836  *) Never call load_dh_param(NULL) in s_server.
5837     [Bodo Moeller]
5838
5839  *) Add engine application.  It can currently list engines by name and
5840     identity, and test if they are actually available.
5841     [Richard Levitte]
5842
5843  *) Improve RPM specification file by forcing symbolic linking and making
5844     sure the installed documentation is also owned by root.root.
5845     [Damien Miller <djm@mindrot.org>]
5846
5847  *) Give the OpenSSL applications more possibilities to make use of
5848     keys (public as well as private) handled by engines.
5849     [Richard Levitte]
5850
5851  *) Add OCSP code that comes from CertCo.
5852     [Richard Levitte]
5853
5854  *) Add VMS support for the Rijndael code.
5855     [Richard Levitte]
5856
5857  *) Added untested support for Nuron crypto accelerator.
5858     [Ben Laurie]
5859
5860  *) Add support for external cryptographic devices.  This code was
5861     previously distributed separately as the "engine" branch.
5862     [Geoff Thorpe, Richard Levitte]
5863
5864  *) Rework the filename-translation in the DSO code. It is now possible to
5865     have far greater control over how a "name" is turned into a filename
5866     depending on the operating environment and any oddities about the
5867     different shared library filenames on each system.
5868     [Geoff Thorpe]
5869
5870  *) Support threads on FreeBSD-elf in Configure.
5871     [Richard Levitte]
5872
5873  *) Fix for SHA1 assembly problem with MASM: it produces
5874     warnings about corrupt line number information when assembling
5875     with debugging information. This is caused by the overlapping
5876     of two sections.
5877     [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
5878
5879  *) NCONF changes.
5880     NCONF_get_number() has no error checking at all.  As a replacement,
5881     NCONF_get_number_e() is defined (_e for "error checking") and is
5882     promoted strongly.  The old NCONF_get_number is kept around for
5883     binary backward compatibility.
5884     Make it possible for methods to load from something other than a BIO,
5885     by providing a function pointer that is given a name instead of a BIO.
5886     For example, this could be used to load configuration data from an
5887     LDAP server.
5888     [Richard Levitte]
5889
5890  *) Fix for non blocking accept BIOs. Added new I/O special reason
5891     BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
5892     with non blocking I/O was not possible because no retry code was
5893     implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
5894     this case.
5895     [Steve Henson]
5896
5897  *) Added the beginnings of Rijndael support.
5898     [Ben Laurie]
5899
5900  *) Fix for bug in DirectoryString mask setting. Add support for
5901     X509_NAME_print_ex() in 'req' and X509_print_ex() function
5902     to allow certificate printing to more controllable, additional
5903     'certopt' option to 'x509' to allow new printing options to be
5904     set.
5905     [Steve Henson]
5906
5907  *) Clean old EAY MD5 hack from e_os.h.
5908     [Richard Levitte]
5909
5910 Changes between 0.9.6l and 0.9.6m  [17 Mar 2004]
5911
5912  *) Fix null-pointer assignment in do_change_cipher_spec() revealed
5913     by using the Codenomicon TLS Test Tool (CVE-2004-0079)
5914     [Joe Orton, Steve Henson]
5915
5916 Changes between 0.9.6k and 0.9.6l  [04 Nov 2003]
5917
5918  *) Fix additional bug revealed by the NISCC test suite:
5919
5920     Stop bug triggering large recursion when presented with
5921     certain ASN.1 tags (CVE-2003-0851)
5922     [Steve Henson]
5923
5924 Changes between 0.9.6j and 0.9.6k  [30 Sep 2003]
5925
5926  *) Fix various bugs revealed by running the NISCC test suite:
5927
5928     Stop out of bounds reads in the ASN1 code when presented with
5929     invalid tags (CVE-2003-0543 and CVE-2003-0544).
5930     
5931     If verify callback ignores invalid public key errors don't try to check
5932     certificate signature with the NULL public key.
5933
5934     [Steve Henson]
5935
5936  *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
5937     if the server requested one: as stated in TLS 1.0 and SSL 3.0
5938     specifications.
5939     [Steve Henson]
5940
5941  *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
5942     extra data after the compression methods not only for TLS 1.0
5943     but also for SSL 3.0 (as required by the specification).
5944     [Bodo Moeller; problem pointed out by Matthias Loepfe]
5945
5946  *) Change X509_certificate_type() to mark the key as exported/exportable
5947     when it's 512 *bits* long, not 512 bytes.
5948     [Richard Levitte]
5949
5950 Changes between 0.9.6i and 0.9.6j  [10 Apr 2003]
5951
5952  *) Countermeasure against the Klima-Pokorny-Rosa extension of
5953     Bleichbacher's attack on PKCS #1 v1.5 padding: treat
5954     a protocol version number mismatch like a decryption error
5955     in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
5956     [Bodo Moeller]
5957
5958  *) Turn on RSA blinding by default in the default implementation
5959     to avoid a timing attack. Applications that don't want it can call
5960     RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
5961     They would be ill-advised to do so in most cases.
5962     [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
5963
5964  *) Change RSA blinding code so that it works when the PRNG is not
5965     seeded (in this case, the secret RSA exponent is abused as
5966     an unpredictable seed -- if it is not unpredictable, there
5967     is no point in blinding anyway).  Make RSA blinding thread-safe
5968     by remembering the creator's thread ID in rsa->blinding and
5969     having all other threads use local one-time blinding factors
5970     (this requires more computation than sharing rsa->blinding, but
5971     avoids excessive locking; and if an RSA object is not shared
5972     between threads, blinding will still be very fast).
5973     [Bodo Moeller]
5974
5975 Changes between 0.9.6h and 0.9.6i  [19 Feb 2003]
5976
5977  *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
5978     via timing by performing a MAC computation even if incorrrect
5979     block cipher padding has been found.  This is a countermeasure
5980     against active attacks where the attacker has to distinguish
5981     between bad padding and a MAC verification error. (CVE-2003-0078)
5982
5983     [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
5984     Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
5985     Martin Vuagnoux (EPFL, Ilion)]
5986
5987 Changes between 0.9.6g and 0.9.6h  [5 Dec 2002]
5988
5989  *) New function OPENSSL_cleanse(), which is used to cleanse a section of
5990     memory from it's contents.  This is done with a counter that will
5991     place alternating values in each byte.  This can be used to solve
5992     two issues: 1) the removal of calls to memset() by highly optimizing
5993     compilers, and 2) cleansing with other values than 0, since those can
5994     be read through on certain media, for example a swap space on disk.
5995     [Geoff Thorpe]
5996
5997  *) Bugfix: client side session caching did not work with external caching,
5998     because the session->cipher setting was not restored when reloading
5999     from the external cache. This problem was masked, when
6000     SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
6001     (Found by Steve Haslam <steve@araqnid.ddts.net>.)
6002     [Lutz Jaenicke]
6003
6004  *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
6005     length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
6006     [Zeev Lieber <zeev-l@yahoo.com>]
6007
6008  *) Undo an undocumented change introduced in 0.9.6e which caused
6009     repeated calls to OpenSSL_add_all_ciphers() and 
6010     OpenSSL_add_all_digests() to be ignored, even after calling
6011     EVP_cleanup().
6012     [Richard Levitte]
6013
6014  *) Change the default configuration reader to deal with last line not
6015     being properly terminated.
6016     [Richard Levitte]
6017
6018  *) Change X509_NAME_cmp() so it applies the special rules on handling
6019     DN values that are of type PrintableString, as well as RDNs of type
6020     emailAddress where the value has the type ia5String.
6021     [stefank@valicert.com via Richard Levitte]
6022
6023  *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
6024     the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
6025     doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
6026     the bitwise-OR of the two for use by the majority of applications
6027     wanting this behaviour, and update the docs. The documented
6028     behaviour and actual behaviour were inconsistent and had been
6029     changing anyway, so this is more a bug-fix than a behavioural
6030     change.
6031     [Geoff Thorpe, diagnosed by Nadav Har'El]
6032
6033  *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
6034     (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
6035     [Bodo Moeller]
6036
6037  *) Fix initialization code race conditions in
6038        SSLv23_method(),  SSLv23_client_method(),   SSLv23_server_method(),
6039        SSLv2_method(),   SSLv2_client_method(),    SSLv2_server_method(),
6040        SSLv3_method(),   SSLv3_client_method(),    SSLv3_server_method(),
6041        TLSv1_method(),   TLSv1_client_method(),    TLSv1_server_method(),
6042        ssl2_get_cipher_by_char(),
6043        ssl3_get_cipher_by_char().
6044     [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
6045
6046  *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
6047     the cached sessions are flushed, as the remove_cb() might use ex_data
6048     contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
6049     (see [openssl.org #212]).
6050     [Geoff Thorpe, Lutz Jaenicke]
6051
6052  *) Fix typo in OBJ_txt2obj which incorrectly passed the content
6053     length, instead of the encoding length to d2i_ASN1_OBJECT.
6054     [Steve Henson]
6055
6056 Changes between 0.9.6f and 0.9.6g  [9 Aug 2002]
6057
6058  *) [In 0.9.6g-engine release:]
6059     Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
6060     [Lynn Gazis <lgazis@rainbow.com>]
6061
6062 Changes between 0.9.6e and 0.9.6f  [8 Aug 2002]
6063
6064  *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
6065     and get fix the header length calculation.
6066     [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
6067	Alon Kantor <alonk@checkpoint.com> (and others),
6068	Steve Henson]
6069
6070  *) Use proper error handling instead of 'assertions' in buffer
6071     overflow checks added in 0.9.6e.  This prevents DoS (the
6072     assertions could call abort()).
6073     [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
6074
6075 Changes between 0.9.6d and 0.9.6e  [30 Jul 2002]
6076
6077  *) Add various sanity checks to asn1_get_length() to reject
6078     the ASN1 length bytes if they exceed sizeof(long), will appear
6079     negative or the content length exceeds the length of the
6080     supplied buffer.
6081     [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
6082
6083  *) Fix cipher selection routines: ciphers without encryption had no flags
6084     for the cipher strength set and where therefore not handled correctly
6085     by the selection routines (PR #130).
6086     [Lutz Jaenicke]
6087
6088  *) Fix EVP_dsa_sha macro.
6089     [Nils Larsch]
6090
6091  *) New option
6092          SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
6093     for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
6094     that was added in OpenSSL 0.9.6d.
6095
6096     As the countermeasure turned out to be incompatible with some
6097     broken SSL implementations, the new option is part of SSL_OP_ALL.
6098     SSL_OP_ALL is usually employed when compatibility with weird SSL
6099     implementations is desired (e.g. '-bugs' option to 's_client' and
6100     's_server'), so the new option is automatically set in many
6101     applications.
6102     [Bodo Moeller]
6103
6104  *) Changes in security patch:
6105
6106     Changes marked "(CHATS)" were sponsored by the Defense Advanced
6107     Research Projects Agency (DARPA) and Air Force Research Laboratory,
6108     Air Force Materiel Command, USAF, under agreement number
6109     F30602-01-2-0537.
6110
6111  *) Add various sanity checks to asn1_get_length() to reject
6112     the ASN1 length bytes if they exceed sizeof(long), will appear
6113     negative or the content length exceeds the length of the
6114     supplied buffer. (CVE-2002-0659)
6115     [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
6116
6117  *) Assertions for various potential buffer overflows, not known to
6118     happen in practice.
6119     [Ben Laurie (CHATS)]
6120
6121  *) Various temporary buffers to hold ASCII versions of integers were
6122     too small for 64 bit platforms. (CVE-2002-0655)
6123     [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
6124
6125  *) Remote buffer overflow in SSL3 protocol - an attacker could
6126     supply an oversized session ID to a client. (CVE-2002-0656)
6127     [Ben Laurie (CHATS)]
6128
6129  *) Remote buffer overflow in SSL2 protocol - an attacker could
6130     supply an oversized client master key. (CVE-2002-0656)
6131     [Ben Laurie (CHATS)]
6132
6133 Changes between 0.9.6c and 0.9.6d  [9 May 2002]
6134
6135  *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
6136     encoded as NULL) with id-dsa-with-sha1.
6137     [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
6138
6139  *) Check various X509_...() return values in apps/req.c.
6140     [Nils Larsch <nla@trustcenter.de>]
6141
6142  *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
6143     an end-of-file condition would erronously be flagged, when the CRLF
6144     was just at the end of a processed block. The bug was discovered when
6145     processing data through a buffering memory BIO handing the data to a
6146     BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
6147     <ptsekov@syntrex.com> and Nedelcho Stanev.
6148     [Lutz Jaenicke]
6149
6150  *) Implement a countermeasure against a vulnerability recently found
6151     in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
6152     before application data chunks to avoid the use of known IVs
6153     with data potentially chosen by the attacker.
6154     [Bodo Moeller]
6155
6156  *) Fix length checks in ssl3_get_client_hello().
6157     [Bodo Moeller]
6158
6159  *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
6160     to prevent ssl3_read_internal() from incorrectly assuming that
6161     ssl3_read_bytes() found application data while handshake
6162     processing was enabled when in fact s->s3->in_read_app_data was
6163     merely automatically cleared during the initial handshake.
6164     [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
6165
6166  *) Fix object definitions for Private and Enterprise: they were not
6167     recognized in their shortname (=lowercase) representation. Extend
6168     obj_dat.pl to issue an error when using undefined keywords instead
6169     of silently ignoring the problem (Svenning Sorensen
6170     <sss@sss.dnsalias.net>).
6171     [Lutz Jaenicke]
6172
6173  *) Fix DH_generate_parameters() so that it works for 'non-standard'
6174     generators, i.e. generators other than 2 and 5.  (Previously, the
6175     code did not properly initialise the 'add' and 'rem' values to
6176     BN_generate_prime().)
6177
6178     In the new general case, we do not insist that 'generator' is
6179     actually a primitive root: This requirement is rather pointless;
6180     a generator of the order-q subgroup is just as good, if not
6181     better.
6182     [Bodo Moeller]
6183 
6184  *) Map new X509 verification errors to alerts. Discovered and submitted by
6185     Tom Wu <tom@arcot.com>.
6186     [Lutz Jaenicke]
6187
6188  *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
6189     returning non-zero before the data has been completely received
6190     when using non-blocking I/O.
6191     [Bodo Moeller; problem pointed out by John Hughes]
6192
6193  *) Some of the ciphers missed the strength entry (SSL_LOW etc).
6194     [Ben Laurie, Lutz Jaenicke]
6195
6196  *) Fix bug in SSL_clear(): bad sessions were not removed (found by
6197     Yoram Zahavi <YoramZ@gilian.com>).
6198     [Lutz Jaenicke]
6199
6200  *) Add information about CygWin 1.3 and on, and preserve proper
6201     configuration for the versions before that.
6202     [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
6203
6204  *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
6205     check whether we deal with a copy of a session and do not delete from
6206     the cache in this case. Problem reported by "Izhar Shoshani Levi"
6207     <izhar@checkpoint.com>.
6208     [Lutz Jaenicke]
6209
6210  *) Do not store session data into the internal session cache, if it
6211     is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
6212     flag is set). Proposed by Aslam <aslam@funk.com>.
6213     [Lutz Jaenicke]
6214
6215  *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
6216     value is 0.
6217     [Richard Levitte]
6218
6219  *) [In 0.9.6d-engine release:]
6220     Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
6221     [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
6222
6223  *) Add the configuration target linux-s390x.
6224     [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
6225
6226  *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
6227     ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
6228     variable as an indication that a ClientHello message has been
6229     received.  As the flag value will be lost between multiple
6230     invocations of ssl3_accept when using non-blocking I/O, the
6231     function may not be aware that a handshake has actually taken
6232     place, thus preventing a new session from being added to the
6233     session cache.
6234
6235     To avoid this problem, we now set s->new_session to 2 instead of
6236     using a local variable.
6237     [Lutz Jaenicke, Bodo Moeller]
6238
6239  *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
6240     if the SSL_R_LENGTH_MISMATCH error is detected.
6241     [Geoff Thorpe, Bodo Moeller]
6242
6243  *) New 'shared_ldflag' column in Configure platform table.
6244     [Richard Levitte]
6245
6246  *) Fix EVP_CIPHER_mode macro.
6247     ["Dan S. Camper" <dan@bti.net>]
6248
6249  *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
6250     type, we must throw them away by setting rr->length to 0.
6251     [D P Chang <dpc@qualys.com>]
6252
6253 Changes between 0.9.6b and 0.9.6c  [21 dec 2001]
6254
6255  *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
6256     <Dominikus.Scherkl@biodata.com>.  (The previous implementation
6257     worked incorrectly for those cases where  range = 10..._2  and
6258     3*range  is two bits longer than  range.)
6259     [Bodo Moeller]
6260
6261  *) Only add signing time to PKCS7 structures if it is not already
6262     present.
6263     [Steve Henson]
6264
6265  *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
6266     OBJ_ld_ce should be OBJ_id_ce.
6267     Also some ip-pda OIDs in crypto/objects/objects.txt were
6268     incorrect (cf. RFC 3039).
6269     [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
6270
6271  *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
6272     returns early because it has nothing to do.
6273     [Andy Schneider <andy.schneider@bjss.co.uk>]
6274
6275  *) [In 0.9.6c-engine release:]
6276     Fix mutex callback return values in crypto/engine/hw_ncipher.c.
6277     [Andy Schneider <andy.schneider@bjss.co.uk>]
6278
6279  *) [In 0.9.6c-engine release:]
6280     Add support for Cryptographic Appliance's keyserver technology.
6281     (Use engine 'keyclient')
6282     [Cryptographic Appliances and Geoff Thorpe]
6283
6284  *) Add a configuration entry for OS/390 Unix.  The C compiler 'c89'
6285     is called via tools/c89.sh because arguments have to be
6286     rearranged (all '-L' options must appear before the first object
6287     modules).
6288     [Richard Shapiro <rshapiro@abinitio.com>]
6289
6290  *) [In 0.9.6c-engine release:]
6291     Add support for Broadcom crypto accelerator cards, backported
6292     from 0.9.7.
6293     [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
6294
6295  *) [In 0.9.6c-engine release:]
6296     Add support for SureWare crypto accelerator cards from 
6297     Baltimore Technologies.  (Use engine 'sureware')
6298     [Baltimore Technologies and Mark Cox]
6299
6300  *) [In 0.9.6c-engine release:]
6301     Add support for crypto accelerator cards from Accelerated
6302     Encryption Processing, www.aep.ie.  (Use engine 'aep')
6303     [AEP Inc. and Mark Cox]
6304
6305  *) Add a configuration entry for gcc on UnixWare.
6306     [Gary Benson <gbenson@redhat.com>]
6307
6308  *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
6309     messages are stored in a single piece (fixed-length part and
6310     variable-length part combined) and fix various bugs found on the way.
6311     [Bodo Moeller]
6312
6313  *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
6314     instead.  BIO_gethostbyname() does not know what timeouts are
6315     appropriate, so entries would stay in cache even when they have
6316     become invalid.
6317     [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
6318
6319  *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
6320     faced with a pathologically small ClientHello fragment that does
6321     not contain client_version: Instead of aborting with an error,
6322     simply choose the highest available protocol version (i.e.,
6323     TLS 1.0 unless it is disabled).  In practice, ClientHello
6324     messages are never sent like this, but this change gives us
6325     strictly correct behaviour at least for TLS.
6326     [Bodo Moeller]
6327
6328  *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
6329     never resets s->method to s->ctx->method when called from within
6330     one of the SSL handshake functions.
6331     [Bodo Moeller; problem pointed out by Niko Baric]
6332
6333  *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
6334     (sent using the client's version number) if client_version is
6335     smaller than the protocol version in use.  Also change
6336     ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
6337     the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
6338     the client will at least see that alert.
6339     [Bodo Moeller]
6340
6341  *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
6342     correctly.
6343     [Bodo Moeller]
6344
6345  *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
6346     client receives HelloRequest while in a handshake.
6347     [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
6348
6349  *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
6350     should end in 'break', not 'goto end' which circuments various
6351     cleanups done in state SSL_ST_OK.   But session related stuff
6352     must be disabled for SSL_ST_OK in the case that we just sent a
6353     HelloRequest.
6354
6355     Also avoid some overhead by not calling ssl_init_wbio_buffer()
6356     before just sending a HelloRequest.
6357     [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
6358
6359  *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
6360     reveal whether illegal block cipher padding was found or a MAC
6361     verification error occured.  (Neither SSLerr() codes nor alerts
6362     are directly visible to potential attackers, but the information
6363     may leak via logfiles.)
6364
6365     Similar changes are not required for the SSL 2.0 implementation
6366     because the number of padding bytes is sent in clear for SSL 2.0,
6367     and the extra bytes are just ignored.  However ssl/s2_pkt.c
6368     failed to verify that the purported number of padding bytes is in
6369     the legal range.
6370     [Bodo Moeller]
6371
6372  *) Add OpenUNIX-8 support including shared libraries
6373     (Boyd Lynn Gerber <gerberb@zenez.com>).
6374     [Lutz Jaenicke]
6375
6376  *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
6377     'wristwatch attack' using huge encoding parameters (cf.
6378     James H. Manger's CRYPTO 2001 paper).  Note that the
6379     RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
6380     encoding parameters and hence was not vulnerable.
6381     [Bodo Moeller]
6382
6383  *) BN_sqr() bug fix.
6384     [Ulf M�ller, reported by Jim Ellis <jim.ellis@cavium.com>]
6385
6386  *) Rabin-Miller test analyses assume uniformly distributed witnesses,
6387     so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
6388     followed by modular reduction.
6389     [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
6390
6391  *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
6392     equivalent based on BN_pseudo_rand() instead of BN_rand().
6393     [Bodo Moeller]
6394
6395  *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
6396     This function was broken, as the check for a new client hello message
6397     to handle SGC did not allow these large messages.
6398     (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
6399     [Lutz Jaenicke]
6400
6401  *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
6402     [Lutz Jaenicke]
6403
6404  *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
6405     for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
6406     [Lutz Jaenicke]
6407
6408  *) Rework the configuration and shared library support for Tru64 Unix.
6409     The configuration part makes use of modern compiler features and
6410     still retains old compiler behavior for those that run older versions
6411     of the OS.  The shared library support part includes a variant that
6412     uses the RPATH feature, and is available through the special
6413     configuration target "alpha-cc-rpath", which will never be selected
6414     automatically.
6415     [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
6416
6417  *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
6418     with the same message size as in ssl3_get_certificate_request().
6419     Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
6420     messages might inadvertently be reject as too long.
6421     [Petr Lampa <lampa@fee.vutbr.cz>]
6422
6423  *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
6424     [Andy Polyakov]
6425
6426  *) Modified SSL library such that the verify_callback that has been set
6427     specificly for an SSL object with SSL_set_verify() is actually being
6428     used. Before the change, a verify_callback set with this function was
6429     ignored and the verify_callback() set in the SSL_CTX at the time of
6430     the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
6431     to allow the necessary settings.
6432     [Lutz Jaenicke]
6433
6434  *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
6435     explicitly to NULL, as at least on Solaris 8 this seems not always to be
6436     done automatically (in contradiction to the requirements of the C
6437     standard). This made problems when used from OpenSSH.
6438     [Lutz Jaenicke]
6439
6440  *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
6441     dh->length and always used
6442
6443          BN_rand_range(priv_key, dh->p).
6444
6445     BN_rand_range() is not necessary for Diffie-Hellman, and this
6446     specific range makes Diffie-Hellman unnecessarily inefficient if
6447     dh->length (recommended exponent length) is much smaller than the
6448     length of dh->p.  We could use BN_rand_range() if the order of
6449     the subgroup was stored in the DH structure, but we only have
6450     dh->length.
6451
6452     So switch back to
6453
6454          BN_rand(priv_key, l, ...)
6455
6456     where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
6457     otherwise.
6458     [Bodo Moeller]
6459
6460  *) In
6461
6462          RSA_eay_public_encrypt
6463          RSA_eay_private_decrypt
6464          RSA_eay_private_encrypt (signing)
6465          RSA_eay_public_decrypt (signature verification)
6466
6467     (default implementations for RSA_public_encrypt,
6468     RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
6469     always reject numbers >= n.
6470     [Bodo Moeller]
6471
6472  *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
6473     to synchronize access to 'locking_thread'.  This is necessary on
6474     systems where access to 'locking_thread' (an 'unsigned long'
6475     variable) is not atomic.
6476     [Bodo Moeller]
6477
6478  *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
6479     *before* setting the 'crypto_lock_rand' flag.  The previous code had
6480     a race condition if 0 is a valid thread ID.
6481     [Travis Vitek <vitek@roguewave.com>]
6482
6483  *) Add support for shared libraries under Irix.
6484     [Albert Chin-A-Young <china@thewrittenword.com>]
6485
6486  *) Add configuration option to build on Linux on both big-endian and
6487     little-endian MIPS.
6488     [Ralf Baechle <ralf@uni-koblenz.de>]
6489
6490  *) Add the possibility to create shared libraries on HP-UX.
6491     [Richard Levitte]
6492
6493 Changes between 0.9.6a and 0.9.6b  [9 Jul 2001]
6494
6495  *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
6496     to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
6497     Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
6498     PRNG state recovery was possible based on the output of
6499     one PRNG request appropriately sized to gain knowledge on
6500     'md' followed by enough consecutive 1-byte PRNG requests
6501     to traverse all of 'state'.
6502
6503     1. When updating 'md_local' (the current thread's copy of 'md')
6504        during PRNG output generation, hash all of the previous
6505        'md_local' value, not just the half used for PRNG output.
6506
6507     2. Make the number of bytes from 'state' included into the hash
6508        independent from the number of PRNG bytes requested.
6509
6510     The first measure alone would be sufficient to avoid
6511     Markku-Juhani's attack.  (Actually it had never occurred
6512     to me that the half of 'md_local' used for chaining was the
6513     half from which PRNG output bytes were taken -- I had always
6514     assumed that the secret half would be used.)  The second
6515     measure makes sure that additional data from 'state' is never
6516     mixed into 'md_local' in small portions; this heuristically
6517     further strengthens the PRNG.
6518     [Bodo Moeller]
6519
6520  *) Fix crypto/bn/asm/mips3.s.
6521     [Andy Polyakov]
6522
6523  *) When only the key is given to "enc", the IV is undefined. Print out
6524     an error message in this case.
6525     [Lutz Jaenicke]
6526
6527  *) Handle special case when X509_NAME is empty in X509 printing routines.
6528     [Steve Henson]
6529
6530  *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
6531     positive and less than q.
6532     [Bodo Moeller]
6533
6534  *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
6535     used: it isn't thread safe and the add_lock_callback should handle
6536     that itself.
6537     [Paul Rose <Paul.Rose@bridge.com>]
6538
6539  *) Verify that incoming data obeys the block size in
6540     ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
6541     [Bodo Moeller]
6542
6543  *) Fix OAEP check.
6544     [Ulf M�ller, Bodo M�ller]
6545
6546  *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
6547     RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
6548     when fixing the server behaviour for backwards-compatible 'client
6549     hello' messages.  (Note that the attack is impractical against
6550     SSL 3.0 and TLS 1.0 anyway because length and version checking
6551     means that the probability of guessing a valid ciphertext is
6552     around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
6553     paper.)
6554
6555     Before 0.9.5, the countermeasure (hide the error by generating a
6556     random 'decryption result') did not work properly because
6557     ERR_clear_error() was missing, meaning that SSL_get_error() would
6558     detect the supposedly ignored error.
6559
6560     Both problems are now fixed.
6561     [Bodo Moeller]
6562
6563  *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
6564     (previously it was 1024).
6565     [Bodo Moeller]
6566
6567  *) Fix for compatibility mode trust settings: ignore trust settings
6568     unless some valid trust or reject settings are present.
6569     [Steve Henson]
6570
6571  *) Fix for blowfish EVP: its a variable length cipher.
6572     [Steve Henson]
6573
6574  *) Fix various bugs related to DSA S/MIME verification. Handle missing
6575     parameters in DSA public key structures and return an error in the
6576     DSA routines if parameters are absent.
6577     [Steve Henson]
6578
6579  *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
6580     in the current directory if neither $RANDFILE nor $HOME was set.
6581     RAND_file_name() in 0.9.6a returned NULL in this case.  This has
6582     caused some confusion to Windows users who haven't defined $HOME.
6583     Thus RAND_file_name() is changed again: e_os.h can define a
6584     DEFAULT_HOME, which will be used if $HOME is not set.
6585     For Windows, we use "C:"; on other platforms, we still require
6586     environment variables.
6587
6588  *) Move 'if (!initialized) RAND_poll()' into regions protected by
6589     CRYPTO_LOCK_RAND.  This is not strictly necessary, but avoids
6590     having multiple threads call RAND_poll() concurrently.
6591     [Bodo Moeller]
6592
6593  *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
6594     combination of a flag and a thread ID variable.
6595     Otherwise while one thread is in ssleay_rand_bytes (which sets the
6596     flag), *other* threads can enter ssleay_add_bytes without obeying
6597     the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
6598     that they do not hold after the first thread unsets add_do_not_lock).
6599     [Bodo Moeller]
6600
6601  *) Change bctest again: '-x' expressions are not available in all
6602     versions of 'test'.
6603     [Bodo Moeller]
6604
6605 Changes between 0.9.6 and 0.9.6a  [5 Apr 2001]
6606
6607  *) Fix a couple of memory leaks in PKCS7_dataDecode()
6608     [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
6609
6610  *) Change Configure and Makefiles to provide EXE_EXT, which will contain
6611     the default extension for executables, if any.  Also, make the perl
6612     scripts that use symlink() to test if it really exists and use "cp"
6613     if it doesn't.  All this made OpenSSL compilable and installable in
6614     CygWin.
6615     [Richard Levitte]
6616
6617  *) Fix for asn1_GetSequence() for indefinite length constructed data.
6618     If SEQUENCE is length is indefinite just set c->slen to the total
6619     amount of data available.
6620     [Steve Henson, reported by shige@FreeBSD.org]
6621     [This change does not apply to 0.9.7.]
6622
6623  *) Change bctest to avoid here-documents inside command substitution
6624     (workaround for FreeBSD /bin/sh bug).
6625     For compatibility with Ultrix, avoid shell functions (introduced
6626     in the bctest version that searches along $PATH).
6627     [Bodo Moeller]
6628
6629  *) Rename 'des_encrypt' to 'des_encrypt1'.  This avoids the clashes
6630     with des_encrypt() defined on some operating systems, like Solaris
6631     and UnixWare.
6632     [Richard Levitte]
6633
6634  *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
6635     On the Importance of Eliminating Errors in Cryptographic
6636     Computations, J. Cryptology 14 (2001) 2, 101-119,
6637     http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
6638     [Ulf Moeller]
6639  
6640  *) MIPS assembler BIGNUM division bug fix. 
6641     [Andy Polyakov]
6642
6643  *) Disabled incorrect Alpha assembler code.
6644     [Richard Levitte]
6645
6646  *) Fix PKCS#7 decode routines so they correctly update the length
6647     after reading an EOC for the EXPLICIT tag.
6648     [Steve Henson]
6649     [This change does not apply to 0.9.7.]
6650
6651  *) Fix bug in PKCS#12 key generation routines. This was triggered
6652     if a 3DES key was generated with a 0 initial byte. Include
6653     PKCS12_BROKEN_KEYGEN compilation option to retain the old
6654     (but broken) behaviour.
6655     [Steve Henson]
6656
6657  *) Enhance bctest to search for a working bc along $PATH and print
6658     it when found.
6659     [Tim Rice <tim@multitalents.net> via Richard Levitte]
6660
6661  *) Fix memory leaks in err.c: free err_data string if necessary;
6662     don't write to the wrong index in ERR_set_error_data.
6663     [Bodo Moeller]
6664
6665  *) Implement ssl23_peek (analogous to ssl23_read), which previously
6666     did not exist.
6667     [Bodo Moeller]
6668
6669  *) Replace rdtsc with _emit statements for VC++ version 5.
6670     [Jeremy Cooper <jeremy@baymoo.org>]
6671
6672  *) Make it possible to reuse SSLv2 sessions.
6673     [Richard Levitte]
6674
6675  *) In copy_email() check for >= 0 as a return value for
6676     X509_NAME_get_index_by_NID() since 0 is a valid index.
6677     [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
6678
6679  *) Avoid coredump with unsupported or invalid public keys by checking if
6680     X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
6681     PKCS7_verify() fails with non detached data.
6682     [Steve Henson]
6683
6684  *) Don't use getenv in library functions when run as setuid/setgid.
6685     New function OPENSSL_issetugid().
6686     [Ulf Moeller]
6687
6688  *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
6689     due to incorrect handling of multi-threading:
6690
6691     1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
6692
6693     2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
6694
6695     3. Count how many times MemCheck_off() has been called so that
6696        nested use can be treated correctly.  This also avoids 
6697        inband-signalling in the previous code (which relied on the
6698        assumption that thread ID 0 is impossible).
6699     [Bodo Moeller]
6700
6701  *) Add "-rand" option also to s_client and s_server.
6702     [Lutz Jaenicke]
6703
6704  *) Fix CPU detection on Irix 6.x.
6705     [Kurt Hockenbury <khockenb@stevens-tech.edu> and
6706      "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
6707
6708  *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
6709     was empty.
6710     [Steve Henson]
6711     [This change does not apply to 0.9.7.]
6712
6713  *) Use the cached encoding of an X509_NAME structure rather than
6714     copying it. This is apparently the reason for the libsafe "errors"
6715     but the code is actually correct.
6716     [Steve Henson]
6717
6718  *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
6719     Bleichenbacher's DSA attack.
6720     Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
6721     to be set and top=0 forces the highest bit to be set; top=-1 is new
6722     and leaves the highest bit random.
6723     [Ulf Moeller, Bodo Moeller]
6724
6725  *) In the NCONF_...-based implementations for CONF_... queries
6726     (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
6727     a temporary CONF structure with the data component set to NULL
6728     (which gives segmentation faults in lh_retrieve).
6729     Instead, use NULL for the CONF pointer in CONF_get_string and
6730     CONF_get_number (which may use environment variables) and directly
6731     return NULL from CONF_get_section.
6732     [Bodo Moeller]
6733
6734  *) Fix potential buffer overrun for EBCDIC.
6735     [Ulf Moeller]
6736
6737  *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
6738     keyUsage if basicConstraints absent for a CA.
6739     [Steve Henson]
6740
6741  *) Make SMIME_write_PKCS7() write mail header values with a format that
6742     is more generally accepted (no spaces before the semicolon), since
6743     some programs can't parse those values properly otherwise.  Also make
6744     sure BIO's that break lines after each write do not create invalid
6745     headers.
6746     [Richard Levitte]
6747
6748  *) Make the CRL encoding routines work with empty SEQUENCE OF. The
6749     macros previously used would not encode an empty SEQUENCE OF
6750     and break the signature.
6751     [Steve Henson]
6752     [This change does not apply to 0.9.7.]
6753
6754  *) Zero the premaster secret after deriving the master secret in
6755     DH ciphersuites.
6756     [Steve Henson]
6757
6758  *) Add some EVP_add_digest_alias registrations (as found in
6759     OpenSSL_add_all_digests()) to SSL_library_init()
6760     aka OpenSSL_add_ssl_algorithms().  This provides improved
6761     compatibility with peers using X.509 certificates
6762     with unconventional AlgorithmIdentifier OIDs.
6763     [Bodo Moeller]
6764
6765  *) Fix for Irix with NO_ASM.
6766     ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
6767
6768  *) ./config script fixes.
6769     [Ulf Moeller, Richard Levitte]
6770
6771  *) Fix 'openssl passwd -1'.
6772     [Bodo Moeller]
6773
6774  *) Change PKCS12_key_gen_asc() so it can cope with non null
6775     terminated strings whose length is passed in the passlen
6776     parameter, for example from PEM callbacks. This was done
6777     by adding an extra length parameter to asc2uni().
6778     [Steve Henson, reported by <oddissey@samsung.co.kr>]
6779
6780  *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
6781     call failed, free the DSA structure.
6782     [Bodo Moeller]
6783
6784  *) Fix to uni2asc() to cope with zero length Unicode strings.
6785     These are present in some PKCS#12 files.
6786     [Steve Henson]
6787
6788  *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
6789     Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
6790     when writing a 32767 byte record.
6791     [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
6792
6793  *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
6794     obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
6795
6796     (RSA objects have a reference count access to which is protected
6797     by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
6798     so they are meant to be shared between threads.)
6799     [Bodo Moeller, Geoff Thorpe; original patch submitted by
6800     "Reddie, Steven" <Steven.Reddie@ca.com>]
6801
6802  *) Fix a deadlock in CRYPTO_mem_leaks().
6803     [Bodo Moeller]
6804
6805  *) Use better test patterns in bntest.
6806     [Ulf M�ller]
6807
6808  *) rand_win.c fix for Borland C.
6809     [Ulf M�ller]
6810 
6811  *) BN_rshift bugfix for n == 0.
6812     [Bodo Moeller]
6813
6814  *) Add a 'bctest' script that checks for some known 'bc' bugs
6815     so that 'make test' does not abort just because 'bc' is broken.
6816     [Bodo Moeller]
6817
6818  *) Store verify_result within SSL_SESSION also for client side to
6819     avoid potential security hole. (Re-used sessions on the client side
6820     always resulted in verify_result==X509_V_OK, not using the original
6821     result of the server certificate verification.)
6822     [Lutz Jaenicke]
6823
6824  *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
6825     SSL3_RT_APPLICATION_DATA, return 0.
6826     Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
6827     [Bodo Moeller]
6828
6829  *) Fix SSL_peek:
6830     Both ssl2_peek and ssl3_peek, which were totally broken in earlier
6831     releases, have been re-implemented by renaming the previous
6832     implementations of ssl2_read and ssl3_read to ssl2_read_internal
6833     and ssl3_read_internal, respectively, and adding 'peek' parameters
6834     to them.  The new ssl[23]_{read,peek} functions are calls to
6835     ssl[23]_read_internal with the 'peek' flag set appropriately.
6836     A 'peek' parameter has also been added to ssl3_read_bytes, which
6837     does the actual work for ssl3_read_internal.
6838     [Bodo Moeller]
6839
6840  *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
6841     the method-specific "init()" handler. Also clean up ex_data after
6842     calling the method-specific "finish()" handler. Previously, this was
6843     happening the other way round.
6844     [Geoff Thorpe]
6845
6846  *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
6847     The previous value, 12, was not always sufficient for BN_mod_exp().
6848     [Bodo Moeller]
6849
6850  *) Make sure that shared libraries get the internal name engine with
6851     the full version number and not just 0.  This should mark the
6852     shared libraries as not backward compatible.  Of course, this should
6853     be changed again when we can guarantee backward binary compatibility.
6854     [Richard Levitte]
6855
6856  *) Fix typo in get_cert_by_subject() in by_dir.c
6857     [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
6858
6859  *) Rework the system to generate shared libraries:
6860
6861     - Make note of the expected extension for the shared libraries and
6862       if there is a need for symbolic links from for example libcrypto.so.0
6863       to libcrypto.so.0.9.7.  There is extended info in Configure for
6864       that.
6865
6866     - Make as few rebuilds of the shared libraries as possible.
6867
6868     - Still avoid linking the OpenSSL programs with the shared libraries.
6869
6870     - When installing, install the shared libraries separately from the
6871       static ones.
6872     [Richard Levitte]
6873
6874  *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
6875
6876     Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
6877     and not in SSL_clear because the latter is also used by the
6878     accept/connect functions; previously, the settings made by
6879     SSL_set_read_ahead would be lost during the handshake.
6880     [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]     
6881
6882  *) Correct util/mkdef.pl to be selective about disabled algorithms.
6883     Previously, it would create entries for disableed algorithms no
6884     matter what.
6885     [Richard Levitte]
6886
6887  *) Added several new manual pages for SSL_* function.
6888     [Lutz Jaenicke]
6889
6890 Changes between 0.9.5a and 0.9.6  [24 Sep 2000]
6891
6892  *) In ssl23_get_client_hello, generate an error message when faced
6893     with an initial SSL 3.0/TLS record that is too small to contain the
6894     first two bytes of the ClientHello message, i.e. client_version.
6895     (Note that this is a pathologic case that probably has never happened
6896     in real life.)  The previous approach was to use the version number
6897     from the record header as a substitute; but our protocol choice
6898     should not depend on that one because it is not authenticated
6899     by the Finished messages.
6900     [Bodo Moeller]
6901
6902  *) More robust randomness gathering functions for Windows.
6903     [Jeffrey Altman <jaltman@columbia.edu>]
6904
6905  *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
6906     not set then we don't setup the error code for issuer check errors
6907     to avoid possibly overwriting other errors which the callback does
6908     handle. If an application does set the flag then we assume it knows
6909     what it is doing and can handle the new informational codes
6910     appropriately.
6911     [Steve Henson]
6912
6913  *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
6914     a general "ANY" type, as such it should be able to decode anything
6915     including tagged types. However it didn't check the class so it would
6916     wrongly interpret tagged types in the same way as their universal
6917     counterpart and unknown types were just rejected. Changed so that the
6918     tagged and unknown types are handled in the same way as a SEQUENCE:
6919     that is the encoding is stored intact. There is also a new type
6920     "V_ASN1_OTHER" which is used when the class is not universal, in this
6921     case we have no idea what the actual type is so we just lump them all
6922     together.
6923     [Steve Henson]
6924
6925  *) On VMS, stdout may very well lead to a file that is written to
6926     in a record-oriented fashion.  That means that every write() will
6927     write a separate record, which will be read separately by the
6928     programs trying to read from it.  This can be very confusing.
6929
6930     The solution is to put a BIO filter in the way that will buffer
6931     text until a linefeed is reached, and then write everything a
6932     line at a time, so every record written will be an actual line,
6933     not chunks of lines and not (usually doesn't happen, but I've
6934     seen it once) several lines in one record.  BIO_f_linebuffer() is
6935     the answer.
6936
6937     Currently, it's a VMS-only method, because that's where it has
6938     been tested well enough.
6939     [Richard Levitte]
6940
6941  *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
6942     it can return incorrect results.
6943     (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
6944     but it was in 0.9.6-beta[12].)
6945     [Bodo Moeller]
6946
6947  *) Disable the check for content being present when verifying detached
6948     signatures in pk7_smime.c. Some versions of Netscape (wrongly)
6949     include zero length content when signing messages.
6950     [Steve Henson]
6951
6952  *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
6953     BIO_ctrl (for BIO pairs).
6954     [Bodo M�ller]
6955
6956  *) Add DSO method for VMS.
6957     [Richard Levitte]
6958
6959  *) Bug fix: Montgomery multiplication could produce results with the
6960     wrong sign.
6961     [Ulf M�ller]
6962
6963  *) Add RPM specification openssl.spec and modify it to build three
6964     packages.  The default package contains applications, application
6965     documentation and run-time libraries.  The devel package contains
6966     include files, static libraries and function documentation.  The
6967     doc package contains the contents of the doc directory.  The original
6968     openssl.spec was provided by Damien Miller <djm@mindrot.org>.
6969     [Richard Levitte]
6970     
6971  *) Add a large number of documentation files for many SSL routines.
6972     [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
6973
6974  *) Add a configuration entry for Sony News 4.
6975     [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
6976
6977  *) Don't set the two most significant bits to one when generating a
6978     random number < q in the DSA library.
6979     [Ulf M�ller]
6980
6981  *) New SSL API mode 'SSL_MODE_AUTO_RETRY'.  This disables the default
6982     behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
6983     the underlying transport is blocking) if a handshake took place.
6984     (The default behaviour is needed by applications such as s_client
6985     and s_server that use select() to determine when to use SSL_read;
6986     but for applications that know in advance when to expect data, it
6987     just makes things more complicated.)
6988     [Bodo Moeller]
6989
6990  *) Add RAND_egd_bytes(), which gives control over the number of bytes read
6991     from EGD.
6992     [Ben Laurie]
6993
6994  *) Add a few more EBCDIC conditionals that make `req' and `x509'
6995     work better on such systems.
6996     [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
6997
6998  *) Add two demo programs for PKCS12_parse() and PKCS12_create().
6999     Update PKCS12_parse() so it copies the friendlyName and the
7000     keyid to the certificates aux info.
7001     [Steve Henson]
7002
7003  *) Fix bug in PKCS7_verify() which caused an infinite loop
7004     if there was more than one signature.
7005     [Sven Uszpelkat <su@celocom.de>]
7006
7007  *) Major change in util/mkdef.pl to include extra information
7008     about each symbol, as well as presentig variables as well
7009     as functions.  This change means that there's n more need
7010     to rebuild the .num files when some algorithms are excluded.
7011     [Richard Levitte]
7012
7013  *) Allow the verify time to be set by an application,
7014     rather than always using the current time.
7015     [Steve Henson]
7016  
7017  *) Phase 2 verify code reorganisation. The certificate
7018     verify code now looks up an issuer certificate by a
7019     number of criteria: subject name, authority key id
7020     and key usage. It also verifies self signed certificates
7021     by the same criteria. The main comparison function is
7022     X509_check_issued() which performs these checks.
7023 
7024     Lot of changes were necessary in order to support this
7025     without completely rewriting the lookup code.
7026 
7027     Authority and subject key identifier are now cached.
7028 
7029     The LHASH 'certs' is X509_STORE has now been replaced
7030     by a STACK_OF(X509_OBJECT). This is mainly because an
7031     LHASH can't store or retrieve multiple objects with
7032     the same hash value.
7033
7034     As a result various functions (which were all internal
7035     use only) have changed to handle the new X509_STORE
7036     structure. This will break anything that messed round
7037     with X509_STORE internally.
7038 
7039     The functions X509_STORE_add_cert() now checks for an
7040     exact match, rather than just subject name.
7041 
7042     The X509_STORE API doesn't directly support the retrieval
7043     of multiple certificates matching a given criteria, however
7044     this can be worked round by performing a lookup first
7045     (which will fill the cache with candidate certificates)
7046     and then examining the cache for matches. This is probably
7047     the best we can do without throwing out X509_LOOKUP
7048     entirely (maybe later...).
7049 
7050     The X509_VERIFY_CTX structure has been enhanced considerably.
7051 
7052     All certificate lookup operations now go via a get_issuer()
7053     callback. Although this currently uses an X509_STORE it
7054     can be replaced by custom lookups. This is a simple way
7055     to bypass the X509_STORE hackery necessary to make this
7056     work and makes it possible to use more efficient techniques
7057     in future. A very simple version which uses a simple
7058     STACK for its trusted certificate store is also provided
7059     using X509_STORE_CTX_trusted_stack().
7060 
7061     The verify_cb() and verify() callbacks now have equivalents
7062     in the X509_STORE_CTX structure.
7063 
7064     X509_STORE_CTX also has a 'flags' field which can be used
7065     to customise the verify behaviour.
7066     [Steve Henson]
7067 
7068  *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 
7069     excludes S/MIME capabilities.
7070     [Steve Henson]
7071
7072  *) When a certificate request is read in keep a copy of the
7073     original encoding of the signed data and use it when outputing
7074     again. Signatures then use the original encoding rather than
7075     a decoded, encoded version which may cause problems if the
7076     request is improperly encoded.
7077     [Steve Henson]
7078
7079  *) For consistency with other BIO_puts implementations, call
7080     buffer_write(b, ...) directly in buffer_puts instead of calling
7081     BIO_write(b, ...).
7082
7083     In BIO_puts, increment b->num_write as in BIO_write.
7084     [Peter.Sylvester@EdelWeb.fr]
7085
7086  *) Fix BN_mul_word for the case where the word is 0. (We have to use
7087     BN_zero, we may not return a BIGNUM with an array consisting of
7088     words set to zero.)
7089     [Bodo Moeller]
7090
7091  *) Avoid calling abort() from within the library when problems are
7092     detected, except if preprocessor symbols have been defined
7093     (such as REF_CHECK, BN_DEBUG etc.).
7094     [Bodo Moeller]
7095
7096  *) New openssl application 'rsautl'. This utility can be
7097     used for low level RSA operations. DER public key
7098     BIO/fp routines also added.
7099     [Steve Henson]
7100
7101  *) New Configure entry and patches for compiling on QNX 4.
7102     [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
7103
7104  *) A demo state-machine implementation was sponsored by
7105     Nuron (http://www.nuron.com/) and is now available in
7106     demos/state_machine.
7107     [Ben Laurie]
7108
7109  *) New options added to the 'dgst' utility for signature
7110     generation and verification.
7111     [Steve Henson]
7112
7113  *) Unrecognized PKCS#7 content types are now handled via a
7114     catch all ASN1_TYPE structure. This allows unsupported
7115     types to be stored as a "blob" and an application can
7116     encode and decode it manually.
7117     [Steve Henson]
7118
7119  *) Fix various signed/unsigned issues to make a_strex.c
7120     compile under VC++.
7121     [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
7122
7123  *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
7124     length if passed a buffer. ASN1_INTEGER_to_BN failed
7125     if passed a NULL BN and its argument was negative.
7126     [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
7127
7128  *) Modification to PKCS#7 encoding routines to output definite
7129     length encoding. Since currently the whole structures are in
7130     memory there's not real point in using indefinite length 
7131     constructed encoding. However if OpenSSL is compiled with
7132     the flag PKCS7_INDEFINITE_ENCODING the old form is used.
7133     [Steve Henson]
7134
7135  *) Added BIO_vprintf() and BIO_vsnprintf().
7136     [Richard Levitte]
7137
7138  *) Added more prefixes to parse for in the the strings written
7139     through a logging bio, to cover all the levels that are available
7140     through syslog.  The prefixes are now:
7141
7142	PANIC, EMERG, EMR	=>	LOG_EMERG
7143	ALERT, ALR		=>	LOG_ALERT
7144	CRIT, CRI		=>	LOG_CRIT
7145	ERROR, ERR		=>	LOG_ERR
7146	WARNING, WARN, WAR	=>	LOG_WARNING
7147	NOTICE, NOTE, NOT	=>	LOG_NOTICE
7148	INFO, INF		=>	LOG_INFO
7149	DEBUG, DBG		=>	LOG_DEBUG
7150
7151     and as before, if none of those prefixes are present at the
7152     beginning of the string, LOG_ERR is chosen.
7153
7154     On Win32, the LOG_* levels are mapped according to this:
7155
7156	LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR	=> EVENTLOG_ERROR_TYPE
7157	LOG_WARNING				=> EVENTLOG_WARNING_TYPE
7158	LOG_NOTICE, LOG_INFO, LOG_DEBUG		=> EVENTLOG_INFORMATION_TYPE
7159
7160     [Richard Levitte]
7161
7162  *) Made it possible to reconfigure with just the configuration
7163     argument "reconf" or "reconfigure".  The command line arguments
7164     are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
7165     and are retrieved from there when reconfiguring.
7166     [Richard Levitte]
7167
7168  *) MD4 implemented.
7169     [Assar Westerlund <assar@sics.se>, Richard Levitte]
7170
7171  *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
7172     [Richard Levitte]
7173
7174  *) The obj_dat.pl script was messing up the sorting of object
7175     names. The reason was that it compared the quoted version
7176     of strings as a result "OCSP" > "OCSP Signing" because
7177     " > SPACE. Changed script to store unquoted versions of
7178     names and add quotes on output. It was also omitting some
7179     names from the lookup table if they were given a default
7180     value (that is if SN is missing it is given the same
7181     value as LN and vice versa), these are now added on the
7182     grounds that if an object has a name we should be able to
7183     look it up. Finally added warning output when duplicate
7184     short or long names are found.
7185     [Steve Henson]
7186
7187  *) Changes needed for Tandem NSK.
7188     [Scott Uroff <scott@xypro.com>]
7189
7190  *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
7191     RSA_padding_check_SSLv23(), special padding was never detected
7192     and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
7193     version rollback attacks was not effective.
7194
7195     In s23_clnt.c, don't use special rollback-attack detection padding
7196     (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
7197     client; similarly, in s23_srvr.c, don't do the rollback check if
7198     SSL 2.0 is the only protocol enabled in the server.
7199     [Bodo Moeller]
7200
7201  *) Make it possible to get hexdumps of unprintable data with 'openssl
7202     asn1parse'.  By implication, the functions ASN1_parse_dump() and
7203     BIO_dump_indent() are added.
7204     [Richard Levitte]
7205
7206  *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
7207     these print out strings and name structures based on various
7208     flags including RFC2253 support and proper handling of
7209     multibyte characters. Added options to the 'x509' utility 
7210     to allow the various flags to be set.
7211     [Steve Henson]
7212
7213  *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
7214     Also change the functions X509_cmp_current_time() and
7215     X509_gmtime_adj() work with an ASN1_TIME structure,
7216     this will enable certificates using GeneralizedTime in validity
7217     dates to be checked.
7218     [Steve Henson]
7219
7220  *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
7221     negative public key encodings) on by default,
7222     NO_NEG_PUBKEY_BUG can be set to disable it.
7223     [Steve Henson]
7224
7225  *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
7226     content octets. An i2c_ASN1_OBJECT is unnecessary because
7227     the encoding can be trivially obtained from the structure.
7228     [Steve Henson]
7229
7230  *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
7231     not read locks (CRYPTO_r_[un]lock).
7232     [Bodo Moeller]
7233
7234  *) A first attempt at creating official support for shared
7235     libraries through configuration.  I've kept it so the
7236     default is static libraries only, and the OpenSSL programs
7237     are always statically linked for now, but there are
7238     preparations for dynamic linking in place.
7239     This has been tested on Linux and Tru64.
7240     [Richard Levitte]
7241
7242  *) Randomness polling function for Win9x, as described in:
7243     Peter Gutmann, Software Generation of Practically Strong
7244     Random Numbers.
7245     [Ulf M�ller]
7246
7247  *) Fix so PRNG is seeded in req if using an already existing
7248     DSA key.
7249     [Steve Henson]
7250
7251  *) New options to smime application. -inform and -outform
7252     allow alternative formats for the S/MIME message including
7253     PEM and DER. The -content option allows the content to be
7254     specified separately. This should allow things like Netscape
7255     form signing output easier to verify.
7256     [Steve Henson]
7257
7258  *) Fix the ASN1 encoding of tags using the 'long form'.
7259     [Steve Henson]
7260
7261  *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
7262     STRING types. These convert content octets to and from the
7263     underlying type. The actual tag and length octets are
7264     already assumed to have been read in and checked. These
7265     are needed because all other string types have virtually
7266     identical handling apart from the tag. By having versions
7267     of the ASN1 functions that just operate on content octets
7268     IMPLICIT tagging can be handled properly. It also allows
7269     the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
7270     and ASN1_INTEGER are identical apart from the tag.
7271     [Steve Henson]
7272
7273  *) Change the handling of OID objects as follows:
7274
7275     - New object identifiers are inserted in objects.txt, following
7276       the syntax given in objects.README.
7277     - objects.pl is used to process obj_mac.num and create a new
7278       obj_mac.h.
7279     - obj_dat.pl is used to create a new obj_dat.h, using the data in
7280       obj_mac.h.
7281
7282     This is currently kind of a hack, and the perl code in objects.pl
7283     isn't very elegant, but it works as I intended.  The simplest way
7284     to check that it worked correctly is to look in obj_dat.h and
7285     check the array nid_objs and make sure the objects haven't moved
7286     around (this is important!).  Additions are OK, as well as
7287     consistent name changes. 
7288     [Richard Levitte]
7289
7290  *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
7291     [Bodo Moeller]
7292
7293  *) Addition of the command line parameter '-rand file' to 'openssl req'.
7294     The given file adds to whatever has already been seeded into the
7295     random pool through the RANDFILE configuration file option or
7296     environment variable, or the default random state file.
7297     [Richard Levitte]
7298
7299  *) mkstack.pl now sorts each macro group into lexical order.
7300     Previously the output order depended on the order the files
7301     appeared in the directory, resulting in needless rewriting
7302     of safestack.h .
7303     [Steve Henson]
7304
7305  *) Patches to make OpenSSL compile under Win32 again. Mostly
7306     work arounds for the VC++ problem that it treats func() as
7307     func(void). Also stripped out the parts of mkdef.pl that
7308     added extra typesafe functions: these no longer exist.
7309     [Steve Henson]
7310
7311  *) Reorganisation of the stack code. The macros are now all 
7312     collected in safestack.h . Each macro is defined in terms of
7313     a "stack macro" of the form SKM_<name>(type, a, b). The 
7314     DEBUG_SAFESTACK is now handled in terms of function casts,
7315     this has the advantage of retaining type safety without the
7316     use of additional functions. If DEBUG_SAFESTACK is not defined
7317     then the non typesafe macros are used instead. Also modified the
7318     mkstack.pl script to handle the new form. Needs testing to see
7319     if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
7320     the default if no major problems. Similar behaviour for ASN1_SET_OF
7321     and PKCS12_STACK_OF.
7322     [Steve Henson]
7323
7324  *) When some versions of IIS use the 'NET' form of private key the
7325     key derivation algorithm is different. Normally MD5(password) is
7326     used as a 128 bit RC4 key. In the modified case
7327     MD5(MD5(password) + "SGCKEYSALT")  is used insted. Added some
7328     new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
7329     as the old Netscape_RSA functions except they have an additional
7330     'sgckey' parameter which uses the modified algorithm. Also added
7331     an -sgckey command line option to the rsa utility. Thanks to 
7332     Adrian Peck <bertie@ncipher.com> for posting details of the modified
7333     algorithm to openssl-dev.
7334     [Steve Henson]
7335
7336  *) The evp_local.h macros were using 'c.##kname' which resulted in
7337     invalid expansion on some systems (SCO 5.0.5 for example).
7338     Corrected to 'c.kname'.
7339     [Phillip Porch <root@theporch.com>]
7340
7341  *) New X509_get1_email() and X509_REQ_get1_email() functions that return
7342     a STACK of email addresses from a certificate or request, these look
7343     in the subject name and the subject alternative name extensions and 
7344     omit any duplicate addresses.
7345     [Steve Henson]
7346
7347  *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
7348     This makes DSA verification about 2 % faster.
7349     [Bodo Moeller]
7350
7351  *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
7352     (meaning that now 2^5 values will be precomputed, which is only 4 KB
7353     plus overhead for 1024 bit moduli).
7354     This makes exponentiations about 0.5 % faster for 1024 bit
7355     exponents (as measured by "openssl speed rsa2048").
7356     [Bodo Moeller]
7357
7358  *) Rename memory handling macros to avoid conflicts with other
7359     software:
7360          Malloc         =>  OPENSSL_malloc
7361          Malloc_locked  =>  OPENSSL_malloc_locked
7362          Realloc        =>  OPENSSL_realloc
7363          Free           =>  OPENSSL_free
7364     [Richard Levitte]
7365
7366  *) New function BN_mod_exp_mont_word for small bases (roughly 15%
7367     faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
7368     [Bodo Moeller]
7369
7370  *) CygWin32 support.
7371     [John Jarvie <jjarvie@newsguy.com>]
7372
7373  *) The type-safe stack code has been rejigged. It is now only compiled
7374     in when OpenSSL is configured with the DEBUG_SAFESTACK option and
7375     by default all type-specific stack functions are "#define"d back to
7376     standard stack functions. This results in more streamlined output
7377     but retains the type-safety checking possibilities of the original
7378     approach.
7379     [Geoff Thorpe]
7380
7381  *) The STACK code has been cleaned up, and certain type declarations
7382     that didn't make a lot of sense have been brought in line. This has
7383     also involved a cleanup of sorts in safestack.h to more correctly
7384     map type-safe stack functions onto their plain stack counterparts.
7385     This work has also resulted in a variety of "const"ifications of
7386     lots of the code, especially "_cmp" operations which should normally
7387     be prototyped with "const" parameters anyway.
7388     [Geoff Thorpe]
7389
7390  *) When generating bytes for the first time in md_rand.c, 'stir the pool'
7391     by seeding with STATE_SIZE dummy bytes (with zero entropy count).
7392     (The PRNG state consists of two parts, the large pool 'state' and 'md',
7393     where all of 'md' is used each time the PRNG is used, but 'state'
7394     is used only indexed by a cyclic counter. As entropy may not be
7395     well distributed from the beginning, 'md' is important as a
7396     chaining variable. However, the output function chains only half
7397     of 'md', i.e. 80 bits.  ssleay_rand_add, on the other hand, chains
7398     all of 'md', and seeding with STATE_SIZE dummy bytes will result
7399     in all of 'state' being rewritten, with the new values depending
7400     on virtually all of 'md'.  This overcomes the 80 bit limitation.)
7401     [Bodo Moeller]
7402
7403  *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
7404     the handshake is continued after ssl_verify_cert_chain();
7405     otherwise, if SSL_VERIFY_NONE is set, remaining error codes
7406     can lead to 'unexplainable' connection aborts later.
7407     [Bodo Moeller; problem tracked down by Lutz Jaenicke]
7408
7409  *) Major EVP API cipher revision.
7410     Add hooks for extra EVP features. This allows various cipher
7411     parameters to be set in the EVP interface. Support added for variable
7412     key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
7413     setting of RC2 and RC5 parameters.
7414
7415     Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
7416     ciphers.
7417
7418     Remove lots of duplicated code from the EVP library. For example *every*
7419     cipher init() function handles the 'iv' in the same way according to the
7420     cipher mode. They also all do nothing if the 'key' parameter is NULL and
7421     for CFB and OFB modes they zero ctx->num.
7422
7423     New functionality allows removal of S/MIME code RC2 hack.
7424
7425     Most of the routines have the same form and so can be declared in terms
7426     of macros.
7427
7428     By shifting this to the top level EVP_CipherInit() it can be removed from
7429     all individual ciphers. If the cipher wants to handle IVs or keys
7430     differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
7431     flags.
7432
7433     Change lots of functions like EVP_EncryptUpdate() to now return a
7434     value: although software versions of the algorithms cannot fail
7435     any installed hardware versions can.
7436     [Steve Henson]
7437
7438  *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
7439     this option is set, tolerate broken clients that send the negotiated
7440     protocol version number instead of the requested protocol version
7441     number.
7442     [Bodo Moeller]
7443
7444  *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
7445     i.e. non-zero for export ciphersuites, zero otherwise.
7446     Previous versions had this flag inverted, inconsistent with
7447     rsa_tmp_cb (..._TMP_RSA_CB).
7448     [Bodo Moeller; problem reported by Amit Chopra]
7449
7450  *) Add missing DSA library text string. Work around for some IIS
7451     key files with invalid SEQUENCE encoding.
7452     [Steve Henson]
7453
7454  *) Add a document (doc/standards.txt) that list all kinds of standards
7455     and so on that are implemented in OpenSSL.
7456     [Richard Levitte]
7457
7458  *) Enhance c_rehash script. Old version would mishandle certificates
7459     with the same subject name hash and wouldn't handle CRLs at all.
7460     Added -fingerprint option to crl utility, to support new c_rehash
7461     features.
7462     [Steve Henson]
7463
7464  *) Eliminate non-ANSI declarations in crypto.h and stack.h.
7465     [Ulf M�ller]
7466
7467  *) Fix for SSL server purpose checking. Server checking was
7468     rejecting certificates which had extended key usage present
7469     but no ssl client purpose.
7470     [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
7471
7472  *) Make PKCS#12 code work with no password. The PKCS#12 spec
7473     is a little unclear about how a blank password is handled.
7474     Since the password in encoded as a BMPString with terminating
7475     double NULL a zero length password would end up as just the
7476     double NULL. However no password at all is different and is
7477     handled differently in the PKCS#12 key generation code. NS
7478     treats a blank password as zero length. MSIE treats it as no
7479     password on export: but it will try both on import. We now do
7480     the same: PKCS12_parse() tries zero length and no password if
7481     the password is set to "" or NULL (NULL is now a valid password:
7482     it wasn't before) as does the pkcs12 application.
7483     [Steve Henson]
7484
7485  *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
7486     perror when PEM_read_bio_X509_REQ fails, the error message must
7487     be obtained from the error queue.
7488     [Bodo Moeller]
7489
7490  *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
7491     it in ERR_remove_state if appropriate, and change ERR_get_state
7492     accordingly to avoid race conditions (this is necessary because
7493     thread_hash is no longer constant once set).
7494     [Bodo Moeller]
7495
7496  *) Bugfix for linux-elf makefile.one.
7497     [Ulf M�ller]
7498
7499  *) RSA_get_default_method() will now cause a default
7500     RSA_METHOD to be chosen if one doesn't exist already.
7501     Previously this was only set during a call to RSA_new()
7502     or RSA_new_method(NULL) meaning it was possible for
7503     RSA_get_default_method() to return NULL.
7504     [Geoff Thorpe]
7505
7506  *) Added native name translation to the existing DSO code
7507     that will convert (if the flag to do so is set) filenames
7508     that are sufficiently small and have no path information
7509     into a canonical native form. Eg. "blah" converted to
7510     "libblah.so" or "blah.dll" etc.
7511     [Geoff Thorpe]
7512
7513  *) New function ERR_error_string_n(e, buf, len) which is like
7514     ERR_error_string(e, buf), but writes at most 'len' bytes
7515     including the 0 terminator.  For ERR_error_string_n, 'buf'
7516     may not be NULL.
7517     [Damien Miller <djm@mindrot.org>, Bodo Moeller]
7518
7519  *) CONF library reworked to become more general.  A new CONF
7520     configuration file reader "class" is implemented as well as a
7521     new functions (NCONF_*, for "New CONF") to handle it.  The now
7522     old CONF_* functions are still there, but are reimplemented to
7523     work in terms of the new functions.  Also, a set of functions
7524     to handle the internal storage of the configuration data is
7525     provided to make it easier to write new configuration file
7526     reader "classes" (I can definitely see something reading a
7527     configuration file in XML format, for example), called _CONF_*,
7528     or "the configuration storage API"...
7529
7530     The new configuration file reading functions are:
7531
7532        NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
7533        NCONF_get_section, NCONF_get_string, NCONF_get_numbre
7534
7535        NCONF_default, NCONF_WIN32
7536
7537        NCONF_dump_fp, NCONF_dump_bio
7538
7539     NCONF_default and NCONF_WIN32 are method (or "class") choosers,
7540     NCONF_new creates a new CONF object.  This works in the same way
7541     as other interfaces in OpenSSL, like the BIO interface.
7542     NCONF_dump_* dump the internal storage of the configuration file,
7543     which is useful for debugging.  All other functions take the same
7544     arguments as the old CONF_* functions wth the exception of the
7545     first that must be a `CONF *' instead of a `LHASH *'.
7546
7547     To make it easer to use the new classes with the old CONF_* functions,
7548     the function CONF_set_default_method is provided.
7549     [Richard Levitte]
7550
7551  *) Add '-tls1' option to 'openssl ciphers', which was already
7552     mentioned in the documentation but had not been implemented.
7553     (This option is not yet really useful because even the additional
7554     experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
7555     [Bodo Moeller]
7556
7557  *) Initial DSO code added into libcrypto for letting OpenSSL (and
7558     OpenSSL-based applications) load shared libraries and bind to
7559     them in a portable way.
7560     [Geoff Thorpe, with contributions from Richard Levitte]
7561
7562 Changes between 0.9.5 and 0.9.5a  [1 Apr 2000]
7563
7564  *) Make sure _lrotl and _lrotr are only used with MSVC.
7565
7566  *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
7567     (the default implementation of RAND_status).
7568
7569  *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
7570     to '-clrext' (= clear extensions), as intended and documented.
7571     [Bodo Moeller; inconsistency pointed out by Michael Attili
7572     <attili@amaxo.com>]
7573
7574  *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
7575     was larger than the MD block size.      
7576     [Steve Henson, pointed out by Yost William <YostW@tce.com>]
7577
7578  *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
7579     fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
7580     using the passed key: if the passed key was a private key the result
7581     of X509_print(), for example, would be to print out all the private key
7582     components.
7583     [Steve Henson]
7584
7585  *) des_quad_cksum() byte order bug fix.
7586     [Ulf M�ller, using the problem description in krb4-0.9.7, where
7587      the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
7588
7589  *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
7590     discouraged.
7591     [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
7592
7593  *) For easily testing in shell scripts whether some command
7594     'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
7595     returns with exit code 0 iff no command of the given name is available.
7596     'no-XXX' is printed in this case, 'XXX' otherwise.  In both cases,
7597     the output goes to stdout and nothing is printed to stderr.
7598     Additional arguments are always ignored.
7599
7600     Since for each cipher there is a command of the same name,
7601     the 'no-cipher' compilation switches can be tested this way.
7602
7603     ('openssl no-XXX' is not able to detect pseudo-commands such
7604     as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
7605     [Bodo Moeller]
7606
7607  *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
7608     [Bodo Moeller]
7609
7610  *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
7611     is set; it will be thrown away anyway because each handshake creates
7612     its own key.
7613     ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
7614     to parameters -- in previous versions (since OpenSSL 0.9.3) the
7615     'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
7616     you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
7617     [Bodo Moeller]
7618
7619  *) New s_client option -ign_eof: EOF at stdin is ignored, and
7620     'Q' and 'R' lose their special meanings (quit/renegotiate).
7621     This is part of what -quiet does; unlike -quiet, -ign_eof
7622     does not suppress any output.
7623     [Richard Levitte]
7624
7625  *) Add compatibility options to the purpose and trust code. The
7626     purpose X509_PURPOSE_ANY is "any purpose" which automatically
7627     accepts a certificate or CA, this was the previous behaviour,
7628     with all the associated security issues.
7629
7630     X509_TRUST_COMPAT is the old trust behaviour: only and
7631     automatically trust self signed roots in certificate store. A
7632     new trust setting X509_TRUST_DEFAULT is used to specify that
7633     a purpose has no associated trust setting and it should instead
7634     use the value in the default purpose.
7635     [Steve Henson]
7636
7637  *) Fix the PKCS#8 DSA private key code so it decodes keys again
7638     and fix a memory leak.
7639     [Steve Henson]
7640
7641  *) In util/mkerr.pl (which implements 'make errors'), preserve
7642     reason strings from the previous version of the .c file, as
7643     the default to have only downcase letters (and digits) in
7644     automatically generated reasons codes is not always appropriate.
7645     [Bodo Moeller]
7646
7647  *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
7648     using strerror.  Previously, ERR_reason_error_string() returned
7649     library names as reason strings for SYSerr; but SYSerr is a special
7650     case where small numbers are errno values, not library numbers.
7651     [Bodo Moeller]
7652
7653  *) Add '-dsaparam' option to 'openssl dhparam' application.  This
7654     converts DSA parameters into DH parameters. (When creating parameters,
7655     DSA_generate_parameters is used.)
7656     [Bodo Moeller]
7657
7658  *) Include 'length' (recommended exponent length) in C code generated
7659     by 'openssl dhparam -C'.
7660     [Bodo Moeller]
7661
7662  *) The second argument to set_label in perlasm was already being used
7663     so couldn't be used as a "file scope" flag. Moved to third argument
7664     which was free.
7665     [Steve Henson]
7666
7667  *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
7668     instead of RAND_bytes for encryption IVs and salts.
7669     [Bodo Moeller]
7670
7671  *) Include RAND_status() into RAND_METHOD instead of implementing
7672     it only for md_rand.c  Otherwise replacing the PRNG by calling
7673     RAND_set_rand_method would be impossible.
7674     [Bodo Moeller]
7675
7676  *) Don't let DSA_generate_key() enter an infinite loop if the random
7677     number generation fails.
7678     [Bodo Moeller]
7679
7680  *) New 'rand' application for creating pseudo-random output.
7681     [Bodo Moeller]
7682
7683  *) Added configuration support for Linux/IA64
7684     [Rolf Haberrecker <rolf@suse.de>]
7685
7686  *) Assembler module support for Mingw32.
7687     [Ulf M�ller]
7688
7689  *) Shared library support for HPUX (in shlib/).
7690     [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
7691
7692  *) Shared library support for Solaris gcc.
7693     [Lutz Behnke <behnke@trustcenter.de>]
7694
7695 Changes between 0.9.4 and 0.9.5  [28 Feb 2000]
7696
7697  *) PKCS7_encrypt() was adding text MIME headers twice because they
7698     were added manually and by SMIME_crlf_copy().
7699     [Steve Henson]
7700
7701  *) In bntest.c don't call BN_rand with zero bits argument.
7702     [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
7703
7704  *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
7705     case was implemented. This caused BN_div_recp() to fail occasionally.
7706     [Ulf M�ller]
7707
7708  *) Add an optional second argument to the set_label() in the perl
7709     assembly language builder. If this argument exists and is set
7710     to 1 it signals that the assembler should use a symbol whose 
7711     scope is the entire file, not just the current function. This
7712     is needed with MASM which uses the format label:: for this scope.
7713     [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
7714
7715  *) Change the ASN1 types so they are typedefs by default. Before
7716     almost all types were #define'd to ASN1_STRING which was causing
7717     STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
7718     for example.
7719     [Steve Henson]
7720
7721  *) Change names of new functions to the new get1/get0 naming
7722     convention: After 'get1', the caller owns a reference count
7723     and has to call ..._free; 'get0' returns a pointer to some
7724     data structure without incrementing reference counters.
7725     (Some of the existing 'get' functions increment a reference
7726     counter, some don't.)
7727     Similarly, 'set1' and 'add1' functions increase reference
7728     counters or duplicate objects.
7729     [Steve Henson]
7730
7731  *) Allow for the possibility of temp RSA key generation failure:
7732     the code used to assume it always worked and crashed on failure.
7733     [Steve Henson]
7734
7735  *) Fix potential buffer overrun problem in BIO_printf().
7736     [Ulf M�ller, using public domain code by Patrick Powell; problem
7737      pointed out by David Sacerdote <das33@cornell.edu>]
7738
7739  *) Support EGD <http://www.lothar.com/tech/crypto/>.  New functions
7740     RAND_egd() and RAND_status().  In the command line application,
7741     the EGD socket can be specified like a seed file using RANDFILE
7742     or -rand.
7743     [Ulf M�ller]
7744
7745  *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
7746     Some CAs (e.g. Verisign) distribute certificates in this form.
7747     [Steve Henson]
7748
7749  *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
7750     list to exclude them. This means that no special compilation option
7751     is needed to use anonymous DH: it just needs to be included in the
7752     cipher list.
7753     [Steve Henson]
7754
7755  *) Change the EVP_MD_CTX_type macro so its meaning consistent with
7756     EVP_MD_type. The old functionality is available in a new macro called
7757     EVP_MD_md(). Change code that uses it and update docs.
7758     [Steve Henson]
7759
7760  *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
7761     where the 'void *' argument is replaced by a function pointer argument.
7762     Previously 'void *' was abused to point to functions, which works on
7763     many platforms, but is not correct.  As these functions are usually
7764     called by macros defined in OpenSSL header files, most source code
7765     should work without changes.
7766     [Richard Levitte]
7767
7768  *) <openssl/opensslconf.h> (which is created by Configure) now contains
7769     sections with information on -D... compiler switches used for
7770     compiling the library so that applications can see them.  To enable
7771     one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
7772     must be defined.  E.g.,
7773        #define OPENSSL_ALGORITHM_DEFINES
7774        #include <openssl/opensslconf.h>
7775     defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
7776     [Richard Levitte, Ulf and Bodo M�ller]
7777
7778  *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
7779     record layer.
7780     [Bodo Moeller]
7781
7782  *) Change the 'other' type in certificate aux info to a STACK_OF
7783     X509_ALGOR. Although not an AlgorithmIdentifier as such it has
7784     the required ASN1 format: arbitrary types determined by an OID.
7785     [Steve Henson]
7786
7787  *) Add some PEM_write_X509_REQ_NEW() functions and a command line
7788     argument to 'req'. This is not because the function is newer or
7789     better than others it just uses the work 'NEW' in the certificate
7790     request header lines. Some software needs this.
7791     [Steve Henson]
7792
7793  *) Reorganise password command line arguments: now passwords can be
7794     obtained from various sources. Delete the PEM_cb function and make
7795     it the default behaviour: i.e. if the callback is NULL and the
7796     usrdata argument is not NULL interpret it as a null terminated pass
7797     phrase. If usrdata and the callback are NULL then the pass phrase
7798     is prompted for as usual.
7799     [Steve Henson]
7800
7801  *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
7802     the support is automatically enabled. The resulting binaries will
7803     autodetect the card and use it if present.
7804     [Ben Laurie and Compaq Inc.]
7805
7806  *) Work around for Netscape hang bug. This sends certificate request
7807     and server done in one record. Since this is perfectly legal in the
7808     SSL/TLS protocol it isn't a "bug" option and is on by default. See
7809     the bugs/SSLv3 entry for more info.
7810     [Steve Henson]
7811
7812  *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
7813     [Andy Polyakov]
7814
7815  *) Add -rand argument to smime and pkcs12 applications and read/write
7816     of seed file.
7817     [Steve Henson]
7818
7819  *) New 'passwd' tool for crypt(3) and apr1 password hashes.
7820     [Bodo Moeller]
7821
7822  *) Add command line password options to the remaining applications.
7823     [Steve Henson]
7824
7825  *) Bug fix for BN_div_recp() for numerators with an even number of
7826     bits.
7827     [Ulf M�ller]
7828
7829  *) More tests in bntest.c, and changed test_bn output.
7830     [Ulf M�ller]
7831
7832  *) ./config recognizes MacOS X now.
7833     [Andy Polyakov]
7834
7835  *) Bug fix for BN_div() when the first words of num and divsor are
7836     equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
7837     [Ulf M�ller]
7838
7839  *) Add support for various broken PKCS#8 formats, and command line
7840     options to produce them.
7841     [Steve Henson]
7842
7843  *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
7844     get temporary BIGNUMs from a BN_CTX.
7845     [Ulf M�ller]
7846
7847  *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
7848     for p == 0.
7849     [Ulf M�ller]
7850
7851  *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
7852     include a #define from the old name to the new. The original intent
7853     was that statically linked binaries could for example just call
7854     SSLeay_add_all_ciphers() to just add ciphers to the table and not
7855     link with digests. This never worked becayse SSLeay_add_all_digests()
7856     and SSLeay_add_all_ciphers() were in the same source file so calling
7857     one would link with the other. They are now in separate source files.
7858     [Steve Henson]
7859
7860  *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
7861     [Steve Henson]
7862
7863  *) Use a less unusual form of the Miller-Rabin primality test (it used
7864     a binary algorithm for exponentiation integrated into the Miller-Rabin
7865     loop, our standard modexp algorithms are faster).
7866     [Bodo Moeller]
7867
7868  *) Support for the EBCDIC character set completed.
7869     [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
7870
7871  *) Source code cleanups: use const where appropriate, eliminate casts,
7872     use void * instead of char * in lhash.
7873     [Ulf M�ller] 
7874
7875  *) Bugfix: ssl3_send_server_key_exchange was not restartable
7876     (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
7877     this the server could overwrite ephemeral keys that the client
7878     has already seen).
7879     [Bodo Moeller]
7880
7881  *) Turn DSA_is_prime into a macro that calls BN_is_prime,
7882     using 50 iterations of the Rabin-Miller test.
7883
7884     DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
7885     iterations of the Rabin-Miller test as required by the appendix
7886     to FIPS PUB 186[-1]) instead of DSA_is_prime.
7887     As BN_is_prime_fasttest includes trial division, DSA parameter
7888     generation becomes much faster.
7889
7890     This implies a change for the callback functions in DSA_is_prime
7891     and DSA_generate_parameters: The callback function is called once
7892     for each positive witness in the Rabin-Miller test, not just
7893     occasionally in the inner loop; and the parameters to the
7894     callback function now provide an iteration count for the outer
7895     loop rather than for the current invocation of the inner loop.
7896     DSA_generate_parameters additionally can call the callback
7897     function with an 'iteration count' of -1, meaning that a
7898     candidate has passed the trial division test (when q is generated 
7899     from an application-provided seed, trial division is skipped).
7900     [Bodo Moeller]
7901
7902  *) New function BN_is_prime_fasttest that optionally does trial
7903     division before starting the Rabin-Miller test and has
7904     an additional BN_CTX * argument (whereas BN_is_prime always
7905     has to allocate at least one BN_CTX).
7906     'callback(1, -1, cb_arg)' is called when a number has passed the
7907     trial division stage.
7908     [Bodo Moeller]
7909
7910  *) Fix for bug in CRL encoding. The validity dates weren't being handled
7911     as ASN1_TIME.
7912     [Steve Henson]
7913
7914  *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
7915     [Steve Henson]
7916
7917  *) New function BN_pseudo_rand().
7918     [Ulf M�ller]
7919
7920  *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
7921     bignum version of BN_from_montgomery() with the working code from
7922     SSLeay 0.9.0 (the word based version is faster anyway), and clean up
7923     the comments.
7924     [Ulf M�ller]
7925
7926  *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
7927     made it impossible to use the same SSL_SESSION data structure in
7928     SSL2 clients in multiple threads.
7929     [Bodo Moeller]
7930
7931  *) The return value of RAND_load_file() no longer counts bytes obtained
7932     by stat().  RAND_load_file(..., -1) is new and uses the complete file
7933     to seed the PRNG (previously an explicit byte count was required).
7934     [Ulf M�ller, Bodo M�ller]
7935
7936  *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
7937     used (char *) instead of (void *) and had casts all over the place.
7938     [Steve Henson]
7939
7940  *) Make BN_generate_prime() return NULL on error if ret!=NULL.
7941     [Ulf M�ller]
7942
7943  *) Retain source code compatibility for BN_prime_checks macro:
7944     BN_is_prime(..., BN_prime_checks, ...) now uses
7945     BN_prime_checks_for_size to determine the appropriate number of
7946     Rabin-Miller iterations.
7947     [Ulf M�ller]
7948
7949  *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
7950     DH_CHECK_P_NOT_SAFE_PRIME.
7951     (Check if this is true? OpenPGP calls them "strong".)
7952     [Ulf M�ller]
7953
7954  *) Merge the functionality of "dh" and "gendh" programs into a new program
7955     "dhparam". The old programs are retained for now but will handle DH keys
7956     (instead of parameters) in future.
7957     [Steve Henson]
7958
7959  *) Make the ciphers, s_server and s_client programs check the return values
7960     when a new cipher list is set.
7961     [Steve Henson]
7962
7963  *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
7964     ciphers. Before when the 56bit ciphers were enabled the sorting was
7965     wrong.
7966
7967     The syntax for the cipher sorting has been extended to support sorting by
7968     cipher-strength (using the strength_bits hard coded in the tables).
7969     The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
7970
7971     Fix a bug in the cipher-command parser: when supplying a cipher command
7972     string with an "undefined" symbol (neither command nor alphanumeric
7973     [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
7974     an error is flagged.
7975
7976     Due to the strength-sorting extension, the code of the
7977     ssl_create_cipher_list() function was completely rearranged. I hope that
7978     the readability was also increased :-)
7979     [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
7980
7981  *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
7982     for the first serial number and places 2 in the serial number file. This
7983     avoids problems when the root CA is created with serial number zero and
7984     the first user certificate has the same issuer name and serial number
7985     as the root CA.
7986     [Steve Henson]
7987
7988  *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
7989     the new code. Add documentation for this stuff.
7990     [Steve Henson]
7991
7992  *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
7993     X509_*() to X509at_*() on the grounds that they don't handle X509
7994     structures and behave in an analagous way to the X509v3 functions:
7995     they shouldn't be called directly but wrapper functions should be used
7996     instead.
7997
7998     So we also now have some wrapper functions that call the X509at functions
7999     when passed certificate requests. (TO DO: similar things can be done with
8000     PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
8001     things. Some of these need some d2i or i2d and print functionality
8002     because they handle more complex structures.)
8003     [Steve Henson]
8004
8005  *) Add missing #ifndefs that caused missing symbols when building libssl
8006     as a shared library without RSA.  Use #ifndef NO_SSL2 instead of
8007     NO_RSA in ssl/s2*.c. 
8008     [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M�ller]
8009
8010  *) Precautions against using the PRNG uninitialized: RAND_bytes() now
8011     has a return value which indicates the quality of the random data
8012     (1 = ok, 0 = not seeded).  Also an error is recorded on the thread's
8013     error queue. New function RAND_pseudo_bytes() generates output that is
8014     guaranteed to be unique but not unpredictable. RAND_add is like
8015     RAND_seed, but takes an extra argument for an entropy estimate
8016     (RAND_seed always assumes full entropy).
8017     [Ulf M�ller]
8018
8019  *) Do more iterations of Rabin-Miller probable prime test (specifically,
8020     3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
8021     instead of only 2 for all lengths; see BN_prime_checks_for_size definition
8022     in crypto/bn/bn_prime.c for the complete table).  This guarantees a
8023     false-positive rate of at most 2^-80 for random input.
8024     [Bodo Moeller]
8025
8026  *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
8027     [Bodo Moeller]
8028
8029  *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
8030     in the 0.9.5 release), this returns the chain
8031     from an X509_CTX structure with a dup of the stack and all
8032     the X509 reference counts upped: so the stack will exist
8033     after X509_CTX_cleanup() has been called. Modify pkcs12.c
8034     to use this.
8035
8036     Also make SSL_SESSION_print() print out the verify return
8037     code.
8038     [Steve Henson]
8039
8040  *) Add manpage for the pkcs12 command. Also change the default
8041     behaviour so MAC iteration counts are used unless the new
8042     -nomaciter option is used. This improves file security and
8043     only older versions of MSIE (4.0 for example) need it.
8044     [Steve Henson]
8045
8046  *) Honor the no-xxx Configure options when creating .DEF files.
8047     [Ulf M�ller]
8048
8049  *) Add PKCS#10 attributes to field table: challengePassword, 
8050     unstructuredName and unstructuredAddress. These are taken from
8051     draft PKCS#9 v2.0 but are compatible with v1.2 provided no 
8052     international characters are used.
8053
8054     More changes to X509_ATTRIBUTE code: allow the setting of types
8055     based on strings. Remove the 'loc' parameter when adding
8056     attributes because these will be a SET OF encoding which is sorted
8057     in ASN1 order.
8058     [Steve Henson]
8059
8060  *) Initial changes to the 'req' utility to allow request generation
8061     automation. This will allow an application to just generate a template
8062     file containing all the field values and have req construct the
8063     request.
8064
8065     Initial support for X509_ATTRIBUTE handling. Stacks of these are
8066     used all over the place including certificate requests and PKCS#7
8067     structures. They are currently handled manually where necessary with
8068     some primitive wrappers for PKCS#7. The new functions behave in a
8069     manner analogous to the X509 extension functions: they allow
8070     attributes to be looked up by NID and added.
8071
8072     Later something similar to the X509V3 code would be desirable to
8073     automatically handle the encoding, decoding and printing of the
8074     more complex types. The string types like challengePassword can
8075     be handled by the string table functions.
8076
8077     Also modified the multi byte string table handling. Now there is
8078     a 'global mask' which masks out certain types. The table itself
8079     can use the flag STABLE_NO_MASK to ignore the mask setting: this
8080     is useful when for example there is only one permissible type
8081     (as in countryName) and using the mask might result in no valid
8082     types at all.
8083     [Steve Henson]
8084
8085  *) Clean up 'Finished' handling, and add functions SSL_get_finished and
8086     SSL_get_peer_finished to allow applications to obtain the latest
8087     Finished messages sent to the peer or expected from the peer,
8088     respectively.  (SSL_get_peer_finished is usually the Finished message
8089     actually received from the peer, otherwise the protocol will be aborted.)
8090
8091     As the Finished message are message digests of the complete handshake
8092     (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
8093     be used for external authentication procedures when the authentication
8094     provided by SSL/TLS is not desired or is not enough.
8095     [Bodo Moeller]
8096
8097  *) Enhanced support for Alpha Linux is added. Now ./config checks if
8098     the host supports BWX extension and if Compaq C is present on the
8099     $PATH. Just exploiting of the BWX extension results in 20-30%
8100     performance kick for some algorithms, e.g. DES and RC4 to mention
8101     a couple. Compaq C in turn generates ~20% faster code for MD5 and
8102     SHA1.
8103     [Andy Polyakov]
8104
8105  *) Add support for MS "fast SGC". This is arguably a violation of the
8106     SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
8107     weak crypto and after checking the certificate is SGC a second one
8108     with strong crypto. MS SGC stops the first handshake after receiving
8109     the server certificate message and sends a second client hello. Since
8110     a server will typically do all the time consuming operations before
8111     expecting any further messages from the client (server key exchange
8112     is the most expensive) there is little difference between the two.
8113
8114     To get OpenSSL to support MS SGC we have to permit a second client
8115     hello message after we have sent server done. In addition we have to
8116     reset the MAC if we do get this second client hello.
8117     [Steve Henson]
8118
8119  *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
8120     if a DER encoded private key is RSA or DSA traditional format. Changed
8121     d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
8122     format DER encoded private key. Newer code should use PKCS#8 format which
8123     has the key type encoded in the ASN1 structure. Added DER private key
8124     support to pkcs8 application.
8125     [Steve Henson]
8126
8127  *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
8128     ciphersuites has been selected (as required by the SSL 3/TLS 1
8129     specifications).  Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
8130     is set, we interpret this as a request to violate the specification
8131     (the worst that can happen is a handshake failure, and 'correct'
8132     behaviour would result in a handshake failure anyway).
8133     [Bodo Moeller]
8134
8135  *) In SSL_CTX_add_session, take into account that there might be multiple
8136     SSL_SESSION structures with the same session ID (e.g. when two threads
8137     concurrently obtain them from an external cache).
8138     The internal cache can handle only one SSL_SESSION with a given ID,
8139     so if there's a conflict, we now throw out the old one to achieve
8140     consistency.
8141     [Bodo Moeller]
8142
8143  *) Add OIDs for idea and blowfish in CBC mode. This will allow both
8144     to be used in PKCS#5 v2.0 and S/MIME.  Also add checking to
8145     some routines that use cipher OIDs: some ciphers do not have OIDs
8146     defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
8147     example.
8148     [Steve Henson]
8149
8150  *) Simplify the trust setting structure and code. Now we just have
8151     two sequences of OIDs for trusted and rejected settings. These will
8152     typically have values the same as the extended key usage extension
8153     and any application specific purposes.
8154
8155     The trust checking code now has a default behaviour: it will just
8156     check for an object with the same NID as the passed id. Functions can
8157     be provided to override either the default behaviour or the behaviour
8158     for a given id. SSL client, server and email already have functions
8159     in place for compatibility: they check the NID and also return "trusted"
8160     if the certificate is self signed.
8161     [Steve Henson]
8162
8163  *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
8164     traditional format into an EVP_PKEY structure.
8165     [Steve Henson]
8166
8167  *) Add a password callback function PEM_cb() which either prompts for
8168     a password if usr_data is NULL or otherwise assumes it is a null
8169     terminated password. Allow passwords to be passed on command line
8170     environment or config files in a few more utilities.
8171     [Steve Henson]
8172
8173  *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
8174     keys. Add some short names for PKCS#8 PBE algorithms and allow them
8175     to be specified on the command line for the pkcs8 and pkcs12 utilities.
8176     Update documentation.
8177     [Steve Henson]
8178
8179  *) Support for ASN1 "NULL" type. This could be handled before by using
8180     ASN1_TYPE but there wasn't any function that would try to read a NULL
8181     and produce an error if it couldn't. For compatibility we also have
8182     ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
8183     don't allocate anything because they don't need to.
8184     [Steve Henson]
8185
8186  *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
8187     for details.
8188     [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
8189
8190  *) Rebuild of the memory allocation routines used by OpenSSL code and
8191     possibly others as well.  The purpose is to make an interface that
8192     provide hooks so anyone can build a separate set of allocation and
8193     deallocation routines to be used by OpenSSL, for example memory
8194     pool implementations, or something else, which was previously hard
8195     since Malloc(), Realloc() and Free() were defined as macros having
8196     the values malloc, realloc and free, respectively (except for Win32
8197     compilations).  The same is provided for memory debugging code.
8198     OpenSSL already comes with functionality to find memory leaks, but
8199     this gives people a chance to debug other memory problems.
8200
8201     With these changes, a new set of functions and macros have appeared:
8202
8203       CRYPTO_set_mem_debug_functions()	        [F]
8204       CRYPTO_get_mem_debug_functions()         [F]
8205       CRYPTO_dbg_set_options()	                [F]
8206       CRYPTO_dbg_get_options()                 [F]
8207       CRYPTO_malloc_debug_init()               [M]
8208
8209     The memory debug functions are NULL by default, unless the library
8210     is compiled with CRYPTO_MDEBUG or friends is defined.  If someone
8211     wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
8212     gives the standard debugging functions that come with OpenSSL) or
8213     CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
8214     provided by the library user) must be used.  When the standard
8215     debugging functions are used, CRYPTO_dbg_set_options can be used to
8216     request additional information:
8217     CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
8218     the CRYPTO_MDEBUG_xxx macro when compiling the library.   
8219
8220     Also, things like CRYPTO_set_mem_functions will always give the
8221     expected result (the new set of functions is used for allocation
8222     and deallocation) at all times, regardless of platform and compiler
8223     options.
8224
8225     To finish it up, some functions that were never use in any other
8226     way than through macros have a new API and new semantic:
8227
8228       CRYPTO_dbg_malloc()
8229       CRYPTO_dbg_realloc()
8230       CRYPTO_dbg_free()
8231
8232     All macros of value have retained their old syntax.
8233     [Richard Levitte and Bodo Moeller]
8234
8235  *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
8236     ordering of SMIMECapabilities wasn't in "strength order" and there
8237     was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
8238     algorithm.
8239     [Steve Henson]
8240
8241  *) Some ASN1 types with illegal zero length encoding (INTEGER,
8242     ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
8243     [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
8244
8245  *) Merge in my S/MIME library for OpenSSL. This provides a simple
8246     S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
8247     functionality to handle multipart/signed properly) and a utility
8248     called 'smime' to call all this stuff. This is based on code I
8249     originally wrote for Celo who have kindly allowed it to be
8250     included in OpenSSL.
8251     [Steve Henson]
8252
8253  *) Add variants des_set_key_checked and des_set_key_unchecked of
8254     des_set_key (aka des_key_sched).  Global variable des_check_key
8255     decides which of these is called by des_set_key; this way
8256     des_check_key behaves as it always did, but applications and
8257     the library itself, which was buggy for des_check_key == 1,
8258     have a cleaner way to pick the version they need.
8259     [Bodo Moeller]
8260
8261  *) New function PKCS12_newpass() which changes the password of a
8262     PKCS12 structure.
8263     [Steve Henson]
8264
8265  *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
8266     dynamic mix. In both cases the ids can be used as an index into the
8267     table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
8268     functions so they accept a list of the field values and the
8269     application doesn't need to directly manipulate the X509_TRUST
8270     structure.
8271     [Steve Henson]
8272
8273  *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
8274     need initialising.
8275     [Steve Henson]
8276
8277  *) Modify the way the V3 extension code looks up extensions. This now
8278     works in a similar way to the object code: we have some "standard"
8279     extensions in a static table which is searched with OBJ_bsearch()
8280     and the application can add dynamic ones if needed. The file
8281     crypto/x509v3/ext_dat.h now has the info: this file needs to be
8282     updated whenever a new extension is added to the core code and kept
8283     in ext_nid order. There is a simple program 'tabtest.c' which checks
8284     this. New extensions are not added too often so this file can readily
8285     be maintained manually.
8286
8287     There are two big advantages in doing things this way. The extensions
8288     can be looked up immediately and no longer need to be "added" using
8289     X509V3_add_standard_extensions(): this function now does nothing.
8290     [Side note: I get *lots* of email saying the extension code doesn't
8291      work because people forget to call this function]
8292     Also no dynamic allocation is done unless new extensions are added:
8293     so if we don't add custom extensions there is no need to call
8294     X509V3_EXT_cleanup().
8295     [Steve Henson]
8296
8297  *) Modify enc utility's salting as follows: make salting the default. Add a
8298     magic header, so unsalted files fail gracefully instead of just decrypting
8299     to garbage. This is because not salting is a big security hole, so people
8300     should be discouraged from doing it.
8301     [Ben Laurie]
8302
8303  *) Fixes and enhancements to the 'x509' utility. It allowed a message
8304     digest to be passed on the command line but it only used this
8305     parameter when signing a certificate. Modified so all relevant
8306     operations are affected by the digest parameter including the
8307     -fingerprint and -x509toreq options. Also -x509toreq choked if a
8308     DSA key was used because it didn't fix the digest.
8309     [Steve Henson]
8310
8311  *) Initial certificate chain verify code. Currently tests the untrusted
8312     certificates for consistency with the verify purpose (which is set
8313     when the X509_STORE_CTX structure is set up) and checks the pathlength.
8314
8315     There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
8316     this is because it will reject chains with invalid extensions whereas
8317     every previous version of OpenSSL and SSLeay made no checks at all.
8318
8319     Trust code: checks the root CA for the relevant trust settings. Trust
8320     settings have an initial value consistent with the verify purpose: e.g.
8321     if the verify purpose is for SSL client use it expects the CA to be
8322     trusted for SSL client use. However the default value can be changed to
8323     permit custom trust settings: one example of this would be to only trust
8324     certificates from a specific "secure" set of CAs.
8325
8326     Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
8327     which should be used for version portability: especially since the
8328     verify structure is likely to change more often now.
8329
8330     SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
8331     to set them. If not set then assume SSL clients will verify SSL servers
8332     and vice versa.
8333
8334     Two new options to the verify program: -untrusted allows a set of
8335     untrusted certificates to be passed in and -purpose which sets the
8336     intended purpose of the certificate. If a purpose is set then the
8337     new chain verify code is used to check extension consistency.
8338     [Steve Henson]
8339
8340  *) Support for the authority information access extension.
8341     [Steve Henson]
8342
8343  *) Modify RSA and DSA PEM read routines to transparently handle
8344     PKCS#8 format private keys. New *_PUBKEY_* functions that handle
8345     public keys in a format compatible with certificate
8346     SubjectPublicKeyInfo structures. Unfortunately there were already
8347     functions called *_PublicKey_* which used various odd formats so
8348     these are retained for compatibility: however the DSA variants were
8349     never in a public release so they have been deleted. Changed dsa/rsa
8350     utilities to handle the new format: note no releases ever handled public
8351     keys so we should be OK.
8352
8353     The primary motivation for this change is to avoid the same fiasco
8354     that dogs private keys: there are several incompatible private key
8355     formats some of which are standard and some OpenSSL specific and
8356     require various evil hacks to allow partial transparent handling and
8357     even then it doesn't work with DER formats. Given the option anything
8358     other than PKCS#8 should be dumped: but the other formats have to
8359     stay in the name of compatibility.
8360
8361     With public keys and the benefit of hindsight one standard format 
8362     is used which works with EVP_PKEY, RSA or DSA structures: though
8363     it clearly returns an error if you try to read the wrong kind of key.
8364
8365     Added a -pubkey option to the 'x509' utility to output the public key.
8366     Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
8367     (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
8368     EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
8369     that do the same as the EVP_PKEY_assign_*() except they up the
8370     reference count of the added key (they don't "swallow" the
8371     supplied key).
8372     [Steve Henson]
8373
8374  *) Fixes to crypto/x509/by_file.c the code to read in certificates and
8375     CRLs would fail if the file contained no certificates or no CRLs:
8376     added a new function to read in both types and return the number
8377     read: this means that if none are read it will be an error. The
8378     DER versions of the certificate and CRL reader would always fail
8379     because it isn't possible to mix certificates and CRLs in DER format
8380     without choking one or the other routine. Changed this to just read
8381     a certificate: this is the best we can do. Also modified the code
8382     in apps/verify.c to take notice of return codes: it was previously
8383     attempting to read in certificates from NULL pointers and ignoring
8384     any errors: this is one reason why the cert and CRL reader seemed
8385     to work. It doesn't check return codes from the default certificate
8386     routines: these may well fail if the certificates aren't installed.
8387     [Steve Henson]
8388
8389  *) Code to support otherName option in GeneralName.
8390     [Steve Henson]
8391
8392  *) First update to verify code. Change the verify utility
8393     so it warns if it is passed a self signed certificate:
8394     for consistency with the normal behaviour. X509_verify
8395     has been modified to it will now verify a self signed
8396     certificate if *exactly* the same certificate appears
8397     in the store: it was previously impossible to trust a
8398     single self signed certificate. This means that:
8399     openssl verify ss.pem
8400     now gives a warning about a self signed certificate but
8401     openssl verify -CAfile ss.pem ss.pem
8402     is OK.
8403     [Steve Henson]
8404
8405  *) For servers, store verify_result in SSL_SESSION data structure
8406     (and add it to external session representation).
8407     This is needed when client certificate verifications fails,
8408     but an application-provided verification callback (set by
8409     SSL_CTX_set_cert_verify_callback) allows accepting the session
8410     anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
8411     but returns 1): When the session is reused, we have to set
8412     ssl->verify_result to the appropriate error code to avoid
8413     security holes.
8414     [Bodo Moeller, problem pointed out by Lutz Jaenicke]
8415
8416  *) Fix a bug in the new PKCS#7 code: it didn't consider the
8417     case in PKCS7_dataInit() where the signed PKCS7 structure
8418     didn't contain any existing data because it was being created.
8419     [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
8420
8421  *) Add a salt to the key derivation routines in enc.c. This
8422     forms the first 8 bytes of the encrypted file. Also add a
8423     -S option to allow a salt to be input on the command line.
8424     [Steve Henson]
8425
8426  *) New function X509_cmp(). Oddly enough there wasn't a function
8427     to compare two certificates. We do this by working out the SHA1
8428     hash and comparing that. X509_cmp() will be needed by the trust
8429     code.
8430     [Steve Henson]
8431
8432  *) SSL_get1_session() is like SSL_get_session(), but increments
8433     the reference count in the SSL_SESSION returned.
8434     [Geoff Thorpe <geoff@eu.c2.net>]
8435
8436  *) Fix for 'req': it was adding a null to request attributes.
8437     Also change the X509_LOOKUP and X509_INFO code to handle
8438     certificate auxiliary information.
8439     [Steve Henson]
8440
8441  *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
8442     the 'enc' command.
8443     [Steve Henson]
8444
8445  *) Add the possibility to add extra information to the memory leak
8446     detecting output, to form tracebacks, showing from where each
8447     allocation was originated: CRYPTO_push_info("constant string") adds
8448     the string plus current file name and line number to a per-thread
8449     stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
8450     is like calling CYRPTO_pop_info() until the stack is empty.
8451     Also updated memory leak detection code to be multi-thread-safe.
8452     [Richard Levitte]
8453
8454  *) Add options -text and -noout to pkcs7 utility and delete the
8455     encryption options which never did anything. Update docs.
8456     [Steve Henson]
8457
8458  *) Add options to some of the utilities to allow the pass phrase
8459     to be included on either the command line (not recommended on
8460     OSes like Unix) or read from the environment. Update the
8461     manpages and fix a few bugs.
8462     [Steve Henson]
8463
8464  *) Add a few manpages for some of the openssl commands.
8465     [Steve Henson]
8466
8467  *) Fix the -revoke option in ca. It was freeing up memory twice,
8468     leaking and not finding already revoked certificates.
8469     [Steve Henson]
8470
8471  *) Extensive changes to support certificate auxiliary information.
8472     This involves the use of X509_CERT_AUX structure and X509_AUX
8473     functions. An X509_AUX function such as PEM_read_X509_AUX()
8474     can still read in a certificate file in the usual way but it
8475     will also read in any additional "auxiliary information". By
8476     doing things this way a fair degree of compatibility can be
8477     retained: existing certificates can have this information added
8478     using the new 'x509' options. 
8479
8480     Current auxiliary information includes an "alias" and some trust
8481     settings. The trust settings will ultimately be used in enhanced
8482     certificate chain verification routines: currently a certificate
8483     can only be trusted if it is self signed and then it is trusted
8484     for all purposes.
8485     [Steve Henson]
8486
8487  *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
8488     The problem was that one of the replacement routines had not been working
8489     since SSLeay releases.  For now the offending routine has been replaced
8490     with non-optimised assembler.  Even so, this now gives around 95%
8491     performance improvement for 1024 bit RSA signs.
8492     [Mark Cox]
8493
8494  *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 
8495     handling. Most clients have the effective key size in bits equal to
8496     the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
8497     A few however don't do this and instead use the size of the decrypted key
8498     to determine the RC2 key length and the AlgorithmIdentifier to determine
8499     the effective key length. In this case the effective key length can still
8500     be 40 bits but the key length can be 168 bits for example. This is fixed
8501     by manually forcing an RC2 key into the EVP_PKEY structure because the
8502     EVP code can't currently handle unusual RC2 key sizes: it always assumes
8503     the key length and effective key length are equal.
8504     [Steve Henson]
8505
8506  *) Add a bunch of functions that should simplify the creation of 
8507     X509_NAME structures. Now you should be able to do:
8508     X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
8509     and have it automatically work out the correct field type and fill in
8510     the structures. The more adventurous can try:
8511     X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
8512     and it will (hopefully) work out the correct multibyte encoding.
8513     [Steve Henson]
8514
8515  *) Change the 'req' utility to use the new field handling and multibyte
8516     copy routines. Before the DN field creation was handled in an ad hoc
8517     way in req, ca, and x509 which was rather broken and didn't support
8518     BMPStrings or UTF8Strings. Since some software doesn't implement
8519     BMPStrings or UTF8Strings yet, they can be enabled using the config file
8520     using the dirstring_type option. See the new comment in the default
8521     openssl.cnf for more info.
8522     [Steve Henson]
8523
8524  *) Make crypto/rand/md_rand.c more robust:
8525     - Assure unique random numbers after fork().
8526     - Make sure that concurrent threads access the global counter and
8527       md serializably so that we never lose entropy in them
8528       or use exactly the same state in multiple threads.
8529       Access to the large state is not always serializable because
8530       the additional locking could be a performance killer, and
8531       md should be large enough anyway.
8532     [Bodo Moeller]
8533
8534  *) New file apps/app_rand.c with commonly needed functionality
8535     for handling the random seed file.
8536
8537     Use the random seed file in some applications that previously did not:
8538          ca,
8539          dsaparam -genkey (which also ignored its '-rand' option), 
8540          s_client,
8541          s_server,
8542          x509 (when signing).
8543     Except on systems with /dev/urandom, it is crucial to have a random
8544     seed file at least for key creation, DSA signing, and for DH exchanges;
8545     for RSA signatures we could do without one.
8546
8547     gendh and gendsa (unlike genrsa) used to read only the first byte
8548     of each file listed in the '-rand' option.  The function as previously
8549     found in genrsa is now in app_rand.c and is used by all programs
8550     that support '-rand'.
8551     [Bodo Moeller]
8552
8553  *) In RAND_write_file, use mode 0600 for creating files;
8554     don't just chmod when it may be too late.
8555     [Bodo Moeller]
8556
8557  *) Report an error from X509_STORE_load_locations
8558     when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
8559     [Bill Perry]
8560
8561  *) New function ASN1_mbstring_copy() this copies a string in either
8562     ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
8563     into an ASN1_STRING type. A mask of permissible types is passed
8564     and it chooses the "minimal" type to use or an error if not type
8565     is suitable.
8566     [Steve Henson]
8567
8568  *) Add function equivalents to the various macros in asn1.h. The old
8569     macros are retained with an M_ prefix. Code inside the library can
8570     use the M_ macros. External code (including the openssl utility)
8571     should *NOT* in order to be "shared library friendly".
8572     [Steve Henson]
8573
8574  *) Add various functions that can check a certificate's extensions
8575     to see if it usable for various purposes such as SSL client,
8576     server or S/MIME and CAs of these types. This is currently 
8577     VERY EXPERIMENTAL but will ultimately be used for certificate chain
8578     verification. Also added a -purpose flag to x509 utility to
8579     print out all the purposes.
8580     [Steve Henson]
8581
8582  *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
8583     functions.
8584     [Steve Henson]
8585
8586  *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
8587     for, obtain and decode and extension and obtain its critical flag.
8588     This allows all the necessary extension code to be handled in a
8589     single function call.
8590     [Steve Henson]
8591
8592  *) RC4 tune-up featuring 30-40% performance improvement on most RISC
8593     platforms. See crypto/rc4/rc4_enc.c for further details.
8594     [Andy Polyakov]
8595
8596  *) New -noout option to asn1parse. This causes no output to be produced
8597     its main use is when combined with -strparse and -out to extract data
8598     from a file (which may not be in ASN.1 format).
8599     [Steve Henson]
8600
8601  *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
8602     when producing the local key id.
8603     [Richard Levitte <levitte@stacken.kth.se>]
8604
8605  *) New option -dhparam in s_server. This allows a DH parameter file to be
8606     stated explicitly. If it is not stated then it tries the first server
8607     certificate file. The previous behaviour hard coded the filename
8608     "server.pem".
8609     [Steve Henson]
8610
8611  *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
8612     a public key to be input or output. For example:
8613     openssl rsa -in key.pem -pubout -out pubkey.pem
8614     Also added necessary DSA public key functions to handle this.
8615     [Steve Henson]
8616
8617  *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
8618     in the message. This was handled by allowing
8619     X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
8620     [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
8621
8622  *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
8623     to the end of the strings whereas this didn't. This would cause problems
8624     if strings read with d2i_ASN1_bytes() were later modified.
8625     [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
8626
8627  *) Fix for base64 decode bug. When a base64 bio reads only one line of
8628     data and it contains EOF it will end up returning an error. This is
8629     caused by input 46 bytes long. The cause is due to the way base64
8630     BIOs find the start of base64 encoded data. They do this by trying a
8631     trial decode on each line until they find one that works. When they
8632     do a flag is set and it starts again knowing it can pass all the
8633     data directly through the decoder. Unfortunately it doesn't reset
8634     the context it uses. This means that if EOF is reached an attempt
8635     is made to pass two EOFs through the context and this causes the
8636     resulting error. This can also cause other problems as well. As is
8637     usual with these problems it takes *ages* to find and the fix is
8638     trivial: move one line.
8639     [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
8640
8641  *) Ugly workaround to get s_client and s_server working under Windows. The
8642     old code wouldn't work because it needed to select() on sockets and the
8643     tty (for keypresses and to see if data could be written). Win32 only
8644     supports select() on sockets so we select() with a 1s timeout on the
8645     sockets and then see if any characters are waiting to be read, if none
8646     are present then we retry, we also assume we can always write data to
8647     the tty. This isn't nice because the code then blocks until we've
8648     received a complete line of data and it is effectively polling the
8649     keyboard at 1s intervals: however it's quite a bit better than not
8650     working at all :-) A dedicated Windows application might handle this
8651     with an event loop for example.
8652     [Steve Henson]
8653
8654  *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
8655     and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
8656     will be called when RSA_sign() and RSA_verify() are used. This is useful
8657     if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
8658     For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
8659     should *not* be used: RSA_sign() and RSA_verify() must be used instead.
8660     This necessitated the support of an extra signature type NID_md5_sha1
8661     for SSL signatures and modifications to the SSL library to use it instead
8662     of calling RSA_public_decrypt() and RSA_private_encrypt().
8663     [Steve Henson]
8664
8665  *) Add new -verify -CAfile and -CApath options to the crl program, these
8666     will lookup a CRL issuers certificate and verify the signature in a
8667     similar way to the verify program. Tidy up the crl program so it
8668     no longer accesses structures directly. Make the ASN1 CRL parsing a bit
8669     less strict. It will now permit CRL extensions even if it is not
8670     a V2 CRL: this will allow it to tolerate some broken CRLs.
8671     [Steve Henson]
8672
8673  *) Initialize all non-automatic variables each time one of the openssl
8674     sub-programs is started (this is necessary as they may be started
8675     multiple times from the "OpenSSL>" prompt).
8676     [Lennart Bang, Bodo Moeller]
8677
8678  *) Preliminary compilation option RSA_NULL which disables RSA crypto without
8679     removing all other RSA functionality (this is what NO_RSA does). This
8680     is so (for example) those in the US can disable those operations covered
8681     by the RSA patent while allowing storage and parsing of RSA keys and RSA
8682     key generation.
8683     [Steve Henson]
8684
8685  *) Non-copying interface to BIO pairs.
8686     (still largely untested)
8687     [Bodo Moeller]
8688
8689  *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
8690     ASCII string. This was handled independently in various places before.
8691     [Steve Henson]
8692
8693  *) New functions UTF8_getc() and UTF8_putc() that parse and generate
8694     UTF8 strings a character at a time.
8695     [Steve Henson]
8696
8697  *) Use client_version from client hello to select the protocol
8698     (s23_srvr.c) and for RSA client key exchange verification
8699     (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
8700     [Bodo Moeller]
8701
8702  *) Add various utility functions to handle SPKACs, these were previously
8703     handled by poking round in the structure internals. Added new function
8704     NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
8705     print, verify and generate SPKACs. Based on an original idea from
8706     Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
8707     [Steve Henson]
8708
8709  *) RIPEMD160 is operational on all platforms and is back in 'make test'.
8710     [Andy Polyakov]
8711
8712  *) Allow the config file extension section to be overwritten on the
8713     command line. Based on an original idea from Massimiliano Pala
8714     <madwolf@comune.modena.it>. The new option is called -extensions
8715     and can be applied to ca, req and x509. Also -reqexts to override
8716     the request extensions in req and -crlexts to override the crl extensions
8717     in ca.
8718     [Steve Henson]
8719
8720  *) Add new feature to the SPKAC handling in ca.  Now you can include
8721     the same field multiple times by preceding it by "XXXX." for example:
8722     1.OU="Unit name 1"
8723     2.OU="Unit name 2"
8724     this is the same syntax as used in the req config file.
8725     [Steve Henson]
8726
8727  *) Allow certificate extensions to be added to certificate requests. These
8728     are specified in a 'req_extensions' option of the req section of the
8729     config file. They can be printed out with the -text option to req but
8730     are otherwise ignored at present.
8731     [Steve Henson]
8732
8733  *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
8734     data read consists of only the final block it would not decrypted because
8735     EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
8736     A misplaced 'break' also meant the decrypted final block might not be
8737     copied until the next read.
8738     [Steve Henson]
8739
8740  *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
8741     a few extra parameters to the DH structure: these will be useful if
8742     for example we want the value of 'q' or implement X9.42 DH.
8743     [Steve Henson]
8744
8745  *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
8746     provides hooks that allow the default DSA functions or functions on a
8747     "per key" basis to be replaced. This allows hardware acceleration and
8748     hardware key storage to be handled without major modification to the
8749     library. Also added low level modexp hooks and CRYPTO_EX structure and 
8750     associated functions.
8751     [Steve Henson]
8752
8753  *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
8754     as "read only": it can't be written to and the buffer it points to will
8755     not be freed. Reading from a read only BIO is much more efficient than
8756     a normal memory BIO. This was added because there are several times when
8757     an area of memory needs to be read from a BIO. The previous method was
8758     to create a memory BIO and write the data to it, this results in two
8759     copies of the data and an O(n^2) reading algorithm. There is a new
8760     function BIO_new_mem_buf() which creates a read only memory BIO from
8761     an area of memory. Also modified the PKCS#7 routines to use read only
8762     memory BIOs.
8763     [Steve Henson]
8764
8765  *) Bugfix: ssl23_get_client_hello did not work properly when called in
8766     state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
8767     a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
8768     but a retry condition occured while trying to read the rest.
8769     [Bodo Moeller]
8770
8771  *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
8772     NID_pkcs7_encrypted by default: this was wrong since this should almost
8773     always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
8774     the encrypted data type: this is a more sensible place to put it and it
8775     allows the PKCS#12 code to be tidied up that duplicated this
8776     functionality.
8777     [Steve Henson]
8778
8779  *) Changed obj_dat.pl script so it takes its input and output files on
8780     the command line. This should avoid shell escape redirection problems
8781     under Win32.
8782     [Steve Henson]
8783
8784  *) Initial support for certificate extension requests, these are included
8785     in things like Xenroll certificate requests. Included functions to allow
8786     extensions to be obtained and added.
8787     [Steve Henson]
8788
8789  *) -crlf option to s_client and s_server for sending newlines as
8790     CRLF (as required by many protocols).
8791     [Bodo Moeller]
8792
8793 Changes between 0.9.3a and 0.9.4  [09 Aug 1999]
8794  
8795  *) Install libRSAglue.a when OpenSSL is built with RSAref.
8796     [Ralf S. Engelschall]
8797
8798  *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
8799     [Andrija Antonijevic <TheAntony2@bigfoot.com>]
8800
8801  *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
8802     program.
8803     [Steve Henson]
8804
8805  *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
8806     DH parameters/keys (q is lost during that conversion, but the resulting
8807     DH parameters contain its length).
8808
8809     For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
8810     much faster than DH_generate_parameters (which creates parameters
8811     where p = 2*q + 1), and also the smaller q makes DH computations
8812     much more efficient (160-bit exponentiation instead of 1024-bit
8813     exponentiation); so this provides a convenient way to support DHE
8814     ciphersuites in SSL/TLS servers (see ssl/ssltest.c).  It is of
8815     utter importance to use
8816         SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
8817     or
8818         SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
8819     when such DH parameters are used, because otherwise small subgroup
8820     attacks may become possible!
8821     [Bodo Moeller]
8822
8823  *) Avoid memory leak in i2d_DHparams.
8824     [Bodo Moeller]
8825
8826  *) Allow the -k option to be used more than once in the enc program:
8827     this allows the same encrypted message to be read by multiple recipients.
8828     [Steve Henson]
8829
8830  *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
8831     an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
8832     it will always use the numerical form of the OID, even if it has a short
8833     or long name.
8834     [Steve Henson]
8835
8836  *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
8837     method only got called if p,q,dmp1,dmq1,iqmp components were present,
8838     otherwise bn_mod_exp was called. In the case of hardware keys for example
8839     no private key components need be present and it might store extra data
8840     in the RSA structure, which cannot be accessed from bn_mod_exp.
8841     By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
8842     private key operations.
8843     [Steve Henson]
8844
8845  *) Added support for SPARC Linux.
8846     [Andy Polyakov]
8847
8848  *) pem_password_cb function type incompatibly changed from
8849          typedef int pem_password_cb(char *buf, int size, int rwflag);
8850     to
8851          ....(char *buf, int size, int rwflag, void *userdata);
8852     so that applications can pass data to their callbacks:
8853     The PEM[_ASN1]_{read,write}... functions and macros now take an
8854     additional void * argument, which is just handed through whenever
8855     the password callback is called.
8856     [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]
8857
8858     New function SSL_CTX_set_default_passwd_cb_userdata.
8859
8860     Compatibility note: As many C implementations push function arguments
8861     onto the stack in reverse order, the new library version is likely to
8862     interoperate with programs that have been compiled with the old
8863     pem_password_cb definition (PEM_whatever takes some data that
8864     happens to be on the stack as its last argument, and the callback
8865     just ignores this garbage); but there is no guarantee whatsoever that
8866     this will work.
8867
8868  *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
8869     (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
8870     problems not only on Windows, but also on some Unix platforms.
8871     To avoid problematic command lines, these definitions are now in an
8872     auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
8873     for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
8874     [Bodo Moeller]
8875
8876  *) MIPS III/IV assembler module is reimplemented.
8877     [Andy Polyakov]
8878
8879  *) More DES library cleanups: remove references to srand/rand and
8880     delete an unused file.
8881     [Ulf M�ller]
8882
8883  *) Add support for the the free Netwide assembler (NASM) under Win32,
8884     since not many people have MASM (ml) and it can be hard to obtain.
8885     This is currently experimental but it seems to work OK and pass all
8886     the tests. Check out INSTALL.W32 for info.
8887     [Steve Henson]
8888
8889  *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
8890     without temporary keys kept an extra copy of the server key,
8891     and connections with temporary keys did not free everything in case
8892     of an error.
8893     [Bodo Moeller]
8894
8895  *) New function RSA_check_key and new openssl rsa option -check
8896     for verifying the consistency of RSA keys.
8897     [Ulf Moeller, Bodo Moeller]
8898
8899  *) Various changes to make Win32 compile work: 
8900     1. Casts to avoid "loss of data" warnings in p5_crpt2.c
8901     2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
8902        comparison" warnings.
8903     3. Add sk_<TYPE>_sort to DEF file generator and do make update.
8904     [Steve Henson]
8905
8906  *) Add a debugging option to PKCS#5 v2 key generation function: when
8907     you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
8908     derived keys are printed to stderr.
8909     [Steve Henson]
8910
8911  *) Copy the flags in ASN1_STRING_dup().
8912     [Roman E. Pavlov <pre@mo.msk.ru>]
8913
8914  *) The x509 application mishandled signing requests containing DSA
8915     keys when the signing key was also DSA and the parameters didn't match.
8916
8917     It was supposed to omit the parameters when they matched the signing key:
8918     the verifying software was then supposed to automatically use the CA's
8919     parameters if they were absent from the end user certificate.
8920
8921     Omitting parameters is no longer recommended. The test was also
8922     the wrong way round! This was probably due to unusual behaviour in
8923     EVP_cmp_parameters() which returns 1 if the parameters match. 
8924     This meant that parameters were omitted when they *didn't* match and
8925     the certificate was useless. Certificates signed with 'ca' didn't have
8926     this bug.
8927     [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>]
8928
8929  *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
8930     The interface is as follows:
8931     Applications can use
8932         CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
8933         CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
8934     "off" is now the default.
8935     The library internally uses
8936         CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
8937         CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
8938     to disable memory-checking temporarily.
8939
8940     Some inconsistent states that previously were possible (and were
8941     even the default) are now avoided.
8942
8943     -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
8944     with each memory chunk allocated; this is occasionally more helpful
8945     than just having a counter.
8946
8947     -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
8948
8949     -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
8950     extensions.
8951     [Bodo Moeller]
8952
8953  *) Introduce "mode" for SSL structures (with defaults in SSL_CTX),
8954     which largely parallels "options", but is for changing API behaviour,
8955     whereas "options" are about protocol behaviour.
8956     Initial "mode" flags are:
8957
8958     SSL_MODE_ENABLE_PARTIAL_WRITE   Allow SSL_write to report success when
8959                                     a single record has been written.
8960     SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER  Don't insist that SSL_write
8961                                     retries use the same buffer location.
8962                                     (But all of the contents must be
8963                                     copied!)
8964     [Bodo Moeller]
8965
8966  *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
8967     worked.
8968
8969  *) Fix problems with no-hmac etc.
8970     [Ulf M�ller, pointed out by Brian Wellington <bwelling@tislabs.com>]
8971
8972  *) New functions RSA_get_default_method(), RSA_set_method() and
8973     RSA_get_method(). These allows replacement of RSA_METHODs without having
8974     to mess around with the internals of an RSA structure.
8975     [Steve Henson]
8976
8977  *) Fix memory leaks in DSA_do_sign and DSA_is_prime.
8978     Also really enable memory leak checks in openssl.c and in some
8979     test programs.
8980     [Chad C. Mulligan, Bodo Moeller]
8981
8982  *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
8983     up the length of negative integers. This has now been simplified to just
8984     store the length when it is first determined and use it later, rather
8985     than trying to keep track of where data is copied and updating it to
8986     point to the end.
8987     [Steve Henson, reported by Brien Wheeler
8988      <bwheeler@authentica-security.com>]
8989
8990  *) Add a new function PKCS7_signatureVerify. This allows the verification
8991     of a PKCS#7 signature but with the signing certificate passed to the
8992     function itself. This contrasts with PKCS7_dataVerify which assumes the
8993     certificate is present in the PKCS#7 structure. This isn't always the
8994     case: certificates can be omitted from a PKCS#7 structure and be
8995     distributed by "out of band" means (such as a certificate database).
8996     [Steve Henson]
8997
8998  *) Complete the PEM_* macros with DECLARE_PEM versions to replace the
8999     function prototypes in pem.h, also change util/mkdef.pl to add the
9000     necessary function names. 
9001     [Steve Henson]
9002
9003  *) mk1mf.pl (used by Windows builds) did not properly read the
9004     options set by Configure in the top level Makefile, and Configure
9005     was not even able to write more than one option correctly.
9006     Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
9007     [Bodo Moeller]
9008
9009  *) New functions CONF_load_bio() and CONF_load_fp() to allow a config
9010     file to be loaded from a BIO or FILE pointer. The BIO version will
9011     for example allow memory BIOs to contain config info.
9012     [Steve Henson]
9013
9014  *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
9015     Whoever hopes to achieve shared-library compatibility across versions
9016     must use this, not the compile-time macro.
9017     (Exercise 0.9.4: Which is the minimum library version required by
9018     such programs?)
9019     Note: All this applies only to multi-threaded programs, others don't
9020     need locks.
9021     [Bodo Moeller]
9022
9023  *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests
9024     through a BIO pair triggered the default case, i.e.
9025     SSLerr(...,SSL_R_UNKNOWN_STATE).
9026     [Bodo Moeller]
9027
9028  *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
9029     can use the SSL library even if none of the specific BIOs is
9030     appropriate.
9031     [Bodo Moeller]
9032
9033  *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
9034     for the encoded length.
9035     [Jeon KyoungHo <khjeon@sds.samsung.co.kr>]
9036
9037  *) Add initial documentation of the X509V3 functions.
9038     [Steve Henson]
9039
9040  *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 
9041     PEM_write_bio_PKCS8PrivateKey() that are equivalent to
9042     PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
9043     secure PKCS#8 private key format with a high iteration count.
9044     [Steve Henson]
9045
9046  *) Fix determination of Perl interpreter: A perl or perl5
9047     _directory_ in $PATH was also accepted as the interpreter.
9048     [Ralf S. Engelschall]
9049
9050  *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking
9051     wrong with it but it was very old and did things like calling
9052     PEM_ASN1_read() directly and used MD5 for the hash not to mention some
9053     unusual formatting.
9054     [Steve Henson]
9055
9056  *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed
9057     to use the new extension code.
9058     [Steve Henson]
9059
9060  *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
9061     with macros. This should make it easier to change their form, add extra
9062     arguments etc. Fix a few PEM prototypes which didn't have cipher as a
9063     constant.
9064     [Steve Henson]
9065
9066  *) Add to configuration table a new entry that can specify an alternative
9067     name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
9068     according to Mark Crispin <MRC@Panda.COM>.
9069     [Bodo Moeller]
9070
9071#if 0
9072  *) DES CBC did not update the IV. Weird.
9073     [Ben Laurie]
9074#else
9075     des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
9076     Changing the behaviour of the former might break existing programs --
9077     where IV updating is needed, des_ncbc_encrypt can be used.
9078#endif
9079
9080  *) When bntest is run from "make test" it drives bc to check its
9081     calculations, as well as internally checking them. If an internal check
9082     fails, it needs to cause bc to give a non-zero result or make test carries
9083     on without noticing the failure. Fixed.
9084     [Ben Laurie]
9085
9086  *) DES library cleanups.
9087     [Ulf M�ller]
9088
9089  *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
9090     used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
9091     ciphers. NOTE: although the key derivation function has been verified
9092     against some published test vectors it has not been extensively tested
9093     yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
9094     of v2.0.
9095     [Steve Henson]
9096
9097  *) Instead of "mkdir -p", which is not fully portable, use new
9098     Perl script "util/mkdir-p.pl".
9099     [Bodo Moeller]
9100
9101  *) Rewrite the way password based encryption (PBE) is handled. It used to
9102     assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
9103     structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
9104     but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
9105     the 'parameter' field of the AlgorithmIdentifier is passed to the
9106     underlying key generation function so it must do its own ASN1 parsing.
9107     This has also changed the EVP_PBE_CipherInit() function which now has a
9108     'parameter' argument instead of literal salt and iteration count values
9109     and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
9110     [Steve Henson]
9111
9112  *) Support for PKCS#5 v1.5 compatible password based encryption algorithms
9113     and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
9114     Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
9115     KEY" because this clashed with PKCS#8 unencrypted string. Since this
9116     value was just used as a "magic string" and not used directly its
9117     value doesn't matter.
9118     [Steve Henson]
9119
9120  *) Introduce some semblance of const correctness to BN. Shame C doesn't
9121     support mutable.
9122     [Ben Laurie]
9123
9124  *) "linux-sparc64" configuration (ultrapenguin).
9125     [Ray Miller <ray.miller@oucs.ox.ac.uk>]
9126     "linux-sparc" configuration.
9127     [Christian Forster <fo@hawo.stw.uni-erlangen.de>]
9128
9129  *) config now generates no-xxx options for missing ciphers.
9130     [Ulf M�ller]
9131
9132  *) Support the EBCDIC character set (work in progress).
9133     File ebcdic.c not yet included because it has a different license.
9134     [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
9135
9136  *) Support BS2000/OSD-POSIX.
9137     [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
9138
9139  *) Make callbacks for key generation use void * instead of char *.
9140     [Ben Laurie]
9141
9142  *) Make S/MIME samples compile (not yet tested).
9143     [Ben Laurie]
9144
9145  *) Additional typesafe stacks.
9146     [Ben Laurie]
9147
9148  *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
9149     [Bodo Moeller]
9150
9151
9152 Changes between 0.9.3 and 0.9.3a  [29 May 1999]
9153
9154  *) New configuration variant "sco5-gcc".
9155
9156  *) Updated some demos.
9157     [Sean O Riordain, Wade Scholine]
9158
9159  *) Add missing BIO_free at exit of pkcs12 application.
9160     [Wu Zhigang]
9161
9162  *) Fix memory leak in conf.c.
9163     [Steve Henson]
9164
9165  *) Updates for Win32 to assembler version of MD5.
9166     [Steve Henson]
9167
9168  *) Set #! path to perl in apps/der_chop to where we found it
9169     instead of using a fixed path.
9170     [Bodo Moeller]
9171
9172  *) SHA library changes for irix64-mips4-cc.
9173     [Andy Polyakov]
9174
9175  *) Improvements for VMS support.
9176     [Richard Levitte]
9177
9178
9179 Changes between 0.9.2b and 0.9.3  [24 May 1999]
9180
9181  *) Bignum library bug fix. IRIX 6 passes "make test" now!
9182     This also avoids the problems with SC4.2 and unpatched SC5.  
9183     [Andy Polyakov <appro@fy.chalmers.se>]
9184
9185  *) New functions sk_num, sk_value and sk_set to replace the previous macros.
9186     These are required because of the typesafe stack would otherwise break 
9187     existing code. If old code used a structure member which used to be STACK
9188     and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
9189     sk_num or sk_value it would produce an error because the num, data members
9190     are not present in STACK_OF. Now it just produces a warning. sk_set
9191     replaces the old method of assigning a value to sk_value
9192     (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
9193     that does this will no longer work (and should use sk_set instead) but
9194     this could be regarded as a "questionable" behaviour anyway.
9195     [Steve Henson]
9196
9197  *) Fix most of the other PKCS#7 bugs. The "experimental" code can now
9198     correctly handle encrypted S/MIME data.
9199     [Steve Henson]
9200
9201  *) Change type of various DES function arguments from des_cblock
9202     (which means, in function argument declarations, pointer to char)
9203     to des_cblock * (meaning pointer to array with 8 char elements),
9204     which allows the compiler to do more typechecking; it was like
9205     that back in SSLeay, but with lots of ugly casts.
9206
9207     Introduce new type const_des_cblock.
9208     [Bodo Moeller]
9209
9210  *) Reorganise the PKCS#7 library and get rid of some of the more obvious
9211     problems: find RecipientInfo structure that matches recipient certificate
9212     and initialise the ASN1 structures properly based on passed cipher.
9213     [Steve Henson]
9214
9215  *) Belatedly make the BN tests actually check the results.
9216     [Ben Laurie]
9217
9218  *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
9219     to and from BNs: it was completely broken. New compilation option
9220     NEG_PUBKEY_BUG to allow for some broken certificates that encode public
9221     key elements as negative integers.
9222     [Steve Henson]
9223
9224  *) Reorganize and speed up MD5.
9225     [Andy Polyakov <appro@fy.chalmers.se>]
9226
9227  *) VMS support.
9228     [Richard Levitte <richard@levitte.org>]
9229
9230  *) New option -out to asn1parse to allow the parsed structure to be
9231     output to a file. This is most useful when combined with the -strparse
9232     option to examine the output of things like OCTET STRINGS.
9233     [Steve Henson]
9234
9235  *) Make SSL library a little more fool-proof by not requiring any longer
9236     that SSL_set_{accept,connect}_state be called before
9237     SSL_{accept,connect} may be used (SSL_set_..._state is omitted
9238     in many applications because usually everything *appeared* to work as
9239     intended anyway -- now it really works as intended).
9240     [Bodo Moeller]
9241
9242  *) Move openssl.cnf out of lib/.
9243     [Ulf M�ller]
9244
9245  *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall
9246     -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
9247     -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 
9248     [Ralf S. Engelschall]
9249
9250  *) Various fixes to the EVP and PKCS#7 code. It may now be able to
9251     handle PKCS#7 enveloped data properly.
9252     [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve]
9253
9254  *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
9255     copying pointers.  The cert_st handling is changed by this in
9256     various ways (and thus what used to be known as ctx->default_cert
9257     is now called ctx->cert, since we don't resort to s->ctx->[default_]cert
9258     any longer when s->cert does not give us what we need).
9259     ssl_cert_instantiate becomes obsolete by this change.
9260     As soon as we've got the new code right (possibly it already is?),
9261     we have solved a couple of bugs of the earlier code where s->cert
9262     was used as if it could not have been shared with other SSL structures.
9263
9264     Note that using the SSL API in certain dirty ways now will result
9265     in different behaviour than observed with earlier library versions:
9266     Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx)
9267     does not influence s as it used to.
9268     
9269     In order to clean up things more thoroughly, inside SSL_SESSION
9270     we don't use CERT any longer, but a new structure SESS_CERT
9271     that holds per-session data (if available); currently, this is
9272     the peer's certificate chain and, for clients, the server's certificate
9273     and temporary key.  CERT holds only those values that can have
9274     meaningful defaults in an SSL_CTX.
9275     [Bodo Moeller]
9276
9277  *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
9278     from the internal representation. Various PKCS#7 fixes: remove some
9279     evil casts and set the enc_dig_alg field properly based on the signing
9280     key type.
9281     [Steve Henson]
9282
9283  *) Allow PKCS#12 password to be set from the command line or the
9284     environment. Let 'ca' get its config file name from the environment
9285     variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
9286     and 'x509').
9287     [Steve Henson]
9288
9289  *) Allow certificate policies extension to use an IA5STRING for the
9290     organization field. This is contrary to the PKIX definition but
9291     VeriSign uses it and IE5 only recognises this form. Document 'x509'
9292     extension option.
9293     [Steve Henson]
9294
9295  *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
9296     without disallowing inline assembler and the like for non-pedantic builds.
9297     [Ben Laurie]
9298
9299  *) Support Borland C++ builder.
9300     [Janez Jere <jj@void.si>, modified by Ulf M�ller]
9301
9302  *) Support Mingw32.
9303     [Ulf M�ller]
9304
9305  *) SHA-1 cleanups and performance enhancements.
9306     [Andy Polyakov <appro@fy.chalmers.se>]
9307
9308  *) Sparc v8plus assembler for the bignum library.
9309     [Andy Polyakov <appro@fy.chalmers.se>]
9310
9311  *) Accept any -xxx and +xxx compiler options in Configure.
9312     [Ulf M�ller]
9313
9314  *) Update HPUX configuration.
9315     [Anonymous]
9316  
9317  *) Add missing sk_<type>_unshift() function to safestack.h
9318     [Ralf S. Engelschall]
9319
9320  *) New function SSL_CTX_use_certificate_chain_file that sets the
9321     "extra_cert"s in addition to the certificate.  (This makes sense
9322     only for "PEM" format files, as chains as a whole are not
9323     DER-encoded.)
9324     [Bodo Moeller]
9325
9326  *) Support verify_depth from the SSL API.
9327     x509_vfy.c had what can be considered an off-by-one-error:
9328     Its depth (which was not part of the external interface)
9329     was actually counting the number of certificates in a chain;
9330     now it really counts the depth.
9331     [Bodo Moeller]
9332
9333  *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
9334     instead of X509err, which often resulted in confusing error
9335     messages since the error codes are not globally unique
9336     (e.g. an alleged error in ssl3_accept when a certificate
9337     didn't match the private key).
9338
9339  *) New function SSL_CTX_set_session_id_context that allows to set a default
9340     value (so that you don't need SSL_set_session_id_context for each
9341     connection using the SSL_CTX).
9342     [Bodo Moeller]
9343
9344  *) OAEP decoding bug fix.
9345     [Ulf M�ller]
9346
9347  *) Support INSTALL_PREFIX for package builders, as proposed by
9348     David Harris.
9349     [Bodo Moeller]
9350
9351  *) New Configure options "threads" and "no-threads".  For systems
9352     where the proper compiler options are known (currently Solaris
9353     and Linux), "threads" is the default.
9354     [Bodo Moeller]
9355
9356  *) New script util/mklink.pl as a faster substitute for util/mklink.sh.
9357     [Bodo Moeller]
9358
9359  *) Install various scripts to $(OPENSSLDIR)/misc, not to
9360     $(INSTALLTOP)/bin -- they shouldn't clutter directories
9361     such as /usr/local/bin.
9362     [Bodo Moeller]
9363
9364  *) "make linux-shared" to build shared libraries.
9365     [Niels Poppe <niels@netbox.org>]
9366
9367  *) New Configure option no-<cipher> (rsa, idea, rc5, ...).
9368     [Ulf M�ller]
9369
9370  *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
9371     extension adding in x509 utility.
9372     [Steve Henson]
9373
9374  *) Remove NOPROTO sections and error code comments.
9375     [Ulf M�ller]
9376
9377  *) Partial rewrite of the DEF file generator to now parse the ANSI
9378     prototypes.
9379     [Steve Henson]
9380
9381  *) New Configure options --prefix=DIR and --openssldir=DIR.
9382     [Ulf M�ller]
9383
9384  *) Complete rewrite of the error code script(s). It is all now handled
9385     by one script at the top level which handles error code gathering,
9386     header rewriting and C source file generation. It should be much better
9387     than the old method: it now uses a modified version of Ulf's parser to
9388     read the ANSI prototypes in all header files (thus the old K&R definitions
9389     aren't needed for error creation any more) and do a better job of
9390     translating function codes into names. The old 'ASN1 error code imbedded
9391     in a comment' is no longer necessary and it doesn't use .err files which
9392     have now been deleted. Also the error code call doesn't have to appear all
9393     on one line (which resulted in some large lines...).
9394     [Steve Henson]
9395
9396  *) Change #include filenames from <foo.h> to <openssl/foo.h>.
9397     [Bodo Moeller]
9398
9399  *) Change behaviour of ssl2_read when facing length-0 packets: Don't return
9400     0 (which usually indicates a closed connection), but continue reading.
9401     [Bodo Moeller]
9402
9403  *) Fix some race conditions.
9404     [Bodo Moeller]
9405
9406  *) Add support for CRL distribution points extension. Add Certificate
9407     Policies and CRL distribution points documentation.
9408     [Steve Henson]
9409
9410  *) Move the autogenerated header file parts to crypto/opensslconf.h.
9411     [Ulf M�ller]
9412
9413  *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
9414     8 of keying material. Merlin has also confirmed interop with this fix
9415     between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
9416     [Merlin Hughes <merlin@baltimore.ie>]
9417
9418  *) Fix lots of warnings.
9419     [Richard Levitte <levitte@stacken.kth.se>]
9420 
9421  *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
9422     the directory spec didn't end with a LIST_SEPARATOR_CHAR.
9423     [Richard Levitte <levitte@stacken.kth.se>]
9424 
9425  *) Fix problems with sizeof(long) == 8.
9426     [Andy Polyakov <appro@fy.chalmers.se>]
9427
9428  *) Change functions to ANSI C.
9429     [Ulf M�ller]
9430
9431  *) Fix typos in error codes.
9432     [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M�ller]
9433
9434  *) Remove defunct assembler files from Configure.
9435     [Ulf M�ller]
9436
9437  *) SPARC v8 assembler BIGNUM implementation.
9438     [Andy Polyakov <appro@fy.chalmers.se>]
9439
9440  *) Support for Certificate Policies extension: both print and set.
9441     Various additions to support the r2i method this uses.
9442     [Steve Henson]
9443
9444  *) A lot of constification, and fix a bug in X509_NAME_oneline() that could
9445     return a const string when you are expecting an allocated buffer.
9446     [Ben Laurie]
9447
9448  *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
9449     types DirectoryString and DisplayText.
9450     [Steve Henson]
9451
9452  *) Add code to allow r2i extensions to access the configuration database,
9453     add an LHASH database driver and add several ctx helper functions.
9454     [Steve Henson]
9455
9456  *) Fix an evil bug in bn_expand2() which caused various BN functions to
9457     fail when they extended the size of a BIGNUM.
9458     [Steve Henson]
9459
9460  *) Various utility functions to handle SXNet extension. Modify mkdef.pl to
9461     support typesafe stack.
9462     [Steve Henson]
9463
9464  *) Fix typo in SSL_[gs]et_options().
9465     [Nils Frostberg <nils@medcom.se>]
9466
9467  *) Delete various functions and files that belonged to the (now obsolete)
9468     old X509V3 handling code.
9469     [Steve Henson]
9470
9471  *) New Configure option "rsaref".
9472     [Ulf M�ller]
9473
9474  *) Don't auto-generate pem.h.
9475     [Bodo Moeller]
9476
9477  *) Introduce type-safe ASN.1 SETs.
9478     [Ben Laurie]
9479
9480  *) Convert various additional casted stacks to type-safe STACK_OF() variants.
9481     [Ben Laurie, Ralf S. Engelschall, Steve Henson]
9482
9483  *) Introduce type-safe STACKs. This will almost certainly break lots of code
9484     that links with OpenSSL (well at least cause lots of warnings), but fear
9485     not: the conversion is trivial, and it eliminates loads of evil casts. A
9486     few STACKed things have been converted already. Feel free to convert more.
9487     In the fullness of time, I'll do away with the STACK type altogether.
9488     [Ben Laurie]
9489
9490  *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate
9491     specified in <certfile> by updating the entry in the index.txt file.
9492     This way one no longer has to edit the index.txt file manually for
9493     revoking a certificate. The -revoke option does the gory details now.
9494     [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall]
9495
9496  *) Fix `openssl crl -noout -text' combination where `-noout' killed the
9497     `-text' option at all and this way the `-noout -text' combination was
9498     inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'.
9499     [Ralf S. Engelschall]
9500
9501  *) Make sure a corresponding plain text error message exists for the
9502     X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
9503     verify callback function determined that a certificate was revoked.
9504     [Ralf S. Engelschall]
9505
9506  *) Bugfix: In test/testenc, don't test "openssl <cipher>" for
9507     ciphers that were excluded, e.g. by -DNO_IDEA.  Also, test
9508     all available cipers including rc5, which was forgotten until now.
9509     In order to let the testing shell script know which algorithms
9510     are available, a new (up to now undocumented) command
9511     "openssl list-cipher-commands" is used.
9512     [Bodo Moeller]
9513
9514  *) Bugfix: s_client occasionally would sleep in select() when
9515     it should have checked SSL_pending() first.
9516     [Bodo Moeller]
9517
9518  *) New functions DSA_do_sign and DSA_do_verify to provide access to
9519     the raw DSA values prior to ASN.1 encoding.
9520     [Ulf M�ller]
9521
9522  *) Tweaks to Configure
9523     [Niels Poppe <niels@netbox.org>]
9524
9525  *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
9526     yet...
9527     [Steve Henson]
9528
9529  *) New variables $(RANLIB) and $(PERL) in the Makefiles.
9530     [Ulf M�ller]
9531
9532  *) New config option to avoid instructions that are illegal on the 80386.
9533     The default code is faster, but requires at least a 486.
9534     [Ulf M�ller]
9535  
9536  *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
9537     SSL2_SERVER_VERSION (not used at all) macros, which are now the
9538     same as SSL2_VERSION anyway.
9539     [Bodo Moeller]
9540
9541  *) New "-showcerts" option for s_client.
9542     [Bodo Moeller]
9543
9544  *) Still more PKCS#12 integration. Add pkcs12 application to openssl
9545     application. Various cleanups and fixes.
9546     [Steve Henson]
9547
9548  *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
9549     modify error routines to work internally. Add error codes and PBE init
9550     to library startup routines.
9551     [Steve Henson]
9552
9553  *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and
9554     packing functions to asn1 and evp. Changed function names and error
9555     codes along the way.
9556     [Steve Henson]
9557
9558  *) PKCS12 integration: and so it begins... First of several patches to
9559     slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
9560     objects to objects.h
9561     [Steve Henson]
9562
9563  *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1
9564     and display support for Thawte strong extranet extension.
9565     [Steve Henson]
9566
9567  *) Add LinuxPPC support.
9568     [Jeff Dubrule <igor@pobox.org>]
9569
9570  *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
9571     bn_div_words in alpha.s.
9572     [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie]
9573
9574  *) Make sure the RSA OAEP test is skipped under -DRSAref because
9575     OAEP isn't supported when OpenSSL is built with RSAref.
9576     [Ulf Moeller <ulf@fitug.de>]
9577
9578  *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 
9579     so they no longer are missing under -DNOPROTO. 
9580     [Soren S. Jorvang <soren@t.dk>]
9581
9582
9583 Changes between 0.9.1c and 0.9.2b  [22 Mar 1999]
9584
9585  *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
9586     doesn't work when the session is reused. Coming soon!
9587     [Ben Laurie]
9588
9589  *) Fix a security hole, that allows sessions to be reused in the wrong
9590     context thus bypassing client cert protection! All software that uses
9591     client certs and session caches in multiple contexts NEEDS PATCHING to
9592     allow session reuse! A fuller solution is in the works.
9593     [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)]
9594
9595  *) Some more source tree cleanups (removed obsolete files
9596     crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
9597     permission on "config" script to be executable) and a fix for the INSTALL
9598     document.
9599     [Ulf Moeller <ulf@fitug.de>]
9600
9601  *) Remove some legacy and erroneous uses of malloc, free instead of
9602     Malloc, Free.
9603     [Lennart Bang <lob@netstream.se>, with minor changes by Steve]
9604
9605  *) Make rsa_oaep_test return non-zero on error.
9606     [Ulf Moeller <ulf@fitug.de>]
9607
9608  *) Add support for native Solaris shared libraries. Configure
9609     solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
9610     if someone would make that last step automatic.
9611     [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>]
9612
9613  *) ctx_size was not built with the right compiler during "make links". Fixed.
9614     [Ben Laurie]
9615
9616  *) Change the meaning of 'ALL' in the cipher list. It now means "everything
9617     except NULL ciphers". This means the default cipher list will no longer
9618     enable NULL ciphers. They need to be specifically enabled e.g. with
9619     the string "DEFAULT:eNULL".
9620     [Steve Henson]
9621
9622  *) Fix to RSA private encryption routines: if p < q then it would
9623     occasionally produce an invalid result. This will only happen with
9624     externally generated keys because OpenSSL (and SSLeay) ensure p > q.
9625     [Steve Henson]
9626
9627  *) Be less restrictive and allow also `perl util/perlpath.pl
9628     /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin',
9629     because this way one can also use an interpreter named `perl5' (which is
9630     usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
9631     installed as `perl').
9632     [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
9633
9634  *) Let util/clean-depend.pl work also with older Perl 5.00x versions.
9635     [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
9636
9637  *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
9638     advapi32.lib to Win32 build and change the pem test comparision
9639     to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
9640     suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
9641     and crypto/des/ede_cbcm_enc.c.
9642     [Steve Henson]
9643
9644  *) DES quad checksum was broken on big-endian architectures. Fixed.
9645     [Ben Laurie]
9646
9647  *) Comment out two functions in bio.h that aren't implemented. Fix up the
9648     Win32 test batch file so it (might) work again. The Win32 test batch file
9649     is horrible: I feel ill....
9650     [Steve Henson]
9651
9652  *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
9653     in e_os.h. Audit of header files to check ANSI and non ANSI
9654     sections: 10 functions were absent from non ANSI section and not exported
9655     from Windows DLLs. Fixed up libeay.num for new functions.
9656     [Steve Henson]
9657
9658  *) Make `openssl version' output lines consistent.
9659     [Ralf S. Engelschall]
9660
9661  *) Fix Win32 symbol export lists for BIO functions: Added
9662     BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
9663     to ms/libeay{16,32}.def.
9664     [Ralf S. Engelschall]
9665
9666  *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
9667     fine under Unix and passes some trivial tests I've now added. But the
9668     whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
9669     added to make sure no one expects that this stuff really works in the
9670     OpenSSL 0.9.2 release.  Additionally I've started to clean the XS sources
9671     up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
9672     openssl_bio.xs.
9673     [Ralf S. Engelschall]
9674
9675  *) Fix the generation of two part addresses in perl.
9676     [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie]
9677
9678  *) Add config entry for Linux on MIPS.
9679     [John Tobey <jtobey@channel1.com>]
9680
9681  *) Make links whenever Configure is run, unless we are on Windoze.
9682     [Ben Laurie]
9683
9684  *) Permit extensions to be added to CRLs using crl_section in openssl.cnf.
9685     Currently only issuerAltName and AuthorityKeyIdentifier make any sense
9686     in CRLs.
9687     [Steve Henson]
9688
9689  *) Add a useful kludge to allow package maintainers to specify compiler and
9690     other platforms details on the command line without having to patch the
9691     Configure script everytime: One now can use ``perl Configure
9692     <id>:<details>'', i.e. platform ids are allowed to have details appended
9693     to them (seperated by colons). This is treated as there would be a static
9694     pre-configured entry in Configure's %table under key <id> with value
9695     <details> and ``perl Configure <id>'' is called.  So, when you want to
9696     perform a quick test-compile under FreeBSD 3.1 with pgcc and without
9697     assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"''
9698     now, which overrides the FreeBSD-elf entry on-the-fly.
9699     [Ralf S. Engelschall]
9700
9701  *) Disable new TLS1 ciphersuites by default: they aren't official yet.
9702     [Ben Laurie]
9703
9704  *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
9705     on the `perl Configure ...' command line. This way one can compile
9706     OpenSSL libraries with Position Independent Code (PIC) which is needed
9707     for linking it into DSOs.
9708     [Ralf S. Engelschall]
9709
9710  *) Remarkably, export ciphers were totally broken and no-one had noticed!
9711     Fixed.
9712     [Ben Laurie]
9713
9714  *) Cleaned up the LICENSE document: The official contact for any license
9715     questions now is the OpenSSL core team under openssl-core@openssl.org.
9716     And add a paragraph about the dual-license situation to make sure people
9717     recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
9718     to the OpenSSL toolkit.
9719     [Ralf S. Engelschall]
9720
9721  *) General source tree makefile cleanups: Made `making xxx in yyy...'
9722     display consistent in the source tree and replaced `/bin/rm' by `rm'.
9723     Additonally cleaned up the `make links' target: Remove unnecessary
9724     semicolons, subsequent redundant removes, inline point.sh into mklink.sh
9725     to speed processing and no longer clutter the display with confusing
9726     stuff. Instead only the actually done links are displayed.
9727     [Ralf S. Engelschall]
9728
9729  *) Permit null encryption ciphersuites, used for authentication only. It used
9730     to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
9731     It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
9732     encryption.
9733     [Ben Laurie]
9734
9735  *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
9736     signed attributes when verifying signatures (this would break them), 
9737     the detached data encoding was wrong and public keys obtained using
9738     X509_get_pubkey() weren't freed.
9739     [Steve Henson]
9740
9741  *) Add text documentation for the BUFFER functions. Also added a work around
9742     to a Win95 console bug. This was triggered by the password read stuff: the
9743     last character typed gets carried over to the next fread(). If you were 
9744     generating a new cert request using 'req' for example then the last
9745     character of the passphrase would be CR which would then enter the first
9746     field as blank.
9747     [Steve Henson]
9748
9749  *) Added the new `Includes OpenSSL Cryptography Software' button as
9750     doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
9751     button and can be used by applications based on OpenSSL to show the
9752     relationship to the OpenSSL project.  
9753     [Ralf S. Engelschall]
9754
9755  *) Remove confusing variables in function signatures in files
9756     ssl/ssl_lib.c and ssl/ssl.h.
9757     [Lennart Bong <lob@kulthea.stacken.kth.se>]
9758
9759  *) Don't install bss_file.c under PREFIX/include/
9760     [Lennart Bong <lob@kulthea.stacken.kth.se>]
9761
9762  *) Get the Win32 compile working again. Modify mkdef.pl so it can handle
9763     functions that return function pointers and has support for NT specific
9764     stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
9765     #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
9766     unsigned to signed types: this was killing the Win32 compile.
9767     [Steve Henson]
9768
9769  *) Add new certificate file to stack functions,
9770     SSL_add_dir_cert_subjects_to_stack() and
9771     SSL_add_file_cert_subjects_to_stack().  These largely supplant
9772     SSL_load_client_CA_file(), and can be used to add multiple certs easily
9773     to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
9774     This means that Apache-SSL and similar packages don't have to mess around
9775     to add as many CAs as they want to the preferred list.
9776     [Ben Laurie]
9777
9778  *) Experiment with doxygen documentation. Currently only partially applied to
9779     ssl/ssl_lib.c.
9780     See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with
9781     openssl.doxy as the configuration file.
9782     [Ben Laurie]
9783  
9784  *) Get rid of remaining C++-style comments which strict C compilers hate.
9785     [Ralf S. Engelschall, pointed out by Carlos Amengual]
9786
9787  *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
9788     compiled in by default: it has problems with large keys.
9789     [Steve Henson]
9790
9791  *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
9792     DH private keys and/or callback functions which directly correspond to
9793     their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
9794     is needed for applications which have to configure certificates on a
9795     per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
9796     (e.g. s_server). 
9797        For the RSA certificate situation is makes no difference, but
9798     for the DSA certificate situation this fixes the "no shared cipher"
9799     problem where the OpenSSL cipher selection procedure failed because the
9800     temporary keys were not overtaken from the context and the API provided
9801     no way to reconfigure them. 
9802        The new functions now let applications reconfigure the stuff and they
9803     are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
9804     SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback.  Additionally a new
9805     non-public-API function ssl_cert_instantiate() is used as a helper
9806     function and also to reduce code redundancy inside ssl_rsa.c.
9807     [Ralf S. Engelschall]
9808
9809  *) Move s_server -dcert and -dkey options out of the undocumented feature
9810     area because they are useful for the DSA situation and should be
9811     recognized by the users.
9812     [Ralf S. Engelschall]
9813
9814  *) Fix the cipher decision scheme for export ciphers: the export bits are
9815     *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
9816     SSL_EXP_MASK.  So, the original variable has to be used instead of the
9817     already masked variable.
9818     [Richard Levitte <levitte@stacken.kth.se>]
9819
9820  *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c
9821     [Richard Levitte <levitte@stacken.kth.se>]
9822
9823  *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
9824     from `int' to `unsigned int' because it's a length and initialized by
9825     EVP_DigestFinal() which expects an `unsigned int *'.
9826     [Richard Levitte <levitte@stacken.kth.se>]
9827
9828  *) Don't hard-code path to Perl interpreter on shebang line of Configure
9829     script. Instead use the usual Shell->Perl transition trick.
9830     [Ralf S. Engelschall]
9831
9832  *) Make `openssl x509 -noout -modulus' functional also for DSA certificates
9833     (in addition to RSA certificates) to match the behaviour of `openssl dsa
9834     -noout -modulus' as it's already the case for `openssl rsa -noout
9835     -modulus'.  For RSA the -modulus is the real "modulus" while for DSA
9836     currently the public key is printed (a decision which was already done by
9837     `openssl dsa -modulus' in the past) which serves a similar purpose.
9838     Additionally the NO_RSA no longer completely removes the whole -modulus
9839     option; it now only avoids using the RSA stuff. Same applies to NO_DSA
9840     now, too.
9841     [Ralf S.  Engelschall]
9842
9843  *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
9844     BIO. See the source (crypto/evp/bio_ok.c) for more info.
9845     [Arne Ansper <arne@ats.cyber.ee>]
9846
9847  *) Dump the old yucky req code that tried (and failed) to allow raw OIDs
9848     to be added. Now both 'req' and 'ca' can use new objects defined in the
9849     config file.
9850     [Steve Henson]
9851
9852  *) Add cool BIO that does syslog (or event log on NT).
9853     [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie]
9854
9855  *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
9856     TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
9857     TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
9858     Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
9859     [Ben Laurie]
9860
9861  *) Add preliminary config info for new extension code.
9862     [Steve Henson]
9863
9864  *) Make RSA_NO_PADDING really use no padding.
9865     [Ulf Moeller <ulf@fitug.de>]
9866
9867  *) Generate errors when private/public key check is done.
9868     [Ben Laurie]
9869
9870  *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
9871     for some CRL extensions and new objects added.
9872     [Steve Henson]
9873
9874  *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private
9875     key usage extension and fuller support for authority key id.
9876     [Steve Henson]
9877
9878  *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
9879     padding method for RSA, which is recommended for new applications in PKCS
9880     #1 v2.0 (RFC 2437, October 1998).
9881     OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
9882     foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
9883     against Bleichbacher's attack on RSA.
9884     [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
9885      Ben Laurie]
9886
9887  *) Updates to the new SSL compression code
9888     [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
9889
9890  *) Fix so that the version number in the master secret, when passed
9891     via RSA, checks that if TLS was proposed, but we roll back to SSLv3
9892     (because the server will not accept higher), that the version number
9893     is 0x03,0x01, not 0x03,0x00
9894     [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
9895
9896  *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory
9897     leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes
9898     in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c
9899     [Steve Henson]
9900
9901  *) Support for RAW extensions where an arbitrary extension can be
9902     created by including its DER encoding. See apps/openssl.cnf for
9903     an example.
9904     [Steve Henson]
9905
9906  *) Make sure latest Perl versions don't interpret some generated C array
9907     code as Perl array code in the crypto/err/err_genc.pl script.
9908     [Lars Weber <3weber@informatik.uni-hamburg.de>]
9909
9910  *) Modify ms/do_ms.bat to not generate assembly language makefiles since
9911     not many people have the assembler. Various Win32 compilation fixes and
9912     update to the INSTALL.W32 file with (hopefully) more accurate Win32
9913     build instructions.
9914     [Steve Henson]
9915
9916  *) Modify configure script 'Configure' to automatically create crypto/date.h
9917     file under Win32 and also build pem.h from pem.org. New script
9918     util/mkfiles.pl to create the MINFO file on environments that can't do a
9919     'make files': perl util/mkfiles.pl >MINFO should work.
9920     [Steve Henson]
9921
9922  *) Major rework of DES function declarations, in the pursuit of correctness
9923     and purity. As a result, many evil casts evaporated, and some weirdness,
9924     too. You may find this causes warnings in your code. Zapping your evil
9925     casts will probably fix them. Mostly.
9926     [Ben Laurie]
9927
9928  *) Fix for a typo in asn1.h. Bug fix to object creation script
9929     obj_dat.pl. It considered a zero in an object definition to mean
9930     "end of object": none of the objects in objects.h have any zeros
9931     so it wasn't spotted.
9932     [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>]
9933
9934  *) Add support for Triple DES Cipher Block Chaining with Output Feedback
9935     Masking (CBCM). In the absence of test vectors, the best I have been able
9936     to do is check that the decrypt undoes the encrypt, so far. Send me test
9937     vectors if you have them.
9938     [Ben Laurie]
9939
9940  *) Correct calculation of key length for export ciphers (too much space was
9941     allocated for null ciphers). This has not been tested!
9942     [Ben Laurie]
9943
9944  *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage
9945     message is now correct (it understands "crypto" and "ssl" on its
9946     command line). There is also now an "update" option. This will update
9947     the util/ssleay.num and util/libeay.num files with any new functions.
9948     If you do a: 
9949     perl util/mkdef.pl crypto ssl update
9950     it will update them.
9951     [Steve Henson]
9952
9953  *) Overhauled the Perl interface (perl/*):
9954     - ported BN stuff to OpenSSL's different BN library
9955     - made the perl/ source tree CVS-aware
9956     - renamed the package from SSLeay to OpenSSL (the files still contain
9957       their history because I've copied them in the repository)
9958     - removed obsolete files (the test scripts will be replaced
9959       by better Test::Harness variants in the future)
9960     [Ralf S. Engelschall]
9961
9962  *) First cut for a very conservative source tree cleanup:
9963     1. merge various obsolete readme texts into doc/ssleay.txt
9964     where we collect the old documents and readme texts.
9965     2. remove the first part of files where I'm already sure that we no
9966     longer need them because of three reasons: either they are just temporary
9967     files which were left by Eric or they are preserved original files where
9968     I've verified that the diff is also available in the CVS via "cvs diff
9969     -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
9970     the crypto/md/ stuff).
9971     [Ralf S. Engelschall]
9972
9973  *) More extension code. Incomplete support for subject and issuer alt
9974     name, issuer and authority key id. Change the i2v function parameters
9975     and add an extra 'crl' parameter in the X509V3_CTX structure: guess
9976     what that's for :-) Fix to ASN1 macro which messed up
9977     IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
9978     [Steve Henson]
9979
9980  *) Preliminary support for ENUMERATED type. This is largely copied from the
9981     INTEGER code.
9982     [Steve Henson]
9983
9984  *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
9985     [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
9986
9987  *) Make sure `make rehash' target really finds the `openssl' program.
9988     [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
9989
9990  *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
9991     like to hear about it if this slows down other processors.
9992     [Ben Laurie]
9993
9994  *) Add CygWin32 platform information to Configure script.
9995     [Alan Batie <batie@aahz.jf.intel.com>]
9996
9997  *) Fixed ms/32all.bat script: `no_asm' -> `no-asm'
9998     [Rainer W. Gerling <gerling@mpg-gv.mpg.de>]
9999  
10000  *) New program nseq to manipulate netscape certificate sequences
10001     [Steve Henson]
10002
10003  *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
10004     few typos.
10005     [Steve Henson]
10006
10007  *) Fixes to BN code.  Previously the default was to define BN_RECURSION
10008     but the BN code had some problems that would cause failures when
10009     doing certificate verification and some other functions.
10010     [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
10011
10012  *) Add ASN1 and PEM code to support netscape certificate sequences.
10013     [Steve Henson]
10014
10015  *) Add ASN1 and PEM code to support netscape certificate sequences.
10016     [Steve Henson]
10017
10018  *) Add several PKIX and private extended key usage OIDs.
10019     [Steve Henson]
10020
10021  *) Modify the 'ca' program to handle the new extension code. Modify
10022     openssl.cnf for new extension format, add comments.
10023     [Steve Henson]
10024
10025  *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
10026     and add a sample to openssl.cnf so req -x509 now adds appropriate
10027     CA extensions.
10028     [Steve Henson]
10029
10030  *) Continued X509 V3 changes. Add to other makefiles, integrate with the
10031     error code, add initial support to X509_print() and x509 application.
10032     [Steve Henson]
10033
10034  *) Takes a deep breath and start addding X509 V3 extension support code. Add
10035     files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
10036     stuff is currently isolated and isn't even compiled yet.
10037     [Steve Henson]
10038
10039  *) Continuing patches for GeneralizedTime. Fix up certificate and CRL
10040     ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
10041     Removed the versions check from X509 routines when loading extensions:
10042     this allows certain broken certificates that don't set the version
10043     properly to be processed.
10044     [Steve Henson]
10045
10046  *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
10047     Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
10048     can still be regenerated with "make depend".
10049     [Ben Laurie]
10050
10051  *) Spelling mistake in C version of CAST-128.
10052     [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>]
10053
10054  *) Changes to the error generation code. The perl script err-code.pl 
10055     now reads in the old error codes and retains the old numbers, only
10056     adding new ones if necessary. It also only changes the .err files if new
10057     codes are added. The makefiles have been modified to only insert errors
10058     when needed (to avoid needlessly modifying header files). This is done
10059     by only inserting errors if the .err file is newer than the auto generated
10060     C file. To rebuild all the error codes from scratch (the old behaviour)
10061     either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
10062     or delete all the .err files.
10063     [Steve Henson]
10064
10065  *) CAST-128 was incorrectly implemented for short keys. The C version has
10066     been fixed, but is untested. The assembler versions are also fixed, but
10067     new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
10068     to regenerate it if needed.
10069     [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
10070      Hagino <itojun@kame.net>]
10071
10072  *) File was opened incorrectly in randfile.c.
10073     [Ulf M�ller <ulf@fitug.de>]
10074
10075  *) Beginning of support for GeneralizedTime. d2i, i2d, check and print
10076     functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
10077     GeneralizedTime. ASN1_TIME is the proper type used in certificates et
10078     al: it's just almost always a UTCTime. Note this patch adds new error
10079     codes so do a "make errors" if there are problems.
10080     [Steve Henson]
10081
10082  *) Correct Linux 1 recognition in config.
10083     [Ulf M�ller <ulf@fitug.de>]
10084
10085  *) Remove pointless MD5 hash when using DSA keys in ca.
10086     [Anonymous <nobody@replay.com>]
10087
10088  *) Generate an error if given an empty string as a cert directory. Also
10089     generate an error if handed NULL (previously returned 0 to indicate an
10090     error, but didn't set one).
10091     [Ben Laurie, reported by Anonymous <nobody@replay.com>]
10092
10093  *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
10094     [Ben Laurie]
10095
10096  *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
10097     parameters. This was causing a warning which killed off the Win32 compile.
10098     [Steve Henson]
10099
10100  *) Remove C++ style comments from crypto/bn/bn_local.h.
10101     [Neil Costigan <neil.costigan@celocom.com>]
10102
10103  *) The function OBJ_txt2nid was broken. It was supposed to return a nid
10104     based on a text string, looking up short and long names and finally
10105     "dot" format. The "dot" format stuff didn't work. Added new function
10106     OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 
10107     OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
10108     OID is not part of the table.
10109     [Steve Henson]
10110
10111  *) Add prototypes to X509 lookup/verify methods, fixing a bug in
10112     X509_LOOKUP_by_alias().
10113     [Ben Laurie]
10114
10115  *) Sort openssl functions by name.
10116     [Ben Laurie]
10117
10118  *) Get the gendsa program working (hopefully) and add it to app list. Remove
10119     encryption from sample DSA keys (in case anyone is interested the password
10120     was "1234").
10121     [Steve Henson]
10122
10123  *) Make _all_ *_free functions accept a NULL pointer.
10124     [Frans Heymans <fheymans@isaserver.be>]
10125
10126  *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use
10127     NULL pointers.
10128     [Anonymous <nobody@replay.com>]
10129
10130  *) s_server should send the CAfile as acceptable CAs, not its own cert.
10131     [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
10132
10133  *) Don't blow it for numeric -newkey arguments to apps/req.
10134     [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
10135
10136  *) Temp key "for export" tests were wrong in s3_srvr.c.
10137     [Anonymous <nobody@replay.com>]
10138
10139  *) Add prototype for temp key callback functions
10140     SSL_CTX_set_tmp_{rsa,dh}_callback().
10141     [Ben Laurie]
10142
10143  *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
10144     DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
10145     [Steve Henson]
10146
10147  *) X509_name_add_entry() freed the wrong thing after an error.
10148     [Arne Ansper <arne@ats.cyber.ee>]
10149
10150  *) rsa_eay.c would attempt to free a NULL context.
10151     [Arne Ansper <arne@ats.cyber.ee>]
10152
10153  *) BIO_s_socket() had a broken should_retry() on Windoze.
10154     [Arne Ansper <arne@ats.cyber.ee>]
10155
10156  *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
10157     [Arne Ansper <arne@ats.cyber.ee>]
10158
10159  *) Make sure the already existing X509_STORE->depth variable is initialized
10160     in X509_STORE_new(), but document the fact that this variable is still
10161     unused in the certificate verification process.
10162     [Ralf S. Engelschall]
10163
10164  *) Fix the various library and apps files to free up pkeys obtained from
10165     X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
10166     [Steve Henson]
10167
10168  *) Fix reference counting in X509_PUBKEY_get(). This makes
10169     demos/maurice/example2.c work, amongst others, probably.
10170     [Steve Henson and Ben Laurie]
10171
10172  *) First cut of a cleanup for apps/. First the `ssleay' program is now named
10173     `openssl' and second, the shortcut symlinks for the `openssl <command>'
10174     are no longer created. This way we have a single and consistent command
10175     line interface `openssl <command>', similar to `cvs <command>'.
10176     [Ralf S. Engelschall, Paul Sutton and Ben Laurie]
10177
10178  *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
10179     BIT STRING wrapper always have zero unused bits.
10180     [Steve Henson]
10181
10182  *) Add CA.pl, perl version of CA.sh, add extended key usage OID.
10183     [Steve Henson]
10184
10185  *) Make the top-level INSTALL documentation easier to understand.
10186     [Paul Sutton]
10187
10188  *) Makefiles updated to exit if an error occurs in a sub-directory
10189     make (including if user presses ^C) [Paul Sutton]
10190
10191  *) Make Montgomery context stuff explicit in RSA data structure.
10192     [Ben Laurie]
10193
10194  *) Fix build order of pem and err to allow for generated pem.h.
10195     [Ben Laurie]
10196
10197  *) Fix renumbering bug in X509_NAME_delete_entry().
10198     [Ben Laurie]
10199
10200  *) Enhanced the err-ins.pl script so it makes the error library number 
10201     global and can add a library name. This is needed for external ASN1 and
10202     other error libraries.
10203     [Steve Henson]
10204
10205  *) Fixed sk_insert which never worked properly.
10206     [Steve Henson]
10207
10208  *) Fix ASN1 macros so they can handle indefinite length construted 
10209     EXPLICIT tags. Some non standard certificates use these: they can now
10210     be read in.
10211     [Steve Henson]
10212
10213  *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
10214     into a single doc/ssleay.txt bundle. This way the information is still
10215     preserved but no longer messes up this directory. Now it's new room for
10216     the new set of documenation files.
10217     [Ralf S. Engelschall]
10218
10219  *) SETs were incorrectly DER encoded. This was a major pain, because they
10220     shared code with SEQUENCEs, which aren't coded the same. This means that
10221     almost everything to do with SETs or SEQUENCEs has either changed name or
10222     number of arguments.
10223     [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>]
10224
10225  *) Fix test data to work with the above.
10226     [Ben Laurie]
10227
10228  *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
10229     was already fixed by Eric for 0.9.1 it seems.
10230     [Ben Laurie - pointed out by Ulf M�ller <ulf@fitug.de>]
10231
10232  *) Autodetect FreeBSD3.
10233     [Ben Laurie]
10234
10235  *) Fix various bugs in Configure. This affects the following platforms:
10236     nextstep
10237     ncr-scde
10238     unixware-2.0
10239     unixware-2.0-pentium
10240     sco5-cc.
10241     [Ben Laurie]
10242
10243  *) Eliminate generated files from CVS. Reorder tests to regenerate files
10244     before they are needed.
10245     [Ben Laurie]
10246
10247  *) Generate Makefile.ssl from Makefile.org (to keep CVS happy).
10248     [Ben Laurie]
10249
10250
10251 Changes between 0.9.1b and 0.9.1c  [23-Dec-1998]
10252
10253  *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 
10254     changed SSLeay to OpenSSL in version strings.
10255     [Ralf S. Engelschall]
10256  
10257  *) Some fixups to the top-level documents.
10258     [Paul Sutton]
10259
10260  *) Fixed the nasty bug where rsaref.h was not found under compile-time
10261     because the symlink to include/ was missing.
10262     [Ralf S. Engelschall]
10263
10264  *) Incorporated the popular no-RSA/DSA-only patches 
10265     which allow to compile a RSA-free SSLeay.
10266     [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall]
10267
10268  *) Fixed nasty rehash problem under `make -f Makefile.ssl links'
10269     when "ssleay" is still not found.
10270     [Ralf S. Engelschall]
10271
10272  *) Added more platforms to Configure: Cray T3E, HPUX 11, 
10273     [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>]
10274
10275  *) Updated the README file.
10276     [Ralf S. Engelschall]
10277
10278  *) Added various .cvsignore files in the CVS repository subdirs
10279     to make a "cvs update" really silent.
10280     [Ralf S. Engelschall]
10281
10282  *) Recompiled the error-definition header files and added
10283     missing symbols to the Win32 linker tables.
10284     [Ralf S. Engelschall]
10285
10286  *) Cleaned up the top-level documents;
10287     o new files: CHANGES and LICENSE
10288     o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 
10289     o merged COPYRIGHT into LICENSE
10290     o removed obsolete TODO file
10291     o renamed MICROSOFT to INSTALL.W32
10292     [Ralf S. Engelschall]
10293
10294  *) Removed dummy files from the 0.9.1b source tree: 
10295     crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
10296     crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
10297     crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
10298     crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
10299     util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
10300     [Ralf S. Engelschall]
10301
10302  *) Added various platform portability fixes.
10303     [Mark J. Cox]
10304
10305  *) The Genesis of the OpenSSL rpject:
10306     We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
10307     Young and Tim J. Hudson created while they were working for C2Net until
10308     summer 1998.
10309     [The OpenSSL Project]
10310 
10311
10312 Changes between 0.9.0b and 0.9.1b  [not released]
10313
10314  *) Updated a few CA certificates under certs/
10315     [Eric A. Young]
10316
10317  *) Changed some BIGNUM api stuff.
10318     [Eric A. Young]
10319
10320  *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 
10321     DGUX x86, Linux Alpha, etc.
10322     [Eric A. Young]
10323
10324  *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 
10325     RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
10326     available).
10327     [Eric A. Young]
10328
10329  *) Add -strparse option to asn1pars program which parses nested 
10330     binary structures 
10331     [Dr Stephen Henson <shenson@bigfoot.com>]
10332
10333  *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
10334     [Eric A. Young]
10335
10336  *) DSA fix for "ca" program.
10337     [Eric A. Young]
10338
10339  *) Added "-genkey" option to "dsaparam" program.
10340     [Eric A. Young]
10341
10342  *) Added RIPE MD160 (rmd160) message digest.
10343     [Eric A. Young]
10344
10345  *) Added -a (all) option to "ssleay version" command.
10346     [Eric A. Young]
10347
10348  *) Added PLATFORM define which is the id given to Configure.
10349     [Eric A. Young]
10350
10351  *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
10352     [Eric A. Young]
10353
10354  *) Extended the ASN.1 parser routines.
10355     [Eric A. Young]
10356
10357  *) Extended BIO routines to support REUSEADDR, seek, tell, etc.
10358     [Eric A. Young]
10359
10360  *) Added a BN_CTX to the BN library.
10361     [Eric A. Young]
10362
10363  *) Fixed the weak key values in DES library
10364     [Eric A. Young]
10365
10366  *) Changed API in EVP library for cipher aliases.
10367     [Eric A. Young]
10368
10369  *) Added support for RC2/64bit cipher.
10370     [Eric A. Young]
10371
10372  *) Converted the lhash library to the crypto/mem.c functions.
10373     [Eric A. Young]
10374
10375  *) Added more recognized ASN.1 object ids.
10376     [Eric A. Young]
10377
10378  *) Added more RSA padding checks for SSL/TLS.
10379     [Eric A. Young]
10380
10381  *) Added BIO proxy/filter functionality.
10382     [Eric A. Young]
10383
10384  *) Added extra_certs to SSL_CTX which can be used
10385     send extra CA certificates to the client in the CA cert chain sending
10386     process. It can be configured with SSL_CTX_add_extra_chain_cert().
10387     [Eric A. Young]
10388
10389  *) Now Fortezza is denied in the authentication phase because
10390     this is key exchange mechanism is not supported by SSLeay at all.
10391     [Eric A. Young]
10392
10393  *) Additional PKCS1 checks.
10394     [Eric A. Young]
10395
10396  *) Support the string "TLSv1" for all TLS v1 ciphers.
10397     [Eric A. Young]
10398
10399  *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
10400     ex_data index of the SSL context in the X509_STORE_CTX ex_data.
10401     [Eric A. Young]
10402
10403  *) Fixed a few memory leaks.
10404     [Eric A. Young]
10405
10406  *) Fixed various code and comment typos.
10407     [Eric A. Young]
10408
10409  *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 
10410     bytes sent in the client random.
10411     [Edward Bishop <ebishop@spyglass.com>]
10412
10413