(draft-rescorla-tls-opaque-prf-input-00.txt), and do some cleanups and
bugfixes on the way. In particular, this fixes the buffer bounds
checks in ssl_add_clienthello_tlsext() and in ssl_add_serverhello_tlsext().
Note that the opaque PRF Input TLS extension is not compiled by default;
see CHANGES.
This change resolves a number of problems and obviates multiple kludges.
A new feature is that you can now say "AES256" or "AES128" (not just
"AES", which enables both).
In some cases the ciphersuite list generated from a given string is
affected by this change. I hope this is just in those cases where the
previous behaviour did not make sense.
(CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
Fix SSL client code which could crash if connecting to a
malicious SSLv2 server. (CVE-2006-4343)
[Tavis Ormandy and Will Drewry, Google Security Team]
This tidies up verify parameters and adds support for integrated policy
checking.
Add support for policy related command line options. Currently only in smime
application.
WARNING: experimental code subject to change.
relates to SSL_CTX flags and the use of "external" session caching. The
existing flag, "SSL_SESS_CACHE_NO_INTERNAL_LOOKUP" remains but is
supplemented with a complimentary flag, "SSL_SESS_CACHE_NO_INTERNAL_STORE".
The bitwise OR of the two flags is also defined as
"SSL_SESS_CACHE_NO_INTERNAL" and is the flag that should be used by most
applications wanting to implement session caching *entirely* by its own
provided callbacks. As the documented behaviour contradicted actual
behaviour up until recently, and since that point behaviour has itself been
inconsistent anyway, this change should not introduce any compatibility
problems. I've adjusted the relevant documentation to elaborate about how
this works.
Kudos to "Nadav Har'El" <nyh@math.technion.ac.il> for diagnosing these
anomalies and testing this patch for correctness.
PR: 311
des_old.h redefines crypt:
#define crypt(b,s)\
DES_crypt((b),(s))
This scheme leads to failure, if header files with the OS's true definition
of crypt() are processed _after_ des_old.h was processed. This is e.g. the
case on HP-UX with unistd.h.
As evp.h now again includes des.h (which includes des_old.h), this problem
only came up after this modification.
Solution: move header files (indirectly) including e_os.h before the header
files (indirectly) including evp.h.
Submitted by:
Reviewed by:
PR:
New macros SSL[_CTX]_set_msg_callback_arg().
Message callback imlementation for SSL 3.0/TLS 1.0 (no SSL 2.0 yet).
New '-msg' option for 'openssl s_client' and 'openssl s_server'
that enable a message callback that displays all protocol messages.
In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert if
client_version is smaller than the protocol version in use.
Also change ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0
if the client demanded SSL 3.0 but only TLS 1.0 is enabled; then the
client will at least see that alert.
Fix SSL[_CTX]_ctrl prototype (void * instead of char * for generic
pointer).
Add/update some OpenSSL copyright notices.
Both have per-SSL_CTX defaults.
These new values can be set by calling SSL[_CTX]_[callback_]ctrl
with codes SSL_CTRL_SET_MSG_CALLBACK and SSL_CTRL_SET_MSG_CALLBACK_ARG.
So far, the callback is never actually called.
Also rearrange some SSL_CTX struct members (some exist just in
SSL_CTXs, others are defaults for SSLs and are either copied
during SSL_new, or used if the value in the SSL is not set;
these three classes of members were not in a logical order),
and add some missing assignments to SSL_dup.
See the commit log message for that for more information.
NB: X509_STORE_CTX's use of "ex_data" support was actually misimplemented
(initialisation by "memset" won't/can't/doesn't work). This fixes that but
requires that X509_STORE_CTX_init() be able to handle errors - so its
prototype has been changed to return 'int' rather than 'void'. All uses of
that function throughout the source code have been tracked down and
adjusted.
setting stack (actually, array) values in ex_data. So only increment the
global counters if the underlying CRYPTO_get_ex_new_index() call succeeds.
This change doesn't make "ex_data" right (see the comment at the head of
ex_data.c to know why), but at least makes the source code marginally less
frustrating.
His comments are:
. adds use of replay cache to protect against replay attacks
. adds functions kssl_tgt_is_available() and
kssl_keytab_is_available() which are used within s3_lib.c
and ssl_lib.c to determine at runtime whether or not
KRB5 ciphers can be supported during the current session.
an SSL_CTX's session cache, it is necessary to compare the ssl_version at
the same time (a conflict is defined, courtesy of SSL_SESSION_cmp(), as a
matching id/id_length pair and a matching ssl_version). However, the
SSL_SESSION that will result from the current negotiation does not
necessarily have the same ssl version as the "SSL_METHOD" in use by the
SSL_CTX - part of the work in a handshake is to agree on an ssl version!
This is fixed by having the check function accept an SSL pointer rather
than the SSL_CTX it belongs to.
[Thanks to Lutz for illuminating the full extent of my stupidity]
the ID will be padded out to 16 bytes if the callback attempted to generate
a shorter one. The problem is that the uniqueness checking function used in
callbacks may mistakenly think a 9-byte ID is unique when in fact its
padded 16-byte version is not. This makes the checking function detect
SSLv2 cases, and ensures the padded form is checked rather than the shorter
one passed by the callback.
SSL/TLS session IDs in a server. According to RFC2246, the session ID is an
arbitrary value chosen by the server. It can be useful to have some control
over this "arbitrary value" so as to choose it in ways that can aid in
things like external session caching and balancing (eg. clustering). The
default session ID generation is to fill the ID with random data.
The callback used by default is built in to ssl_sess.c, but registering a
callback in an SSL_CTX or in a particular SSL overrides this. BTW: SSL
callbacks will override SSL_CTX callbacks, and a new SSL structure inherits
any callback set in its 'parent' SSL_CTX. The header comments describe how
this mechanism ticks, and source code comments describe (hopefully) why it
ticks the way it does.
Man pages are on the way ...
[NB: Lutz was also hacking away and helping me to figure out how best to do
this.]
DECLARE/IMPLEMENT macros now exist to create type (and prototype) safe
wrapper functions that avoid the use of function pointer casting yet retain
type-safety for type-specific callbacks. However, most of the usage within
OpenSSL itself doesn't really require the extra function because the hash
and compare callbacks are internal functions declared only for use by the
hash table. So this change catches all those cases and reimplements the
functions using the base-level LHASH prototypes and does per-variable
casting inside those functions to convert to the appropriate item type.
The exception so far is in ssl_lib.c where the hash and compare callbacks
are not static - they're exposed in ssl.h so their prototypes should not be
changed. In this last case, the IMPLEMENT_LHASH_*** macros have been left
intact.
casts) used in the lhash code are about as horrible and evil as they can
be. For starters, the callback prototypes contain empty parameter lists.
Yuck.
This first change defines clearer prototypes - including "typedef"'d
function pointer types to use as "hash" and "compare" callbacks, as well as
the callbacks passed to the lh_doall and lh_doall_arg iteration functions.
Now at least more explicit (and clear) casting is required in all of the
dependant code - and that should be included in this commit.
The next step will be to hunt down and obliterate some of the function
pointer casting being used when it's not necessary - a particularly evil
variant exists in the implementation of lh_doall.
like Malloc, Realloc and especially Free conflict with already existing names
on some operating systems or other packages. That is reason enough to change
the names of the OpenSSL memory allocation macros to something that has a
better chance of being unique, like prepending them with OPENSSL_.
This change includes all the name changes needed throughout all C files.
yet tighter, and also put some heat on the rest of the library by
insisting (correctly) that compare callbacks used in stacks are prototyped
with "const" parameters. This has led to a depth-first explosion of
compiler warnings in the code where 1 constification has led to 3 or 4
more. Fortunately these have all been resolved to completion and the code
seems cleaner as a result - in particular many of the _cmp() functions
should have been prototyped with "const"s, and now are. There was one
little problem however;
X509_cmp() should by rights compare "const X509 *" pointers, and it is now
declared as such. However, it's internal workings can involve
recalculating hash values and extensions if they have not already been
setup. Someone with a more intricate understanding of the flow control of
X509 might be able to tighten this up, but for now - this seemed the
obvious place to stop the "depth-first" constification of the code by
using an evil cast (they have migrated all the way here from safestack.h).
Fortunately, this is the only place in the code where this was required
to complete these type-safety changes, and it's reasonably clear and
commented, and seemed the least unacceptable of the options. Trying to
take the constification further ends up exploding out considerably, and
indeed leads directly into generalised ASN functions which are not likely
to cooperate well with this.
designed for that. This removes the potential error to mix data and
function pointers.
Please note that I'm a little unsure how incorrect calls to the old
ctrl functions should be handled, in som cases. I currently return 0
and that's it, but it may be more correct to generate a genuine error
in those cases.
non-function pointers to function pointers and vice versa.
The current solution is to have unions that describe the
conversion we want to do, and gives us the ability to extract
the type of data we want.
The current solution is a quick fix, and can probably be made
in a more general or elegant way.
yet.
Add a function X509_STORE_CTX_purpose_inherit() which implements the logic
of "inheriting" purpose and trust from a parent structure and using a default:
this will be used in the SSL code and possibly future S/MIME.
Partial documentation of the 'verify' utility. Still need to document how all
the extension checking works and the various error messages.
called sess_cert instead of just cert. This is in preparation of further
changes: Probably often when s->session->sess_cert is used, we should
use s->cert instead; s->session->sess_cert should be a new structure
containing only the stuff that is for just one connection (e.g.
the peer's certificate, which the SSL client implementations currently
store in s->session->[sess_]cert, which is a very confusing thing to do).
Submitted by:
Reviewed by:
PR:
ssl/ssl_lib.c and ssl/ssl.h. At least the double ctx-variable
confused some compilers.
Submitted by: Lennart Bong <lob@kulthea.stacken.kth.se>
Reviewed by: Ralf S. Engelschall
private keys and/or callback functions which directly correspond to their
SSL_CTX_xxx() counterparts but work on a per-connection basis. This is needed
for applications which have to configure certificates on a per-connection
basis (e.g. Apache+mod_ssl) instead of a per-context basis (e.g.
s_server).
For the RSA certificate situation is makes no difference, but for the DSA
certificate situation this fixes the "no shared cipher" problem where the
OpenSSL cipher selection procedure failed because the temporary keys were not
overtaken from the context and the API provided no way to reconfigure them.
The new functions now let applications reconfigure the stuff and they are in
detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new
non-public-API function ssl_cert_instantiate() is used as a helper function
and also to reduce code redundancy inside ssl_rsa.c.
Submitted by: Ralf S. Engelschall
Reviewed by: Ben Laurie
[Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
Fix so that the version number in the master secret, when passed
via RSA, checks that if TLS was proposed, but we roll back to SSLv3
(because the server will not accept higher), that the version number
is 0x03,0x01, not 0x03,0x00
[Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
Submitted by:
Reviewed by:
PR:
1. Add *lots* of missing prototypes for static ssl functions.
2. VC++ doesn't understand the 'LL' suffix for 64 bits constants: change bn.org
3. Add a few missing prototypes in pem.org
Fix mk1mf.pl so it outputs a Makefile that doesn't choke Win95.
Fix mkdef.pl so it doesn't truncate longer names.
1. The already released version was 0.9.1c and not 0.9.1b
2. The next release should be 0.9.2 and not 0.9.1d, because
first the changes are already too large, second we should avoid any more
0.9.1x confusions and third, the Apache version semantics of
VERSION.REVISION.PATCHLEVEL for the version string is reasonable (and here
.2 is already just a patchlevel and not major change).
tVS: ----------------------------------------------------------------------
