*) Fix additional bug revealed by the NISCC test suite:
Stop bug triggering large recursion when presented with
certain ASN.1 tags (CAN-2003-0851)
[Steve Henson]
Changes between 0.9.6j and 0.9.6k [30 Sep 2003]
*) Fix various bugs revealed by running the NISCC test suite:
Stop out of bounds reads in the ASN1 code when presented with
invalid tags (CAN-2003-0543 and CAN-2003-0544).
If verify callback ignores invalid public key errors don't try to check
certificate signature with the NULL public key.
[Steve Henson]
*) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
if the server requested one: as stated in TLS 1.0 and SSL 3.0
specifications.
[Steve Henson]
*) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
extra data after the compression methods not only for TLS 1.0
but also for SSL 3.0 (as required by the specification).
[Bodo Moeller; problem pointed out by Matthias Loepfe]
*) Change X509_certificate_type() to mark the key as exported/exportable
when it's 512 *bits* long, not 512 bytes.
[Richard Levitte]
Changes between 0.9.6i and 0.9.6j [10 Apr 2003]
*) Countermeasure against the Klima-Pokorny-Rosa extension of
Bleichbacher's attack on PKCS #1 v1.5 padding: treat
a protocol version number mismatch like a decryption error
in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
[Bodo Moeller]
*) Turn on RSA blinding by default in the default implementation
to avoid a timing attack. Applications that don't want it can call
RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
They would be ill-advised to do so in most cases.
[Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
*) Change RSA blinding code so that it works when the PRNG is not
seeded (in this case, the secret RSA exponent is abused as
an unpredictable seed -- if it is not unpredictable, there
is no point in blinding anyway). Make RSA blinding thread-safe
by remembering the creator's thread ID in rsa->blinding and
having all other threads use local one-time blinding factors
(this requires more computation than sharing rsa->blinding, but
avoids excessive locking; and if an RSA object is not shared
between threads, blinding will still be very fast).
[Bodo Moeller]
Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
*) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
via timing by performing a MAC computation even if incorrrect
block cipher padding has been found. This is a countermeasure
against active attacks where the attacker has to distinguish
between bad padding and a MAC verification error. (CAN-2003-0078)
[Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
Martin Vuagnoux (EPFL, Ilion)]
Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
*) New function OPENSSL_cleanse(), which is used to cleanse a section of
memory from it's contents. This is done with a counter that will
place alternating values in each byte. This can be used to solve
two issues: 1) the removal of calls to memset() by highly optimizing
compilers, and 2) cleansing with other values than 0, since those can
be read through on certain media, for example a swap space on disk.
[Geoff Thorpe]
*) Bugfix: client side session caching did not work with external caching,
because the session->cipher setting was not restored when reloading
from the external cache. This problem was masked, when
SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
(Found by Steve Haslam <steve@araqnid.ddts.net>.)
[Lutz Jaenicke]
*) Fix client_certificate (ssl/s2_clnt.c): The permissible total
length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
[Zeev Lieber <zeev-l@yahoo.com>]
*) Undo an undocumented change introduced in 0.9.6e which caused
repeated calls to OpenSSL_add_all_ciphers() and
OpenSSL_add_all_digests() to be ignored, even after calling
EVP_cleanup().
[Richard Levitte]
*) Change the default configuration reader to deal with last line not
being properly terminated.
[Richard Levitte]
*) Change X509_NAME_cmp() so it applies the special rules on handling
DN values that are of type PrintableString, as well as RDNs of type
emailAddress where the value has the type ia5String.
[stefank@valicert.com via Richard Levitte]
*) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
the bitwise-OR of the two for use by the majority of applications
wanting this behaviour, and update the docs. The documented
behaviour and actual behaviour were inconsistent and had been
changing anyway, so this is more a bug-fix than a behavioural
change.
[Geoff Thorpe, diagnosed by Nadav Har'El]
*) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
(the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
[Bodo Moeller]
*) Fix initialization code race conditions in
SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
ssl2_get_cipher_by_char(),
ssl3_get_cipher_by_char().
[Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
*) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
the cached sessions are flushed, as the remove_cb() might use ex_data
contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
(see [openssl.org #212]).
[Geoff Thorpe, Lutz Jaenicke]
*) Fix typo in OBJ_txt2obj which incorrectly passed the content
length, instead of the encoding length to d2i_ASN1_OBJECT.
[Steve Henson]
an operating system does not have a 'make' (ie only bmake), or if the OS
supplied 'make' is sufficiently broken (Irix), this will cause the build to
fail (interestingly enough apparently only if build as a dependency, not
if build from this directory).
Patch Makefiles to use @MAKE@, which then, after patching, is substituted with
the actual ${MAKE} (can't use "MAKE= ${MAKE} -f Makefile.ssl").
While here, tweak Irix configure a bit.
Researchers have discovered a timing attack on RSA keys, to which
OpenSSL is generally vulnerable, unless RSA blinding has been turned
on.
Typically, it will not have been, because it is not easily possible to
do so when using OpenSSL to provide SSL or TLS.
The enclosed patch switches blinding on by default. Applications that
wish to can remove the blinding with RSA_blinding_off(), but this is
not generally advised. It is also possible to disable it completely by
defining OPENSSL_NO_FORCE_RSA_BLINDING at compile-time.
The performance impact of blinding appears to be small (a few
percent).
This problem affects many applications using OpenSSL, in particular,
almost all SSL-enabled Apaches. You should rebuild and reinstall
OpenSSL, and all affected applications.
The Common Vulnerabilities and Exposures project (cve.mitre.org) has
assigned the name CAN-2003-0147 to this issue.
* Add patch from http://www.openssl.org/news/secadv_20030319.txt:
Czech cryptologists Vlastimil Klima, Ondrej Pokorny, and Tomas Rosa
have come up with an extension of the "Bleichenbacher attack" on RSA
with PKCS #1 v1.5 padding as used in SSL 3.0 and TLS 1.0. Their
attack requires the attacker to open millions of SSL/TLS connections
to the server under attack; the server's behaviour when faced with
specially made-up RSA ciphertexts can reveal information that in
effect allows the attacker to perform a single RSA private key
operation on a ciphertext of its choice using the server's RSA key.
Note that the server's RSA key is not compromised in this attack.
* Bump PKGREVISION.
In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
via timing by performing a MAC computation even if incorrrect
block cipher padding has been found. This is a countermeasure
against active attacks where the attacker has to distinguish
between bad padding and a MAC verification error. (CAN-2003-0078)
Bump PKGREVISION.
a stunning DoS vulnerability, fixed in 0.9.6f:
*) Use proper error handling instead of 'assertions' in buffer
overflow checks added in 0.9.6e. This prevents DoS (the
assertions could call abort()).
[Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
Regenerate the netbsd patch. This is now a clean diff against the
vendor tag, with version-number-only changes elided.
Partially revert "crypto/dist/openssl/crypto/rand/randfile.c", version
1.4 (via additional pkgsrc patch), to give this a shot to compile on
NetBSD-1.4.2 and earlier, which had no strlcpy() or strlcat().
Assemble the shared library without "-Bsymbolic", mainly to give this
a shot at linking on NetBSD-a.out (untested).
and also changes the ABI of "libcrypto" and "libssl". (So the shared
library majors and buildlink requirements are bumped, too.) The code
base is now synced perfectly with NetBSD HEAD and netbsd-1-6 branches
as of 2002-08-04, the optimization levels are reduced to "-O2", but
I've retained some of the processor optimization flags and different code
path #defines in the "Configure" script, just to keep things interesting.
The default "certs" directory on NetBSD is now "/etc/openssl/certs", to
give continuity to those who find themselves using the package system's
"openssl" after upgrading a package that formerly used the base system's.
[Suggested by itojun.] The best way to avoid such problems, however, is
to upgrade your base system *first*.
I'm making use of the new and improved build system as much as possible.
This gives us a cleaner way to make shared libraries and real man pages,
but loses many of the symlinks to the openssl binary.
I've culled items from the "CHANGES" file that appear to have security
implications or are particularly interesting for NetBSD users, below.
My comments are marked off with '===>'.
===> This is from the netbsd-20020804-patch
*) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
and get fix the header length calculation.
[Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
Alon Kantor <alonk@checkpoint.com> (and others),
Steve Henson]
Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
*) New option
SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
that was added in OpenSSL 0.9.6d.
As the countermeasure turned out to be incompatible with some
broken SSL implementations, the new option is part of SSL_OP_ALL.
SSL_OP_ALL is usually employed when compatibility with weird SSL
implementations is desired (e.g. '-bugs' option to 's_client' and
's_server'), so the new option is automatically set in many
applications.
[Bodo Moeller]
*) Changes in security patch:
Changes marked "(CHATS)" were sponsored by the Defense Advanced
Research Projects Agency (DARPA) and Air Force Research Laboratory,
Air Force Materiel Command, USAF, under agreement number
F30602-01-2-0537.
*) Add various sanity checks to asn1_get_length() to reject
the ASN1 length bytes if they exceed sizeof(long), will appear
negative or the content length exceeds the length of the
supplied buffer.
[Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
*) Assertions for various potential buffer overflows, not known to
happen in practice.
[Ben Laurie (CHATS)]
*) Various temporary buffers to hold ASCII versions of integers were
too small for 64 bit platforms. (CAN-2002-0655)
[Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
*) Remote buffer overflow in SSL3 protocol - an attacker could
supply an oversized session ID to a client. (CAN-2002-0656)
[Ben Laurie (CHATS)]
*) Remote buffer overflow in SSL2 protocol - an attacker could
supply an oversized client master key. (CAN-2002-0656)
[Ben Laurie (CHATS)]
Changes between 0.9.6c and 0.9.6d [9 May 2002]
*) Implement a countermeasure against a vulnerability recently found
in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
before application data chunks to avoid the use of known IVs
with data potentially chosen by the attacker.
[Bodo Moeller]
Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
*) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
PRNG state recovery was possible based on the output of
one PRNG request appropriately sized to gain knowledge on
'md' followed by enough consecutive 1-byte PRNG requests
to traverse all of 'state'.
1. When updating 'md_local' (the current thread's copy of 'md')
during PRNG output generation, hash all of the previous
'md_local' value, not just the half used for PRNG output.
2. Make the number of bytes from 'state' included into the hash
independent from the number of PRNG bytes requested.
The first measure alone would be sufficient to avoid
Markku-Juhani's attack. (Actually it had never occurred
to me that the half of 'md_local' used for chaining was the
half from which PRNG output bytes were taken -- I had always
assumed that the secret half would be used.) The second
measure makes sure that additional data from 'state' is never
mixed into 'md_local' in small portions; this heuristically
further strengthens the PRNG.
[Bodo Moeller]
*) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
when fixing the server behaviour for backwards-compatible 'client
hello' messages. (Note that the attack is impractical against
SSL 3.0 and TLS 1.0 anyway because length and version checking
means that the probability of guessing a valid ciphertext is
around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
paper.)
Before 0.9.5, the countermeasure (hide the error by generating a
random 'decryption result') did not work properly because
ERR_clear_error() was missing, meaning that SSL_get_error() would
detect the supposedly ignored error.
Both problems are now fixed.
[Bodo Moeller]
Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
===> This is our ABI change.
*) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
with des_encrypt() defined on some operating systems, like Solaris
and UnixWare.
[Richard Levitte]
*) Don't use getenv in library functions when run as setuid/setgid.
New function OPENSSL_issetugid().
[Ulf Moeller]
*) Store verify_result within SSL_SESSION also for client side to
avoid potential security hole. (Re-used sessions on the client side
always resulted in verify_result==X509_V_OK, not using the original
result of the server certificate verification.)
[Lutz Jaenicke]
===> package doesn't doesn't do this. We'll bump major versions
===> as necessary.
*) Make sure that shared libraries get the internal name engine with
the full version number and not just 0. This should mark the
shared libraries as not backward compatible. Of course, this should
be changed again when we can guarantee backward binary compatibility.
[Richard Levitte]
*) Rework the system to generate shared libraries:
- Make note of the expected extension for the shared libraries and
if there is a need for symbolic links from for example libcrypto.so.0
to libcrypto.so.0.9.7. There is extended info in Configure for
that.
- Make as few rebuilds of the shared libraries as possible.
- Still avoid linking the OpenSSL programs with the shared libraries.
- When installing, install the shared libraries separately from the
static ones.
