The asn1d2ireadbio function in crypto/asn1/ad2ifp.c in OpenSSL before 0.9.8v, 1.0.0 before 1.0.0i, and 1.0.1 before 1.0.1a does not properly interpret integer data, which allows remote attackers to conduct buffer overflow attacks, and cause a denial of service (memory corruption) or possibly have unspecified other impact, via crafted DER data, as demonstrated by an X.509 certificate or an RSA public key.

Integer underflow in OpenSSL before 0.9.8x, 1.0.0 before 1.0.0j, and 1.0.1 before 1.0.1c, when TLS 1.1, TLS 1.2, or DTLS is used with CBC encryption, allows remote attackers to cause a denial of service (buffer over-read) or possibly have unspecified other impact via a crafted TLS packet that is not properly handled during a certain explicit IV calculation.

crypto/bn/bnnist.c in OpenSSL before 0.9.8h on 32-bit platforms, as used in stunnel and other products, in certain circumstances involving ECDH or ECDHE cipher suites, uses an incorrect modular reduction algorithm in its implementation of the P-256 and P-384 NIST elliptic curves, which allows remote attackers to obtain the private key of a TLS server via multiple handshake attempts.

Draft advisory from OpenSSL team:

OpenSSL Security Advisory [07-Jan-2009]

Incorrect checks for malformed signatures -------------------------------------------

Several functions inside OpenSSL incorrectly checked the result after calling the EVPVerifyFinal function, allowing a malformed signature to be treated as a good signature rather than as an error. This issue affected the signature checks on DSA and ECDSA keys used with SSL/TLS.

One way to exploit this flaw would be for a remote attacker who is in control of a malicious server or who can use a 'man in the middle' attack to present a malformed SSL/TLS signature from a certificate chain to a vulnerable client, bypassing validation.

This vulnerability is tracked as CVE-2008-5077.

The OpenSSL security team would like to thank the Google Security Team for reporting this issue.

Who is affected? -----------------

Everyone using OpenSSL releases prior to 0.9.8j as an SSL/TLS client when connecting to a server whose certificate contains a DSA or ECDSA key.

Use of OpenSSL as an SSL/TLS client when connecting to a server whose certificate uses an RSA key is NOT affected.

Verification of client certificates by OpenSSL servers for any key type is NOT affected.

Recommendations for users of OpenSSL ------------------------------------

Users of OpenSSL 0.9.8 should update to the OpenSSL 0.9.8j release which contains a patch to correct this issue.

The patch used is also appended to this advisory for users or distributions who wish to backport this patch to versions they build from source. Please note: this patch also includes fixes for a few other cases where return codes are not correctly checked, but these do not have a security implication

Recommendations for projects using OpenSSL ------------------------------------------

Projects and products using OpenSSL should audit any use of the routine EVPVerifyFinal() to ensure that the return code is being correctly handled. As documented, this function returns 1 for a successful verification, 0 for failure, and -1 for an error.

General recommendations -----------------------

Any SSL/TLS server with clients that OpenSSL to verify DSA or ECDSA certificates, regardless of the software used by the server, should either ensure that all clients are upgraded or should stop using DSA/ECDSA certificates. Note that unless certificates are revoked (and clients check for revocation) impersonation will still be possible until the certificate expires.

A flaw was found in the way that OpenSSL handled OCSP response verification, which could be exploited to conduct a denial of service attack. This flaw affects all versions of OpenSSL and is fixed in versions 1.0.1d, 1.0.0k, and 0.9.8y.

External References:

http://www.openssl.org/news/secadv20130205.txt

The mimeparamcmp function in crypto/asn1/asnmime.c in OpenSSL before 0.9.8u and 1.x before 1.0.0h allows remote attackers to cause a denial of service (NULL pointer dereference and application crash) via a crafted S/MIME message, a different vulnerability than CVE-2006-7250.

The implementation of Cryptographic Message Syntax (CMS) and PKCS #7 in OpenSSL before 0.9.8u and 1.x before 1.0.0h does not properly restrict certain oracle behavior, which makes it easier for context-dependent attackers to decrypt data via a Million Message Attack (MMA) adaptive chosen ciphertext attack.

Memory leak in the zlibstatefulfinish function in crypto/comp/czlib.c in OpenSSL 0.9.8l and earlier and 1.0.0 Beta through Beta 4 allows remote attackers to cause a denial of service (memory consumption) via vectors that trigger incorrect calls to the CRYPTOcleanupallexdata function, as demonstrated by use of SSLv3 and PHP with the Apache HTTP Server, a related issue to CVE-2008-1678.

The SSL 3.0 implementation in OpenSSL before 0.9.8s and 1.x before 1.0.0f does not properly initialize data structures for block cipher padding, which might allow remote attackers to obtain sensitive information by decrypting the padding data sent by an SSL peer.

The Server Gated Cryptography (SGC) implementation in OpenSSL before 0.9.8s and 1.x before 1.0.0f does not properly handle handshake restarts, which allows remote attackers to cause a denial of service (CPU consumption) via unspecified vectors.

OpenSSL before 0.9.8s and 1.x before 1.0.0f, when RFC 3779 support is enabled, allows remote attackers to cause a denial of service (assertion failure) via an X.509 certificate containing certificate-extension data associated with (1) IP address blocks or (2) Autonomous System (AS) identifiers.

The DTLS implementation in OpenSSL before 0.9.8s and 1.x before 1.0.0f performs a MAC check only if certain padding is valid, which makes it easier for remote attackers to recover plaintext via a padding oracle attack.

OpenSSL before 0.9.8j, when SSLOPNETSCAPEREUSECIPHERCHANGEBUG is enabled, does not prevent modification of the ciphersuite in the session cache, which allows remote attackers to force the use of a disabled cipher via vectors involving sniffing network traffic to discover a session identifier, a different vulnerability than CVE-2010-4180.

The elliptic curve cryptography (ECC) subsystem in OpenSSL 1.0.0d and earlier, when the Elliptic Curve Digital Signature Algorithm (ECDSA) is used for the ECDHEECDSA cipher suite, does not properly implement curves over binary fields, which makes it easier for context-dependent attackers to determine private keys via a timing attack and a lattice calculation.