CVE-2016-2849

Botan before 1.10.13 and 1.11.x before 1.11.29 do not use a constant-time algorithm to perform a modular inverse on the signature nonce k, which might allow remote attackers to obtain ECDSA secret keys via a timing side-channel attack.

Botan before 1.10.13 and 1.11.x before 1.11.29 do not use a constant-time algorithm to perform a modular inverse on the signature nonce k, which might allow remote attackers to obtain ECDSA secret keys via a timing side-channel attack.

Botan before 1.10.13 and 1.11.x before 1.11.29 do not use a constant-time algorithm to perform a modular inverse on the signature nonce k, which might allow remote attackers to obtain ECDSA secret keys via a timing side-channel attack.

Botan before 1.10.13 and 1.11.x before 1.11.29 do not use a constant-time algorithm to perform a modular inverse on the signature nonce k, which might allow remote attackers to obtain ECDSA secret keys via a timing side-channel attack.

Botan before 1.10.13 and 1.11.x before 1.11.29 do not use a constant-time algorithm to perform a modular inverse on the signature nonce k, which might allow remote attackers to obtain ECDSA secret keys via a timing side-channel attack.

Botan before 1.10.13 and 1.11.x before 1.11.29 do not use a constant-time algorithm to perform a modular inverse on the signature nonce k, which might allow remote attackers to obtain ECDSA secret keys via a timing side-channel attack.

Updated botan packages fix security vulnerabilities:

During RSA decryption, how long decoding of PKCS #1 v1.5 padding took was input dependent. If these differences could be measured by an attacker, it could be used to mount a Bleichenbacher million-message attack (CVE-2015-7827).

ECDSA (and DSA) signature algorithms perform a modular inverse on the signature nonce k. The modular inverse algorithm used had input dependent loops, and it is possible a side channel attack could recover sufficient information about the nonce to eventually recover the ECDSA secret key (CVE-2016-2849).

These are all security issues fixed in the libbotan-1_10-1-1.10.13-1.1 package on the GA media of openSUSE Tumbleweed.

This update for Botan to version 1.10.9 fixes the following issues:

These security issues were fixed:

• CVE-2015-5726: The BER decoder in Botan 0.10.x allowed remote attackers to cause a denial of service (application crash) via an empty BIT STRING in ASN.1 data (bsc#968025).
• CVE-2015-5727: The BER decoder in Botan 1.10.x allowed remote attackers to cause a denial of service (memory consumption) via unspecified vectors, related to a length field (bsc#968026).
• CVE-2015-7827: Botan make it easier for remote attackers to conduct million-message attacks by measuring time differences, related to decoding of PKCS#1 padding (bsc#968030).
• CVE-2016-2849: Botan do not use a constant-time algorithm to perform a modular inverse on the signature nonce k, which might allowed remote attackers to obtain ECDSA secret keys via a timing side-channel attack (bsc#977420).
• CVE-2016-9132: In Botan 1.8.0 when decoding BER data an integer overflow could occur, which would cause an incorrect length field to be computed. Some API callers may use the returned (incorrect and attacker controlled) length field in a way which later caused memory corruption or other failure (bsc#1013209).
• CVE-2016-2194: The ressol function in Botan allowed remote attackers to cause a denial of service (infinite loop) via unspecified input to the OS2ECP function, related to a composite modulus (bsc#965621).
• CVE-2016-2195: Integer overflow in the PointGFp constructor in Botan allowed remote attackers to overwrite memory and possibly execute arbitrary code via a crafted ECC point, which triggers a heap-based buffer overflow (bsc#965620).
• CVE-2017-2801: Incorrect comparison in X.509 DN strings (bsc#1033605).
• CVE-2014-9742: The Miller-Rabin primality check in Botan improperly used a single random base, which made it easier for remote attackers to defeat cryptographic protection mechanisms via a DH group (bsc#974521).

These non-security issues were fixed:

• Fixed EAX tag verification to run in constant time
• The default TLS policy now disables SSLv3.
• A crash could have occured when reading from a blocking random device if the device initially indicated that entropy was available but a concurrent process drained the entropy pool before the read was initiated.
• Fixed decoding indefinite length BER constructs that contain a context sensitive tag of zero.
• The key length limit on HMAC has been raised to 512 bytes, allowing the use of very long passphrases with PBKDF2.
• OAEP had two bugs, one of which allowed it to be used even if the key was too small, and the other of which would cause a crash during decryption if the EME data was too large for the associated key.