axTLS v2.1.5 was discovered to contain a heap buffer overflow in the bi_import function in axtls-code/crypto/bigint.c. This vulnerability allows attackers to cause a Denial of Service (DoS) when parsing a private key.

process_certificate in tls1.c in Cameron Hamilton-Rich axTLS through 2.1.5 has a Buffer Overflow via a crafted TLS certificate handshake message with zero certificates.

tls1.c in Cameron Hamilton-Rich axTLS before 2.1.5 has a Buffer Overflow via a crafted sequence of TLS packets because the need_bytes value is mismanaged.

The asn1_signature function in asn1.c in Cameron Hamilton-Rich axTLS through 2.1.5 has a Buffer Overflow that allows remote attackers to cause a denial of service (memory and CPU consumption) via a crafted certificate in the TLS certificate handshake message, because the result of get_asn1_length() is not checked for a minimum or maximum size.

In sig_verify() in x509.c in axTLS version 2.1.3 and before, the PKCS#1 v1.5 signature verification does not properly verify the ASN.1 metadata. Consequently, a remote attacker can forge signatures when small public exponents are being used, which could lead to impersonation through fake X.509 certificates. This is an even more permissive variant of CVE-2006-4790 and CVE-2014-1568.

In sig_verify() in x509.c in axTLS version 2.1.3 and before, the PKCS#1 v1.5 signature verification does not reject excess data after the hash value. Consequently, a remote attacker can forge signatures when small public exponents are being used, which could lead to impersonation through fake X.509 certificates. This is a variant of CVE-2006-4340.

In sig_verify() in x509.c in axTLS version 2.1.3 and before, the PKCS#1 v1.5 signature verification blindly trusts the declared lengths in the ASN.1 structure. Consequently, when small public exponents are being used, a remote attacker can generate purposefully crafted signatures (and put them on X.509 certificates) to induce illegal memory access and crash the verifier.

axTLS version 1.5.3 has a coding error in the ASN.1 parser resulting in the year (19)50 of UTCTime being misinterpreted as 2050.