Vulnerabilities > CVE-2022-33682 - Improper Certificate Validation vulnerability in Apache Pulsar

047910
CVSS 5.9 - MEDIUM
Attack vector
NETWORK
Attack complexity
HIGH
Privileges required
NONE
Confidentiality impact
HIGH
Integrity impact
NONE
Availability impact
NONE
network
high complexity
apache
CWE-295

Summary

TLS hostname verification cannot be enabled in the Pulsar Broker's Java Client, the Pulsar Broker's Java Admin Client, the Pulsar WebSocket Proxy's Java Client, and the Pulsar Proxy's Admin Client leaving intra-cluster connections and geo-replication connections vulnerable to man in the middle attacks, which could leak credentials, configuration data, message data, and any other data sent by these clients. The vulnerability is for both the pulsar+ssl protocol and HTTPS. An attacker can only take advantage of this vulnerability by taking control of a machine 'between' the client and the server. The attacker must then actively manipulate traffic to perform the attack by providing the client with a cryptographically valid certificate for an unrelated host. This issue affects Apache Pulsar Broker, Proxy, and WebSocket Proxy versions 2.7.0 to 2.7.4; 2.8.0 to 2.8.3; 2.9.0 to 2.9.2; 2.10.0; 2.6.4 and earlier.

Common Weakness Enumeration (CWE)

Common Attack Pattern Enumeration and Classification (CAPEC)

  • Creating a Rogue Certificate Authority Certificate
    An attacker exploits a weakness in the MD5 hash algorithm (weak collision resistance) to generate a certificate signing request (CSR) that contains collision blocks in the "to be signed" part. The attacker specially crafts two different, but valid X.509 certificates that when hashed with the MD5 algorithm would yield the same value. The attacker then sends the CSR for one of the certificates to the Certification Authority which uses the MD5 hashing algorithm. That request is completely valid and the Certificate Authority issues an X.509 certificate to the attacker which is signed with its private key. An attacker then takes that signed blob and inserts it into another X.509 certificate that the attacker generated. Due to the MD5 collision, both certificates, though different, hash to the same value and so the signed blob works just as well in the second certificate. The net effect is that the attackers' second X.509 certificate, which the Certification Authority has never seen, is now signed and validated by that Certification Authority. To make the attack more interesting, the second certificate could be not just a regular certificate, but rather itself a signing certificate. Thus the attacker is able to start their own Certification Authority that is anchored in its root of trust in the legitimate Certification Authority that has signed the attackers' first X.509 certificate. If the original Certificate Authority was accepted by default by browsers, so will now the Certificate Authority set up by the attacker and of course any certificates that it signs. So the attacker is now able to generate any SSL certificates to impersonate any web server, and the user's browser will not issue any warning to the victim. This can be used to compromise HTTPS communications and other types of systems where PKI and X.509 certificates may be used (e.g., VPN, IPSec) .