Deserialization of Untrusted Data Affecting verl package, versions [,0.4.0)

verl is a verl: Volcano Engine Reinforcement Learning for LLM

Affected versions of this package are vulnerable to Deserialization of Untrusted Data via the torch.load function in the model_merger.py script when processing user-supplied .pt files with weights_only=False. An attacker can achieve arbitrary code execution by providing a maliciously crafted model file that is loaded by the script. This is only exploitable if the victim downloads and places a malicious model file in a local directory with a specific filename pattern.

Note: In version 0.4.0, model_merger.py was removed.

Serialization is a process of converting an object into a sequence of bytes which can be persisted to a disk or database or can be sent through streams. The reverse process of creating object from sequence of bytes is called deserialization. Serialization is commonly used for communication (sharing objects between multiple hosts) and persistence (store the object state in a file or a database). It is an integral part of popular protocols like Remote Method Invocation (RMI), Java Management Extension (JMX), Java Messaging System (JMS), Action Message Format (AMF), Java Server Faces (JSF) ViewState, etc.

Deserialization of untrusted data (CWE-502) is when the application deserializes untrusted data without sufficiently verifying that the resulting data will be valid, thus allowing the attacker to control the state or the flow of the execution.

• Vulnerability Report
• Vulnerable Code