Vulnerabilities

MessagePack for Ruby (the msgpack-ruby gem) prior to version 1.8.2 contains a stale-pointer flaw (CWE-416) in MessagePack::Buffer#clear: when the C extension returns an internal read-memory (rmem) page to the shared pool via _msgpack_buffer_shift_chunk, the rmem_last, rmem_end, and rmem_owner fields are left pointing at the freed page. A subsequent Buffer#write on the cleared buffer, combined with a second MessagePack::Buffer in the same process allocating that recycled page, causes both buffers to alias the same memory. An attacker with local access or the ability to influence serialized data in such a process can read or corrupt data belonging to the other buffer; the CVSS 4.0 score of 2.1 assesses the impact as low-severity information disclosure with a local attack vector. Any Ruby application using MessagePack serialization (commonly for caching, RPC, or message handling) on versions before 1.8.2 is affected, but triggering requires a specific clear-then-reuse buffer pattern. No in-the-wild exploitation is known: the flaw is not in CISA KEV, EPSS estimates a 0.2% probability of exploitation within 30 days, and the only public reference is the GitHub advisory GHSA-4mrv-5p47-p938.

Do: Upgrade msgpack-ruby to version 1.8.2 or later. As an interim mitigation, avoid calling MessagePack::Buffer#clear on buffers that are subsequently reused (allocate fresh buffers instead) and audit code paths that clear and reuse MessagePack buffers within a single process. Given the local attack vector, prioritize hosts that run untrusted local code or multi-tenant Ruby processes.

MessagePack is the serializer implementation for Python msgpack.org. Prior to 1.2.1, there is an Out-of-bounds read/crash on Unpacker reuse after a caught error, potentially leading to a DoS attack. If the Unpacker is used repeatedly after an error occurs, the process may crash with a SEGV. This issue has been fixed in version 1.2.1.

MessagePack for Java is a serializer implementation for Java. A denial-of-service vulnerability exists in versions prior to 0.9.11 when deserializing .msgpack files containing EXT32 objects with attacker-controlled payload lengths. While MessagePack-Java parses extension headers lazily, it later trusts the declared EXT payload length when materializing the extension data. When ExtensionValue.getData() is invoked, the library attempts to allocate a byte array of the declared length without enforcing any upper bound. A malicious .msgpack file of only a few bytes can therefore trigger unbounded heap allocation, resulting in JVM heap exhaustion, process termination, or service unavailability. This vulnerability is triggered during model loading / deserialization, making it a model format vulnerability suitable for remote exploitation. The vulnerability enables a remote denial-of-service attack against applications that deserialize untrusted .msgpack model files using MessagePack for Java. A specially crafted but syntactically valid .msgpack file containing an EXT32 object with an attacker-controlled, excessively large payload length can trigger unbounded memory allocation during deserialization. When the model file is loaded, the library trusts the declared length metadata and attempts to allocate a byte array of that size, leading to rapid heap exhaustion, excessive garbage collection, or immediate JVM termination with an OutOfMemoryError. The attack requires no malformed bytes, user interaction, or elevated privileges and can be exploited remotely in real-world environments such as model registries, inference services, CI/CD pipelines, and cloud-based model hosting platforms that accept or fetch .msgpack artifacts. Because the malicious file is extremely small yet valid, it can bypass basic validation and scanning mechanisms, resulting in complete service unavailability and potential cascading failures in production systems. Version 0.9.11 fixes the vulnerability.

msgpack5 is a msgpack v5 implementation for node.js and the browser. In msgpack5 before versions 3.6.1, 4.5.1, and 5.2.1 there is a "Prototype Poisoning" vulnerability. When msgpack5 decodes a map containing a key "__proto__", it assigns the decoded value to __proto__. Object.prototype.__proto__ is an accessor property for the receiver's prototype. If the value corresponding to the key __proto__ decodes to an object or null, msgpack5 sets the decoded object's prototype to that value. An attacker who can submit crafted MessagePack data to a service can use this to produce values that appear to be of other types; may have unexpected prototype properties and methods (for example length, numeric properties, and push et al if __proto__'s value decodes to an Array); and/or may throw unexpected exceptions when used (for example if the __proto__ value decodes to a Map or Date). Other unexpected behavior might be produced for other types. There is no effect on the global prototype. This "prototype poisoning" is sort of a very limited inversion of a prototype pollution attack. Only the decoded value's prototype is affected, and it can only be set to msgpack5 values (though if the victim makes use of custom codecs, anything could be a msgpack5 value). We have not found a way to escalate this to true prototype pollution (absent other bugs in the consumer's code). This has been fixed in msgpack5 version 3.6.1, 4.5.1, and 5.2.1. See the referenced GitHub Security Advisory for an example and more details.