CVE-2025-66455

{"id": "CVE-2025-66455", "cveTags": [], "metrics": {"ssvcV203": [{"source": "134c704f-9b21-4f2e-91b3-4a467353bcc0", "ssvcData": {"id": "CVE-2025-66455", "role": "CISA Coordinator", "options": [{"exploitation": "none"}, {"automatable": "yes"}, {"technicalImpact": "total"}], "version": "2.0.3", "timestamp": "2026-09-18T19:54:08.825899Z"}}], "cvssMetricV31": [{"type": "Secondary", "source": "security-advisories@github.com", "cvssData": {"scope": "UNCHANGED", "version": "3.1", "baseScore": 9.8, "attackVector": "NETWORK", "baseSeverity": "CRITICAL", "vectorString": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H", "integrityImpact": "HIGH", "userInteraction": "NONE", "attackComplexity": "LOW", "availabilityImpact": "HIGH", "privilegesRequired": "NONE", "confidentialityImpact": "HIGH"}, "impactScore": 5.9, "exploitabilityScore": 3.9}]}, "affected": [{"source": "security-advisories@github.com", "affectedData": [{"vendor": "InternLM", "product": "lmdeploy", "versions": [{"status": "affected", "version": ">= 0.9.2, < 0.16.0"}]}]}], "published": "2026-09-18T18:17:04.420", "references": [{"url": "https://github.com/InternLM/lmdeploy/commit/f05b4ad8bf2e2d84101a1d63b3c44fadd99223b2", "source": "security-advisories@github.com"}, {"url": "https://github.com/InternLM/lmdeploy/releases/tag/v0.16.0", "source": "security-advisories@github.com"}, {"url": "https://github.com/InternLM/lmdeploy/security/advisories/GHSA-2vh9-42vm-xmv2", "source": "security-advisories@github.com"}], "vulnStatus": "Awaiting Analysis", "weaknesses": [{"type": "Secondary", "source": "security-advisories@github.com", "description": [{"lang": "en", "value": "CWE-502"}]}], "descriptions": [{"lang": "en", "value": "LMDeploy is a toolkit for compressing, deploying, and serving large language models. Starting in version 0.9.2 and prior to version 0.16.0, LMDeploy's PyTorch DistServe/PD-disaggregation control plane used `recv_pyobj()` to deserialize messages received through a ZeroMQ PULL socket. PyZMQ implements `recv_pyobj()` using Python pickle deserialization, which can execute arbitrary code while reconstructing an object. The peer address used by the receiver was supplied through the `POST /distserve/p2p_connect` HTTP endpoint. An attacker who could reach an affected DistServe API server could cause the server to connect to an attacker-controlled ZeroMQ endpoint and deserialize a crafted pickle payload. API-key authentication is not enabled unless the operator explicitly configures it. As a result, affected DistServe deployments without API keys allowed unauthenticated remote code execution with the privileges of the LMDeploy serving process. This issue affects the PyTorch backend when PD-disaggregation/DistServe is enabled. Ordinary deployments that do not use the affected disaggregated-serving path do not expose this data flow. The fix was released in LMDeploy 0.16.0. Users who cannot upgrade immediately should prevent untrusted clients from reaching `/distserve/*` endpoints, restrict the DistServe HTTP and ZeroMQ control planes to trusted cluster networks, configure API-key authentication, and block arbitrary outbound ZeroMQ connections from serving nodes. These measures reduce exposure but do not make pickle deserialization safe."}, {"lang": "es", "value": "LMDeploy es un kit de herramientas para comprimir, desplegar y servir modelos de lenguaje grandes. A partir de la versi\u00f3n 0.9.2 y versiones anteriores a la versi\u00f3n 0.16.0, el plano de control PyTorch DistServe/PD-disaggregation de LMDeploy utilizaba 'recv_pyobj()' para deserializar mensajes recibidos a trav\u00e9s de un socket ZeroMQ PULL. PyZMQ implementa 'recv_pyobj()' utilizando la deserializaci\u00f3n de Python pickle, que puede ejecutar c\u00f3digo arbitrario al reconstruir un objeto. La direcci\u00f3n del par utilizada por el receptor se suministraba a trav\u00e9s del endpoint HTTP 'POST /distserve/p2p_connect'. Un atacante que pudiera alcanzar un servidor API DistServe afectado podr\u00eda hacer que el servidor se conectara a un endpoint ZeroMQ controlado por el atacante y deserializara una carga \u00fatil pickle manipulada. La autenticaci\u00f3n con clave API no est\u00e1 habilitada a menos que el operador la configure expl\u00edcitamente. Como resultado, las implementaciones de DistServe afectadas sin claves API permit\u00edan la ejecuci\u00f3n remota de c\u00f3digo no autenticada con los privilegios del proceso de servicio de LMDeploy. Este problema afecta al backend de PyTorch cuando PD-disaggregation/DistServe est\u00e1 habilitado. Las implementaciones ordinarias que no utilizan la ruta de servicio desagregado afectada no exponen este flujo de datos. La correcci\u00f3n se lanz\u00f3 en LMDeploy 0.16.0. Los usuarios que no puedan actualizar de inmediato deben evitar que clientes no confiables accedan a los endpoints '/distserve/*', restringir los planos de control HTTP y ZeroMQ de DistServe a redes de cl\u00faster confiables, configurar la autenticaci\u00f3n con clave API y bloquear las conexiones ZeroMQ salientes arbitrarias desde los nodos de servicio. Estas medidas reducen la exposici\u00f3n, pero no hacen que la deserializaci\u00f3n de pickle sea segura."}], "lastModified": "2026-10-14T13:10:00.187", "sourceIdentifier": "security-advisories@github.com"}