llama.cpp: Unauthenticated RCE via GRAPH_COMPUTE buffer=0 bypass in llama.cpp RPC backend_CVE-2026-34159

{
    "lastseen": "",
    "description": "llama.cpp is an inference of several LLM models in C/C++. Prior to version b8492, the RPC backend's deserialize_tensor() skips all bounds validation when a tensor's buffer field is 0. An unauthenticated attacker can read and write arbitrary process memory via crafted GRAPH_COMPUTE messages. Combined with pointer leaks from ALLOC_BUFFER/BUFFER_GET_BASE, this gives full ASLR bypass and remote code execution. No authentication required, just TCP access to the RPC server port. This issue has been patched in version b8492.",
    "published": "2026-04-01T16:59:59.967Z",
    "modified": "2026-04-02T03:56:11.820Z",
    "type": "cve",
    "title": "llama.cpp: Unauthenticated RCE via GRAPH_COMPUTE buffer=0 bypass in llama.cpp RPC backend",
    "source": "GitHub_M",
    "references": "https://github.com/ggml-org/llama.cpp/security/advisories/GHSA-j8rj-fmpv-wcxw\nhttps://github.com/ggml-org/llama.cpp/pull/20908\nhttps://github.com/ggml-org/llama.cpp/commit/39bf0d3c6a95803e0f41aaba069ffbee26721042",
    "id": "CVE-2026-34159",
    "bulletinFamily": "",
    "cwe": [
        "CWE-119"
    ],
    "cvelist": null,
    "sourceData": "ggml-org llama.cpp < b8492",
    "sourceHref": "",
    "cvss": {
        "score": 9.8,
        "severity": "CRITICAL",
        "vector": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H",
        "version": "3.1"
    },
    "cvss2": [],
    "cvss3": {
        "version": "",
        "vectorString": "",
        "baseScore": 0,
        "baseSeverity": "",
        "attackVector": "",
        "attackComplexity": "",
        "privilegesRequired": "",
        "userInteraction": "",
        "scope": "",
        "confidentialityImpact": "",
        "integrityImpact": "",
        "availabilityImpact": "",
        "cvssV3": {
            "version": "",
            "vectorString": "",
            "baseScore": 0,
            "baseSeverity": "",
            "attackVector": "",
            "attackComplexity": "",
            "privilegesRequired": "",
            "userInteraction": "",
            "scope": "",
            "confidentialityImpact": "",
            "integrityImpact": "",
            "availabilityImpact": ""
        }
    },
    "href": "",
    "category_name": "CVE",
    "post_link": "",
    "product": "llama.cpp",
    "version": "< b8492",
    "vendor": "ggml-org",
    "ai_description": "",
    "ai_severity": "",
    "ai_vendor": "",
    "ai_product": "",
    "ai_version": "",
    "ai_score": 0
}