vLLM from 0.8.5 before 0.31.0 has a denial of service flaw. LogprobsTensors.empty_cpu() in vllm/v1/outputs.py allocates an uninitialized int32 tensor, and the V1 model runner leaves prompt positions reported as computed by a KV connector unwritten. A client sharing the engine can send prompt_logprobs (or echo with logprobs) against a connector-cached prefix, making tokenizer decode raise OverflowError, which crashes EngineCore and drops all co-tenant requests.
{
"cna_assigner": "VulnCheck",
"cwe_ids": [
"CWE-908"
],
"osv_generated_from": "https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/116xxx/CVE-2026-116180.json"
}{
"extracted_events": [
{
"range": [
{
"introduced": "0.8.5"
},
{
"fixed": "0.31.0"
}
],
"source": "AFFECTED_FIELD",
"strategy": "StandardRange"
},
{
"cpe": "cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:*",
"range": [
{
"introduced": "0.8.5"
},
{
"fixed": "0.31.0"
}
],
"source": "CPE_RANGE"
}
]
}