GHSA-VC6M-HM49-G9QG
vLLM: Quadratic Time Complexity in Input Token Processing leads to denial of service
Ciclo de vida de la vulnerabilidad
| Fecha | Fase | Evento | Detalle |
|---|---|---|---|
| Remediación | Parche 0.8.5 | vllm · -1 d desde divulgación | |
| Divulgación | Publicada | GHSA-VC6M-HM49-G9QG divulgada | |
| Divulgación | Modificada | Última actualización del registro | |
| Monitoreo | Detectada | Ingresó al índice |
### Summary
A critical performance vulnerability has been identified in the input preprocessing logic of the multimodal tokenizer. The code dynamically replaces placeholder tokens (e.g., <|audio_*|>, <|image_*|>) with repeated tokens based on precomputed lengths. Due to inefficient list concatenation operations, the algorithm exhibits quadratic time complexity (O(n²)), allowing malicious actors to trigger resource exhaustion via specially crafted inputs.
### Details
Affected Component: input_processor_for_phi4mm function.
https://github.com/vllm-project/vllm/blob/8cac35ba435906fb7eb07e44fe1a8c26e8744f4e/vllm/model_executor/models/phi4mm.py#L1182-L1197
The code modifies the input_ids list in-place using input_ids = input_ids[:i] + tokens + input_ids[i+1:]. Each concatenation operation copies the entire list, leading to O(n) operations per replacement. For k placeholders expanding to m tokens, total time becomes O(kmn), approximating O(n²) in worst-case scenarios.
### PoC
Test data demonstrates exponential time growth:
test_cases = [100, 200, 400, 800, 1600, 3200, 6400]
run_times = [0.002, 0.007, 0.028, 0.136, 0.616, 2.707, 11.854] # secondsDoubling input size increases runtime by ~4x (consistent with O(n²)).
### Impact
Denial-of-Service (DoS): An attacker could submit inputs with many placeholders (e.g., 10,000 <|audio_1|> tokens), causing CPU/memory exhaustion.
Example: 10,000 placeholders → ~100 million operations.
### Remediation Recommendations
Precompute all placeholder positions and expansion lengths upfront.
Replace dynamic list concatenation with a single preallocated array.
# Pseudocode for O(n) solution
new_input_ids = []
for token in input_ids:
if token is placeholder:
new_input_ids.extend([token] * precomputed_length)
else:
new_input_ids.append(token)Incluye detalles técnicos publicados por el investigador. Úselos solo para validar y corregir sus propios sistemas.
Rangos afectados
Estado del parche por paquete
| Paquete / producto | Estado | Corregida en | Desde divulgación | Última versión |
|---|---|---|---|---|
pypivllm |
Corregida | 0.8.5 | -1 d | 0.34.0 no afectada |
0.8.5Arsenal público
Sin exploits, módulos ni PoC públicos conocidos a la fecha de verificación.
https://nvd.nist.gov/vuln/detail/CVE-2025-46560https://github.com/advisories/GHSA-vc6m-hm49-g9qghttps://github.com/vllm-project/vllm/security/advisories/GHSA-vc6m-hm49-g9qghttps://github.com/pypa/advisory-database/tree/main/vulns/vllm/PYSEC-2026-2022.yamlhttps://github.com/vllm-project/vllmhttps://github.com/vllm-project/vllm/blob/8cac35ba435906fb7eb07e44fe1a8c26e8744f4e/vllm/model_executor/models/phi4mm.py#L1182-L1197https://pypi.org/project/vllmLas referencias externas se muestran para copiar: cve.llama.ec no lo redirige a otros sitios.
