CVE-2026-54235

## Summary All temperature validation gates use comparison operators (`<`, `>`), which silently evaluate to `False` for `NaN` and for positive `Infinity` in Python's IEEE 754 float semantics. Both values pass every guard and propagate to GPU sampling kernels, where they produce undefined behavior or CUDA errors that can crash the inference worker. Note: `-Infinity` is correctly caught. ## Root Cause `sampling_params.py:384`: ```python if 0 < self.temperature < _MAX_TEMP: # NaN → False; +Inf → False ``` `sampling_params.py:462`: ```python if self.temperature < 0.0: # NaN → False; +Inf → False raise VLLMValidationError(...) ``` No `math.isnan()` or `math.isinf()` check exists anywhere in `sampling_params.py`. Python semantics (verified): `float('nan') < 0.0` → `False`, `float('inf') < 0.0` → `False`. ## Impact Crash of inference worker on GPU kernel execution with NaN/Inf softmax input, degrading service for all concurrent users. ## Remediation Add `math.isfinite(self.temperature)` check in `_verify_args()`. Reject non-finite float values with a 400 error. ## Fix A fix for this vulnerability was merged here: https://github.com/vllm-project/vllm/pull/45116

## Summary All temperature validation gates use comparison operators (`<`, `>`), which silently evaluate to `False` for `NaN` and for positive `Infinity` in Python's IEEE 754 float semantics. Both values pass every guard and propagate to GPU sampling kernels, where they produce undefined behavior or CUDA errors that can crash the inference worker. Note: `-Infinity` is correctly caught. ## Root Cause `sampling_params.py:384`: ```python if 0 < self.temperature < _MAX_TEMP: # NaN → False; +Inf → False ``` `sampling_params.py:462`: ```python if self.temperature < 0.0: # NaN → False; +Inf → False raise VLLMValidationError(...) ``` No `math.isnan()` or `math.isinf()` check exists anywhere in `sampling_params.py`. Python semantics (verified): `float('nan') < 0.0` → `False`, `float('inf') < 0.0` → `False`. ## Impact Crash of inference worker on GPU kernel execution with NaN/Inf softmax input, degrading service for all concurrent users. ## Remediation Add `math.isfinite(self.temperature)` check in `_verify_args()`. Reject non-finite float values with a 400 error. ## Fix A fix for this vulnerability was merged here: https://github.com/vllm-project/vllm/pull/45116

vLLM es un motor de inferencia y servicio para modelos de lenguaje grandes (LLMs). Antes de 0.23.1rc0, las compuertas de validación de temperatura de ll usan operadores de comparación (<, >), que se evalúan silenciosamente como Falso para NaN y para Infinito positivo en la semántica de punto flotante IEEE 754 de Python. Ambos valores pasan cada protección y se propagan a los kernels de muestreo de GPU, donde producen un comportamiento indefinido o errores de CUDA que pueden bloquear el trabajador de inferencia. Esta vulnerabilidad está corregida en 0.23.1rc0.