llamaCVEVulnerability Intelligence
/
Advisory · GitHub · PyPI · vllm

GHSA-GRG2-63FW-F2QR

vLLM is vulnerable to DoS in Idefics3 vision models via image payload with ambiguous dimensions

Parche disponible
Descargar PNG JSON
CVSS
—
medium
Fuente: —
ExplotaciónSin evidenciaNo figura en CISA KEV
Exploits públicos0Ninguno conocido
Parche1 versiónPrimer parche a -41 d de la divulgación
Exposición310 dDesde la divulgación, 13 ene 2026, 1:44 p. m.
Detección307 dDesde la divulgación hasta su ingreso al índice
Timeline

Ciclo de vida de la vulnerabilidad

Hoy
Divulgación
Publicada
Modificada
Explotación
Remediación
Parche 0.12.0
Monitoreo
Detectada
3 dic 25
13 ene 26
10 sept 26
17 nov 26
19 nov 26
3 dic 25
1 mar 26
28 may 26
24 ago 26
19 nov 26
FechaFaseEventoDetalle
Remediación Parche 0.12.0 vllm · -41 d desde divulgación
Divulgación Publicada GHSA-GRG2-63FW-F2QR divulgada
Divulgación Modificada Última actualización del registro
Monitoreo Detectada Ingresó al índice
Resumen
Descripción del advisory
Texto original · GitHub · en inglés

### Summary
Users can crash the vLLM engine serving multimodal models that use the _Idefics3_ vision model implementation by sending a specially crafted 1x1 pixel image. This causes a tensor dimension mismatch that results in an unhandled runtime error, leading to complete server termination.

### Details
The vulnerability is triggered when the image processor encounters a 1x1 pixel image with shape (1, 1, 3) in HWC (Height, Width, Channel) format. Due to the ambiguous dimensions, the processor incorrectly assumes the image is in CHW (Channel, Height, Width) format with shape (3, H, W). This misinterpretation causes an incorrect calculation of the number of image patches, resulting in a fatal tensor split operation failure.

Crash location: `vllm/model_executor/models/idefics3.py` line 672:

def _process_image_input(self, image_input: ImageInputs) -> torch.Tensor | list[torch.Tensor]:
    # ...
    num_patches = image_input["num_patches"]
    return [e.flatten(0, 1) for e in image_features.split(num_patches.tolist())]

The `split()` call fails because the computed `num_patches` value (17) does not match the actual tensor dimension (9):

RuntimeError: split_with_sizes expects split_sizes to sum exactly to 9 
(input tensor's size at dimension 0), but got split_sizes=[17]

This unhandled exception terminates the EngineCore process, crashing the server.

#### Affected Models
Any model using the Idefics3 architecture. The vulnerability was tested with `HuggingFaceTB/SmolVLM-Instruct`.

### Impact
Denial of service by crashing the engine

### Mitigation
Validating the input:

def _validate_image_dimensions(self, image_shape):
    h, w = image_shape[:2] if len(image_shape) == 3 else image_shape
    if h < MIN_IMAGE_SIZE or w < MIN_IMAGE_SIZE:
        raise ValueError(f"Image dimensions too small: {h}x{w}")

Managing the exception:

try:
    return [e.flatten(0, 1) for e in image_features.split(num_patches.tolist())]
except RuntimeError as e:
    logger.error(f"Image processing failed: {e}")
    raise InvalidImageError("Failed to process image features") from e
Fixes
  • https://github.com/vllm-project/vllm/pull/29881
Versiones

Rangos afectados

Afectada Corregida Última
vllm≥ 0.6.4 < 0.12.0
0.6.4
0.12.0
0.34.0
Remediación

Estado del parche por paquete

Paquete / productoEstadoCorregida enDesde divulgaciónÚltima versión
pypivllm
Corregida 0.12.0 -41 d 0.34.0 no afectada
Versiones corregidas0.12.0
Explotación

Arsenal público

Sin exploits, módulos ni PoC públicos conocidos a la fecha de verificación.

Referencias
Advisoryhttps://nvd.nist.gov/vuln/detail/CVE-2026-22773
Otrahttps://github.com/vllm-project/vllm/security/advisories/GHSA-grg2-63fw-f2qr
Otrahttps://github.com/vllm-project/vllm/pull/29881
Otrahttps://github.com/vllm-project/vllm/commit/0ec84221718d920c3f46da879cc354f94b8fb59e
Otrahttps://github.com/pypa/advisory-database/tree/main/vulns/vllm/PYSEC-2026-143.yaml
Otrahttps://github.com/vllm-project/vllm

Las referencias externas se muestran para copiar: cve.llama.ec no lo redirige a otros sitios.