vllm-project vLLM up to 0.28.x Disaggregated Serving serving.py InputProcessor._validate_prompt_len token_ids resource consumption

Summaryinfo

A vulnerability was found in vllm-project vLLM up to 0.28.x. It has been declared as problematic. This affects the function InputProcessor._validate_prompt_len of the file vllm/entrypoints/serve/disagg/serving.py of the component Disaggregated Serving. The manipulation of the argument token_ids results in resource consumption. This vulnerability is reported as CVE-2026-100651. The attack can be launched remotely. No exploit exists. It is recommended to upgrade the affected component.

Detailsinfo

A vulnerability, which was classified as problematic, was found in vllm-project vLLM up to 0.28.x. Affected is the function InputProcessor._validate_prompt_len of the file vllm/entrypoints/serve/disagg/serving.py of the component Disaggregated Serving. The manipulation of the argument token_ids with an unknown input leads to a resource consumption vulnerability. CWE is classifying the issue as CWE-400. The product does not properly control the allocation and maintenance of a limited resource, thereby enabling an actor to influence the amount of resources consumed, eventually leading to the exhaustion of available resources. This is going to have an impact on availability. CVE summarizes:

vLLM before 0.29.0 fails to enforce decoder prompt-length validation on the disaggregated serving endpoint /inference/v1/generate. When the request contains a 'features' (multimodal) payload, vllm/entrypoints/serve/disagg/serving.py builds a multimodal EngineInput directly from the caller-supplied token_ids, and GenerateRequest.token_ids (vllm/entrypoints/serve/disagg/protocol.py) is not checked against model_config.max_model_len. For multimodal processors that report skip_prompt_length_check=True (for example Nemotron Parse, Whisper, and FireRedLID), InputProcessor._validate_prompt_len() returns immediately for both encoder and decoder prompts, so an overlong prompt becomes an EngineCoreRequest and reaches the worker input-batch copy into a fixed max_model_len-wide NumPy row. A client able to reach the endpoint on an affected model configuration can therefore submit an overlong token_ids list to trigger a worker failure and denial of service. Fixed in 0.29.0.

The advisory is available at github.com. This vulnerability is traded as CVE-2026-100651 since 09/26/2026. The exploitability is told to be easy. It is possible to launch the attack remotely. Technical details are known, but there is no available exploit. This vulnerability is assigned to T1499 by the MITRE ATT&CK project.

Upgrading to version 0.29.0 eliminates this vulnerability.

If you want to get best quality of vulnerability data, you may have to visit VulDB.

Productinfo

Type

• Project Management Software

Vendor

• vllm-project

Name

• vLLM

Version

• 0.1
• 0.2
• 0.3
• 0.4
• 0.5
• 0.6
• 0.7
• 0.8
• 0.9
• 0.10
• 0.11
• 0.12
• 0.13
• 0.14
• 0.15
• 0.16
• 0.17
• 0.18
• 0.19
• 0.20
• 0.21
• 0.22
• 0.23
• 0.24
• 0.25
• 0.26
• 0.27
• 0.28

Website

• Product: https://github.com/vllm-project/vllm/

CPE 2.3info

• 🔒
• 🔒
• 🔒

CPE 2.2info

• 🔒
• 🔒
• 🔒

CVSSv4info

CVSSv3info

CVSSv2info

Exploitinginfo

Threat Intelligenceinfo

Countermeasuresinfo

Timelineinfo

Sourcesinfo

Entryinfo

If you want to get best quality of vulnerability data, you may have to visit VulDB.

Discussion