CVE-2026-93840

Publié : 18 septembre 2026

Modifié : 18 septembre 2026

Score CVSS

3.7

LOW

Description détaillée

vLLM before 0.29.0 validates allowed_token_ids against tokenizer length instead of model output logits width in SamplingParams._validate_allowed_token_ids(). Attackers can supply token IDs above the output vocabulary that pass validation, causing LogitBiasState to corrupt GPU logits state and allow concurrent requests to sample tokens outside their allowlists.

Vecteur d'attaque (CVSS)

Vecteur brut :CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:L/A:N

Références et Patchs

https://github.com/vllm-project/vllm https://github.com/vllm-project/vllm/blob/v0.28.0/vllm/sampling_params.py#L881-L903 https://github.com/vllm-project/vllm/blob/v0.28.0/vllm/v1/worker/gpu/sample/logit_bias.py#L179-L191 https://github.com/vllm-project/vllm/commit/5b0e5b69ac1a3884a6479c9537789c95263cc804 https://github.com/vllm-project/vllm/pull/49080 https://www.vulncheck.com/advisories/vllm-before-0.29.0-cross-request-logits-corruption-via-allowed-token-ids

Dernières Vulnérabilités

CVE-2026-93740

A vulnerability was identified in Totolink A3002MU Hh-B20211125.1046. Affected is the function formWlEncrypt of the file /boafrm/formWlEncrypt. The manipulation of the argument submit-url leads to buffer overflow. It is possible to initiate the attack remotely. The exploit is publicly available and might be used.

VOIR DÉTAILS

CVE-2026-93739

A vulnerability was determined in Totolink A3002MU Hh-B20211125.1046. This impacts the function formWlAc of the file /boafrm/formWlAc. Executing a manipulation of the argument submit-url can lead to buffer overflow. The attack may be performed from remote. The exploit has been publicly disclosed and may be utilized.

VOIR DÉTAILS

CVE-2026-75885

A flaw was found in the OpenShift console. Unauthenticated access to the `/api/devfile/` and `/api/devfile/samples/` endpoints allows a remote attacker to send crafted devfile payloads. This can lead to Server-Side Request Forgery (SSRF), where the console pod makes requests to internal services and reflects partial responses to the attacker. Additionally, by sending repeated large requests without a specified content length, an attacker can cause unbounded memory growth, leading to a Denial of Service (DoS).

VOIR DÉTAILS