{"dataType": "CVE_RECORD", "dataVersion": "5.1", "cveMetadata": {"cveId": "CVE-2025-46560", "assignerOrgId": "a0819718-46f1-4df5-94e2-005712e83aaa", "state": "PUBLISHED", "assignerShortName": "GitHub_M", "dateReserved": "2025-04-24T21:10:48.174Z", "datePublished": "2025-04-30T00:24:53.750Z", "dateUpdated": "2025-04-30T13:09:13.422Z"}, "containers": {"cna": {"title": "vLLM phi4mm: Quadratic Time Complexity in Input Token Processing\u200b leads to denial of service", "problemTypes": [{"descriptions": [{"cweId": "CWE-1333", "lang": "en", "description": "CWE-1333: Inefficient Regular Expression Complexity", "type": "CWE"}]}], "metrics": [{"cvssV3_1": {"attackComplexity": "LOW", "attackVector": "NETWORK", "availabilityImpact": "HIGH", "baseScore": 6.5, "baseSeverity": "MEDIUM", "confidentialityImpact": "NONE", "integrityImpact": "NONE", "privilegesRequired": "LOW", "scope": "UNCHANGED", "userInteraction": "NONE", "vectorString": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H", "version": "3.1"}}], "references": [{"name": "https://github.com/vllm-project/vllm/security/advisories/GHSA-vc6m-hm49-g9qg", "tags": ["x_refsource_CONFIRM"], "url": "https://github.com/vllm-project/vllm/security/advisories/GHSA-vc6m-hm49-g9qg"}, {"name": "https://github.com/vllm-project/vllm/blob/8cac35ba435906fb7eb07e44fe1a8c26e8744f4e/vllm/model_executor/models/phi4mm.py#L1182-L1197", "tags": ["x_refsource_MISC"], "url": "https://github.com/vllm-project/vllm/blob/8cac35ba435906fb7eb07e44fe1a8c26e8744f4e/vllm/model_executor/models/phi4mm.py#L1182-L1197"}], "affected": [{"vendor": "vllm-project", "product": "vllm", "versions": [{"version": ">= 0.8.0, < 0.8.5", "status": "affected"}]}], "providerMetadata": {"orgId": "a0819718-46f1-4df5-94e2-005712e83aaa", "shortName": "GitHub_M", "dateUpdated": "2025-04-30T00:24:53.750Z"}, "descriptions": [{"lang": "en", "value": "vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Versions starting from 0.8.0 and prior to 0.8.5 are affected by a critical performance vulnerability 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 \u200b\u200binefficient list concatenation operations\u200b\u200b, the algorithm exhibits \u200b\u200bquadratic time complexity (O(n\u00b2))\u200b\u200b, allowing malicious actors to trigger resource exhaustion via specially crafted inputs. This issue has been patched in version 0.8.5."}], "source": {"advisory": "GHSA-vc6m-hm49-g9qg", "discovery": "UNKNOWN"}}, "adp": [{"references": [{"url": "https://github.com/vllm-project/vllm/security/advisories/GHSA-vc6m-hm49-g9qg", "tags": ["exploit"]}], "metrics": [{"other": {"type": "ssvc", "content": {"timestamp": "2025-04-30T13:09:10.349287Z", "id": "CVE-2025-46560", "options": [{"Exploitation": "poc"}, {"Automatable": "no"}, {"Technical Impact": "partial"}], "role": "CISA Coordinator", "version": "2.0.3"}}}], "title": "CISA ADP Vulnrichment", "providerMetadata": {"orgId": "134c704f-9b21-4f2e-91b3-4a467353bcc0", "shortName": "CISA-ADP", "dateUpdated": "2025-04-30T13:09:13.422Z"}}]}}

{"dataType": "CVE_RECORD", "containers": {"adp": [{"title": "CISA ADP Vulnrichment", "metrics": [{"other": {"type": "ssvc", "content": {"id": "CVE-2025-46560", "role": "CISA Coordinator", "options": [{"Exploitation": "poc"}, {"Automatable": "no"}, {"Technical Impact": "partial"}], "version": "2.0.3", "timestamp": "2025-04-30T13:09:10.349287Z"}}}], "references": [{"url": "https://github.com/vllm-project/vllm/security/advisories/GHSA-vc6m-hm49-g9qg", "tags": ["exploit"]}], "providerMetadata": {"orgId": "134c704f-9b21-4f2e-91b3-4a467353bcc0", "shortName": "CISA-ADP", "dateUpdated": "2025-04-30T13:09:04.715Z"}}], "cna": {"title": "vLLM phi4mm: Quadratic Time Complexity in Input Token Processing\u200b leads to denial of service", "source": {"advisory": "GHSA-vc6m-hm49-g9qg", "discovery": "UNKNOWN"}, "metrics": [{"cvssV3_1": {"scope": "UNCHANGED", "version": "3.1", "baseScore": 6.5, "attackVector": "NETWORK", "baseSeverity": "MEDIUM", "vectorString": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H", "integrityImpact": "NONE", "userInteraction": "NONE", "attackComplexity": "LOW", "availabilityImpact": "HIGH", "privilegesRequired": "LOW", "confidentialityImpact": "NONE"}}], "affected": [{"vendor": "vllm-project", "product": "vllm", "versions": [{"status": "affected", "version": ">= 0.8.0, < 0.8.5"}]}], "references": [{"url": "https://github.com/vllm-project/vllm/security/advisories/GHSA-vc6m-hm49-g9qg", "name": "https://github.com/vllm-project/vllm/security/advisories/GHSA-vc6m-hm49-g9qg", "tags": ["x_refsource_CONFIRM"]}, {"url": "https://github.com/vllm-project/vllm/blob/8cac35ba435906fb7eb07e44fe1a8c26e8744f4e/vllm/model_executor/models/phi4mm.py#L1182-L1197", "name": "https://github.com/vllm-project/vllm/blob/8cac35ba435906fb7eb07e44fe1a8c26e8744f4e/vllm/model_executor/models/phi4mm.py#L1182-L1197", "tags": ["x_refsource_MISC"]}], "descriptions": [{"lang": "en", "value": "vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Versions starting from 0.8.0 and prior to 0.8.5 are affected by a critical performance vulnerability 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 \u200b\u200binefficient list concatenation operations\u200b\u200b, the algorithm exhibits \u200b\u200bquadratic time complexity (O(n\u00b2))\u200b\u200b, allowing malicious actors to trigger resource exhaustion via specially crafted inputs. This issue has been patched in version 0.8.5."}], "problemTypes": [{"descriptions": [{"lang": "en", "type": "CWE", "cweId": "CWE-1333", "description": "CWE-1333: Inefficient Regular Expression Complexity"}]}], "providerMetadata": {"orgId": "a0819718-46f1-4df5-94e2-005712e83aaa", "shortName": "GitHub_M", "dateUpdated": "2025-04-30T00:24:53.750Z"}}}, "cveMetadata": {"cveId": "CVE-2025-46560", "state": "PUBLISHED", "dateUpdated": "2025-04-30T13:09:13.422Z", "dateReserved": "2025-04-24T21:10:48.174Z", "assignerOrgId": "a0819718-46f1-4df5-94e2-005712e83aaa", "datePublished": "2025-04-30T00:24:53.750Z", "assignerShortName": "GitHub_M"}, "dataVersion": "5.1"}

{"cveTags": [], "descriptions": [{"lang": "en", "value": "vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Versions starting from 0.8.0 and prior to 0.8.5 are affected by a critical performance vulnerability 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 \u200b\u200binefficient list concatenation operations\u200b\u200b, the algorithm exhibits \u200b\u200bquadratic time complexity (O(n\u00b2))\u200b\u200b, allowing malicious actors to trigger resource exhaustion via specially crafted inputs. This issue has been patched in version 0.8.5."}, {"lang": "es", "value": "vLLM es un motor de inferencia y servicio de alto rendimiento y eficiente en memoria para LLM. Las versiones a partir de la 0.8.0 y anteriores a la 0.8.5 se ven afectadas por una vulnerabilidad cr\u00edtica de rendimiento en la l\u00f3gica de preprocesamiento de entrada del tokenizador multimodal. El c\u00f3digo reemplaza din\u00e1micamente los tokens de marcador de posici\u00f3n (p. ej., &lt;|audio_|&gt;, &lt;|image_|&gt;) con tokens repetidos basados ??en longitudes precalculadas. Debido a las ineficientes operaciones de concatenaci\u00f3n de listas, el algoritmo presenta una complejidad temporal cuadr\u00e1tica (O(n\u00b2)), lo que permite a los actores maliciosos activar el agotamiento de recursos mediante entradas especialmente manipuladas. Este problema se ha corregido en la versi\u00f3n 0.8.5."}], "id": "CVE-2025-46560", "lastModified": "2025-05-02T13:53:40.163", "metrics": {"cvssMetricV31": [{"cvssData": {"attackComplexity": "LOW", "attackVector": "NETWORK", "availabilityImpact": "HIGH", "baseScore": 6.5, "baseSeverity": "MEDIUM", "confidentialityImpact": "NONE", "integrityImpact": "NONE", "privilegesRequired": "LOW", "scope": "UNCHANGED", "userInteraction": "NONE", "vectorString": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H", "version": "3.1"}, "exploitabilityScore": 2.8, "impactScore": 3.6, "source": "[email protected]", "type": "Secondary"}]}, "published": "2025-04-30T01:15:52.097", "references": [{"source": "[email protected]", "url": "https://github.com/vllm-project/vllm/blob/8cac35ba435906fb7eb07e44fe1a8c26e8744f4e/vllm/model_executor/models/phi4mm.py#L1182-L1197"}, {"source": "[email protected]", "url": "https://github.com/vllm-project/vllm/security/advisories/GHSA-vc6m-hm49-g9qg"}, {"source": "134c704f-9b21-4f2e-91b3-4a467353bcc0", "url": "https://github.com/vllm-project/vllm/security/advisories/GHSA-vc6m-hm49-g9qg"}], "sourceIdentifier": "[email protected]", "vulnStatus": "Awaiting Analysis", "weaknesses": [{"description": [{"lang": "en", "value": "CWE-1333"}], "source": "[email protected]", "type": "Secondary"}]}

{"bugzilla": {"description": "vllm: vLLM phi4mm: Quadratic Time Complexity in Input Token Processing\u200b leads to denial of service", "id": "2363025", "url": "https://bugzilla.redhat.com/show_bug.cgi?id=2363025"}, "csaw": false, "cvss3": {"cvss3_base_score": "6.5", "cvss3_scoring_vector": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H", "status": "draft"}, "cwe": "CWE-1333", "details": ["vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Versions starting from 0.8.0 and prior to 0.8.5 are affected by a critical performance vulnerability 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 \u200b\u200binefficient list concatenation operations\u200b\u200b, the algorithm exhibits \u200b\u200bquadratic time complexity (O(n\u00b2))\u200b\u200b, allowing malicious actors to trigger resource exhaustion via specially crafted inputs. This issue has been patched in version 0.8.5."], "mitigation": {"lang": "en:us", "value": "To mitigate this vulnerability:\n- Precompute all placeholder positions and expansion lengths upfront.\n- Replace dynamic list concatenation with a single preallocated array."}, "name": "CVE-2025-46560", "package_state": [{"cpe": "cpe:/a:redhat:enterprise_linux_ai:1", "fix_state": "Not affected", "package_name": "rhelai1/bootc-amd-rhel9", "product_name": "Red Hat Enterprise Linux AI (RHEL AI)"}, {"cpe": "cpe:/a:redhat:enterprise_linux_ai:1", "fix_state": "Not affected", "package_name": "rhelai1/bootc-aws-nvidia-rhel9", "product_name": "Red Hat Enterprise Linux AI (RHEL AI)"}, {"cpe": "cpe:/a:redhat:enterprise_linux_ai:1", "fix_state": "Not affected", "package_name": "rhelai1/bootc-azure-amd-rhel9", "product_name": "Red Hat Enterprise Linux AI (RHEL AI)"}, {"cpe": "cpe:/a:redhat:enterprise_linux_ai:1", "fix_state": "Not affected", "package_name": "rhelai1/bootc-azure-nvidia-rhel9", "product_name": "Red Hat Enterprise Linux AI (RHEL AI)"}, {"cpe": "cpe:/a:redhat:enterprise_linux_ai:1", "fix_state": "Not affected", "package_name": "rhelai1/bootc-gcp-nvidia-rhel9", "product_name": "Red Hat Enterprise Linux AI (RHEL AI)"}, {"cpe": "cpe:/a:redhat:enterprise_linux_ai:1", "fix_state": "Not affected", "package_name": "rhelai1/bootc-ibm-nvidia-rhel9", "product_name": "Red Hat Enterprise Linux AI (RHEL AI)"}, {"cpe": "cpe:/a:redhat:enterprise_linux_ai:1", "fix_state": "Not affected", "package_name": "rhelai1/bootc-intel-rhel9", "product_name": "Red Hat Enterprise Linux AI (RHEL AI)"}, {"cpe": "cpe:/a:redhat:enterprise_linux_ai:1", "fix_state": "Not affected", "package_name": "rhelai1/bootc-nvidia-rhel9", "product_name": "Red Hat Enterprise Linux AI (RHEL AI)"}, {"cpe": "cpe:/a:redhat:enterprise_linux_ai:1", "fix_state": "Not affected", "package_name": "rhelai1/instructlab-amd-rhel9", "product_name": "Red Hat Enterprise Linux AI (RHEL AI)"}, {"cpe": "cpe:/a:redhat:enterprise_linux_ai:1", "fix_state": "Not affected", "package_name": "rhelai1/instructlab-intel-rhel9", "product_name": "Red Hat Enterprise Linux AI (RHEL AI)"}, {"cpe": "cpe:/a:redhat:enterprise_linux_ai:1", "fix_state": "Not affected", "package_name": "rhelai1/instructlab-nvidia-rhel9", "product_name": "Red Hat Enterprise Linux AI (RHEL AI)"}, {"cpe": "cpe:/a:redhat:enterprise_linux_ai:1", "fix_state": "Not affected", "package_name": "rhelai1/ui-rhel9", "product_name": "Red Hat Enterprise Linux AI (RHEL AI)"}], "public_date": "2025-04-30T00:24:53Z", "references": ["https://www.cve.org/CVERecord?id=CVE-2025-46560\nhttps://nvd.nist.gov/vuln/detail/CVE-2025-46560\nhttps://github.com/vllm-project/vllm/blob/8cac35ba435906fb7eb07e44fe1a8c26e8744f4e/vllm/model_executor/models/phi4mm.py#L1182-L1197\nhttps://github.com/vllm-project/vllm/security/advisories/GHSA-vc6m-hm49-g9qg"], "statement": "Red Hat products are not using any affected version of the vllm component. Hence, Red Hat products are not affected by this vulnerability.", "threat_severity": "Moderate"}