Vulnerability-Lookup

GHSA-MW4Q-PWHM-35F7

Vulnerability from github – Published: 2026-10-12 18:31 – Updated: 2026-10-12 18:31

Details

ExLlamaV3’s EXL3 loader uses an untrusted, model-supplied index to select a CUDA kernel without first verifying that the index is within bounds. A malicious model can force the code to read and invoke an out‑of‑range function pointer, resulting in a reproducible crash. Under specific conditions involving certain host memory layouts, it may be possible for an attacker to exploit this vulnerability to achieve controlled indirect kernel dispatch during model load.

Show details on source website

JSON

To clipboard

{
  "affected": [],
  "aliases": [
    "CVE-2026-84286"
  ],
  "database_specific": {
    "cwe_ids": [],
    "github_reviewed": false,
    "github_reviewed_at": null,
    "nvd_published_at": "2026-10-12T17:17:35Z",
    "severity": null
  },
  "details": "ExLlamaV3\u2019s EXL3 loader uses an untrusted, model-supplied index to select a CUDA kernel without first verifying that the index is within bounds. A malicious model can force the code to read and invoke an out\u2011of\u2011range function pointer, resulting in a reproducible crash. Under specific conditions involving certain host memory layouts, it may be possible for an attacker to exploit this vulnerability to achieve controlled indirect kernel dispatch during model load.",
  "id": "GHSA-mw4q-pwhm-35f7",
  "modified": "2026-10-12T18:31:55Z",
  "published": "2026-10-12T18:31:55Z",
  "references": [
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-84286"
    },
    {
      "type": "WEB",
      "url": "https://github.com/turboderp-org/exllamav3/pull/310"
    },
    {
      "type": "WEB",
      "url": "https://kb.cert.org/vuls/id/369611"
    },
    {
      "type": "WEB",
      "url": "https://www.kb.cert.org/vuls/id/369611"
    }
  ],
  "schema_version": "1.4.0",
  "severity": []
}

CVE-2026-84286 (GCVE-0-2026-84286)

Vulnerability from cvelistv5 – Published: 2026-10-12 17:01 – Updated: 2026-10-12 17:14

Title

OOB Read/Write via Negative Array Index in exllamav3_ext Kernel Table

Summary

Severity

No CVSS data available.

CWE

Assigner

certcc CNA under the mitre root

CNA scorecard E 42/100 over 156 records in the last 180 days details

References

3 references

URL	Tags
https://kb.cert.org/vuls/id/369611
https://github.com/turboderp-org/exllamav3/pull/310
https://www.kb.cert.org/vuls/id/369611

Impacted products

1 product

Vendor	Product	Version
Turboderp-org	ExLlamaV3	Affected: 3

Show details on NVD website

JSON

To clipboard

{
  "containers": {
    "adp": [
      {
        "providerMetadata": {
          "dateUpdated": "2026-10-12T17:14:02.436Z",
          "orgId": "af854a3a-2127-422b-91ae-364da2661108",
          "shortName": "CVE"
        },
        "references": [
          {
            "url": "https://www.kb.cert.org/vuls/id/369611"
          }
        ],
        "title": "CVE Program Container"
      }
    ],
    "cna": {
      "affected": [
        {
          "product": "ExLlamaV3",
          "vendor": "Turboderp-org",
          "versions": [
            {
              "status": "affected",
              "version": "3"
            }
          ]
        }
      ],
      "descriptions": [
        {
          "lang": "en",
          "value": "ExLlamaV3\u2019s EXL3 loader uses an untrusted, model-supplied index to select a CUDA kernel without first verifying that the index is within bounds. A malicious model can force the code to read and invoke an out\u2011of\u2011range function pointer, resulting in a reproducible crash. Under specific conditions involving certain host memory layouts, it may be possible for an attacker to exploit this vulnerability to achieve controlled indirect kernel dispatch during model load."
        }
      ],
      "problemTypes": [
        {
          "descriptions": [
            {
              "description": "CWE-125 Out-of-bounds Read",
              "lang": "en"
            }
          ]
        },
        {
          "descriptions": [
            {
              "description": "CWE-787 Out-of-bounds Write",
              "lang": "en"
            }
          ]
        },
        {
          "descriptions": [
            {
              "description": "CWE-20 Improper Input Validation",
              "lang": "en"
            }
          ]
        }
      ],
      "providerMetadata": {
        "dateUpdated": "2026-10-12T17:06:12.603Z",
        "orgId": "37e5125f-f79b-445b-8fad-9564f167944b",
        "shortName": "certcc"
      },
      "references": [
        {
          "url": "https://kb.cert.org/vuls/id/369611"
        },
        {
          "url": "https://github.com/turboderp-org/exllamav3/pull/310"
        }
      ],
      "source": {
        "discovery": "UNKNOWN"
      },
      "title": "OOB Read/Write via Negative Array Index in exllamav3_ext Kernel Table",
      "x_generator": {
        "engine": "VINCE 3.0.50",
        "env": "prod",
        "origin": "https://cveawg.mitre.org/api/cve/CVE-2026-84286"
      }
    }
  },
  "cveMetadata": {
    "assignerOrgId": "37e5125f-f79b-445b-8fad-9564f167944b",
    "assignerShortName": "certcc",
    "cveId": "CVE-2026-84286",
    "datePublished": "2026-10-12T17:01:16.032Z",
    "dateReserved": "2026-09-01T15:47:57.172Z",
    "dateUpdated": "2026-10-12T17:14:02.436Z",
    "state": "PUBLISHED"
  },
  "dataType": "CVE_RECORD",
  "dataVersion": "5.2"
}

Sightings

Author	Source	Type	Date	Other

Nomenclature

Seen: The vulnerability was mentioned, discussed, or observed by the user.
Confirmed: The vulnerability has been validated from an analyst's perspective.
Published Proof of Concept: A public proof of concept is available for this vulnerability.
Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
Not confirmed: The user expressed doubt about the validity of the vulnerability.
Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.

The MITRE ATT&CK techniques below are AI-generated suggestions, inferred from the description of the vulnerability by the CIRCL/vulnerability-attack-technique-classification-roberta-base model, served locally by ML-Gateway. They have not been verified by an analyst and are provided for guidance only.
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.

Related by attack behaviour

Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.

Action not permitted

GHSA-MW4Q-PWHM-35F7

CVE-2026-84286 (GCVE-0-2026-84286)

Tags

Sightings

Nomenclature

Related by attack behaviour