GHSA-2HJF-7455-W946

Vulnerability from github – Published: 2026-11-20 14:43 – Updated: 2026-11-20 14:43
VLAI
Summary
aMQTT: Inefficient Regular Expression Complexity and Uncontrolled Resource Consumption and Improper Input Validation
Details

Impact

amqtt is vulnerable to a remote denial of service caused by improper MQTT topic-filter validation combined with catastrophic regular-expression backtracking.

A single client can freeze the entire broker by sending one specially crafted SUBSCRIBE packet followed by one PUBLISH packet.

The vulnerability is caused by two defects:

  1. add_subscription() validates the placement of the single-level wildcard + using substring checks for /+ and +/, rather than verifying that each /-delimited level containing + is exactly +. This allows malformed filters such as "+/" + "+" * 60 to be accepted instead of rejected with return code 0x80, as required by MQTT-4.7.1-3.
  2. _matches() converts every + character into the greedy regular-expression fragment [^/]* and evaluates the resulting expression with re.fullmatch(). A filter containing many adjacent + characters can cause catastrophic backtracking when matched against a topic that partially matches but ultimately fails.

Topic matching runs synchronously inside the broker’s global broadcast loop. When the malicious match is evaluated, it blocks the asyncio event loop and prevents the broker from processing any other work.

The issue was reproduced using a single raw MQTT 3.1.1 TCP connection with no third-party client dependencies.

Successful exploitation causes:

  • Existing message delivery to stop
  • New MQTT connections to receive no CONNACK
  • Keepalive processing to stop
  • Broadcast processing to stop
  • The broker to remain unavailable until the process is restarted

Under the default configuration, anonymous clients can exploit the vulnerability without credentials. When authentication is enabled, any authenticated client permitted to create arbitrary subscriptions and publish messages may still exploit it. Authentication alone is therefore not a sufficient mitigation.

  • Ecosystem: pip
  • Package: amqtt
  • Affected versions: >= 0.11.0, <= 0.11.3

The vulnerable code is present in the v0.11.0 release commit, 352678c87e50e501a8a5bb91fab73c9400fa046c, and remained present on the main branch as of eeb1e254da9a1e8d9cc68ae754fcedac652e8fc0. Versions before 0.11.0 were not evaluated.

This is a denial-of-service vulnerability only. re.escape() prevents arbitrary regular-expression injection, and there is no known path to remote code execution.

No confidentiality or integrity impact has been identified. Successful exploitation causes a complete loss of availability for the affected broker process.

Patches

The issue has been patched in the v0.11.4 released version; all users should upgrade.

The completed fix:

  • Validate topic filters one level at a time
  • Require any level containing + to be exactly +
  • As part of comprehensive MQTT topic-filter validation, require # to occupy an entire level and appear only as the final level
  • Replace the regular-expression matcher with deterministic, level-by-level MQTT topic matching

Simply changing the validation logic is not sufficient defense in depth if malformed filters can reach the existing regular-expression matcher through another code path.

Workarounds

This is the temporary mitigation that was recommended prior to the patched release: administrators can install a topic-filtering plugin that rejects malformed wildcard subscriptions before they are added to the broker. Place the plugin in an importable Python module, enable it in the broker configuration, and restart the broker.

from amqtt.contexts import Action
from amqtt.plugins.base import BaseTopicPlugin
from amqtt.session import Session

class RejectInvalidTopicFilters(BaseTopicPlugin):
    """A plugin to prevent invalid wildcard subscriptions."""
    async def topic_filtering(
        self, *, session: Session | None = None,
        topic: str | None = None,
        action: Action | None = None
    ) -> bool | None:

        if topic and action == Action.SUBSCRIBE:
            levels = topic.split("/")
            return all(
                level in {"+", "#"} or ("+" not in level and "#" not in level)
                for level in levels
            ) and "#" not in levels[:-1]
        return True

config = {
    ...
    'plugins':{
        ...
        'myplugins.RejectInvalidTopicFilters': {}
    }
}

This workaround applies when subscription requests pass through the plugin. If the broker is configured to use a persistent session plugin, restarting the broker alone is not sufficient; any persisted sessions containing malformed topic filters need to be removed.

Restarting the broker restores availability after exploitation, but the broker remains vulnerable until the patch or workaround is installed.

Acknowledgments

aMQTT thanks the following security researchers for their professional, coordinated private disclosure:

  • Yongkang Xiao (Wuhan University)
  • Min Shi (Wuhan University)
  • Qiyi Deng (Wuhan University)
  • Jing Chen (Wuhan University)
Show details on source website

{
  "affected": [
    {
      "database_specific": {
        "last_known_affected_version_range": "\u003c= 0.11.3"
      },
      "package": {
        "ecosystem": "PyPI",
        "name": "amqtt"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0.11.0"
            },
            {
              "fixed": "0.11.4"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [
    "CVE-2026-104199"
  ],
  "database_specific": {
    "cwe_ids": [
      "CWE-1333",
      "CWE-20",
      "CWE-400"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2026-11-20T14:43:41Z",
    "nvd_published_at": "2026-11-19T20:17:14Z",
    "severity": "HIGH"
  },
  "details": "### Impact\n\n`amqtt` is vulnerable to a remote denial of service caused by improper MQTT topic-filter validation combined with catastrophic regular-expression backtracking.\n\nA single client can freeze the entire broker by sending one specially crafted `SUBSCRIBE` packet followed by one `PUBLISH` packet.\n\nThe vulnerability is caused by two defects:\n\n1. `add_subscription()` validates the placement of the single-level wildcard `+` using substring checks for `/+` and `+/`, rather than verifying that each `/`-delimited level containing `+` is exactly `+`. This allows malformed filters such as `\"+/\" + \"+\" * 60` to be accepted instead of rejected with return code `0x80`, as required by MQTT-4.7.1-3.\n2. `_matches()` converts every `+` character into the greedy regular-expression fragment `[^/]*` and evaluates the resulting expression with `re.fullmatch()`. A filter containing many adjacent `+` characters can cause catastrophic backtracking when matched against a topic that partially matches but ultimately fails.\n\nTopic matching runs synchronously inside the broker\u2019s global broadcast loop. When the malicious match is evaluated, it blocks the asyncio event loop and prevents the broker from processing any other work.\n\nThe issue was reproduced using a single raw MQTT 3.1.1 TCP connection with no third-party client dependencies.\n\nSuccessful exploitation causes:\n\n- Existing message delivery to stop\n- New MQTT connections to receive no `CONNACK`\n- Keepalive processing to stop\n- Broadcast processing to stop\n- The broker to remain unavailable until the process is restarted\n\nUnder the default configuration, anonymous clients can exploit the vulnerability without credentials. When authentication is enabled, any authenticated client permitted to create arbitrary subscriptions and publish messages may still exploit it. Authentication alone is therefore not a sufficient mitigation.\n\n- Ecosystem: `pip`\n- Package: `amqtt`\n- Affected versions: `\u003e= 0.11.0, \u003c= 0.11.3`\n\nThe vulnerable code is present in the `v0.11.0` release commit, `352678c87e50e501a8a5bb91fab73c9400fa046c`, and remained present on the main branch as of `eeb1e254da9a1e8d9cc68ae754fcedac652e8fc0`. Versions before 0.11.0 were not evaluated.\n\nThis is a denial-of-service vulnerability only. `re.escape()` prevents arbitrary regular-expression injection, and there is no known path to remote code execution.\n\nNo confidentiality or integrity impact has been identified. Successful exploitation causes a complete loss of availability for the affected broker process.\n\n### Patches\n\nThe issue has been patched in the `v0.11.4` released version; all users should upgrade.\n\nThe completed fix:\n\n- Validate topic filters one level at a time\n- Require any level containing `+` to be exactly `+`\n- As part of comprehensive MQTT topic-filter validation, require `#` to occupy an entire level and appear only as the final level\n- Replace the regular-expression matcher with deterministic, level-by-level MQTT topic matching\n\nSimply changing the validation logic is not sufficient defense in depth if malformed filters can reach the existing regular-expression matcher through another code path.\n\n### Workarounds\n\nThis is the temporary mitigation that was recommended prior to the patched release: administrators can install a topic-filtering plugin that rejects malformed wildcard subscriptions before they are added to the broker. Place the plugin in an importable Python module, enable it in the broker configuration, and restart the broker.\n\n```python\nfrom amqtt.contexts import Action\nfrom amqtt.plugins.base import BaseTopicPlugin\nfrom amqtt.session import Session\n\nclass RejectInvalidTopicFilters(BaseTopicPlugin):\n    \"\"\"A plugin to prevent invalid wildcard subscriptions.\"\"\"\n    async def topic_filtering(\n        self, *, session: Session | None = None,\n        topic: str | None = None,\n        action: Action | None = None\n    ) -\u003e bool | None:\n\n        if topic and action == Action.SUBSCRIBE:\n            levels = topic.split(\"/\")\n            return all(\n                level in {\"+\", \"#\"} or (\"+\" not in level and \"#\" not in level)\n                for level in levels\n            ) and \"#\" not in levels[:-1]\n        return True\n        \nconfig = {\n    ...\n    \u0027plugins\u0027:{\n        ...\n        \u0027myplugins.RejectInvalidTopicFilters\u0027: {}\n    }\n}\n``` \n\nThis workaround applies when subscription requests pass through the plugin. If the broker is configured to use a persistent session plugin, restarting the broker alone is not sufficient; any persisted sessions containing malformed topic filters need to be removed.\n\nRestarting the broker restores availability after exploitation, but the broker remains vulnerable until the patch or workaround is installed.\n\n### Acknowledgments\n\naMQTT thanks the following security researchers for their professional, coordinated private disclosure:\n\n* **Yongkang Xiao** (Wuhan University)\n* **Min Shi** (Wuhan University)\n* **Qiyi Deng** (Wuhan University)\n* **Jing Chen** (Wuhan University)",
  "id": "GHSA-2hjf-7455-w946",
  "modified": "2026-11-20T14:43:41Z",
  "published": "2026-11-20T14:43:41Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/Yakifo/amqtt/security/advisories/GHSA-2hjf-7455-w946"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-104199"
    },
    {
      "type": "WEB",
      "url": "https://github.com/Yakifo/amqtt/pull/361"
    },
    {
      "type": "WEB",
      "url": "https://github.com/Yakifo/amqtt/commit/e48a8a65027667f936d415db7f01d3ac350104e0"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/Yakifo/amqtt"
    },
    {
      "type": "WEB",
      "url": "https://github.com/Yakifo/amqtt/releases/tag/v0.11.4"
    }
  ],
  "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/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X",
      "type": "CVSS_V4"
    }
  ],
  "summary": "aMQTT: Inefficient Regular Expression Complexity and Uncontrolled Resource Consumption and Improper Input Validation"
}



Log in or create an account to share your comment.




Tags
Taxonomy of the tags.


Loading…

Loading…

Loading…

Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.

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.
Vulnerability timeline

Sightings and recorded dates, grouped by day · UTC

JSON
How durations are calculated

Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.

Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.

Events included in the timeline

Loading timeline…


    Loading…

    Loading…

    Loading…

    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.