GHSA-253C-MCHW-3W2R
Vulnerability from github – Published: 2026-09-29 17:57 – Updated: 2026-09-29 17:57Summary
Two independent quadratic paths in the linkify: true handling. Both are in markdown-it's own code rather than in linkify-it, which stays linear on both payloads.
src/rules_core/linkify.ts calls arrayReplaceAt once per linkified text token, and that rebuilds the whole children array each time. A paragraph of N soft-broken lines is one inline token with about 2N children, so you get N rebuilds over a 2N array. Schema-less emails are what reach it. A http:// link gets consumed by the inline rule first and never arrives as a text token, so those stay linear.
src/rules_inline/linkify.ts runs state.pending.match(SCHEME_RE) at every :// in the source. state.pending only gets truncated once a link is actually produced, so an unregistered scheme leaves it growing and every :// rescans the lot.
Proof of concept
Clean install of 15.0.0 from npm, new MarkdownIt({ linkify: true }).render(payload), Node 24.
| N | 5,000 | 10,000 | 20,000 | 40,000 |
|---|---|---|---|---|
'a@b.co\n'.repeat(N), 34KB to 273KB |
1.1s | 4.7s | 19.6s | 89s |
'a://'.repeat(N), 20KB to 156KB |
0.25s | 0.76s | 3.1s | 15.0s |
Doubling the input roughly quadruples the time in both. With linkify: false the same inputs run in 25 to 131ms and stay flat.
Controls for the first one, all at N=20000: putting every email in a single text token ('a@b.co ') takes 0.59s, one email per paragraph ('a@b.co\n\n') takes 0.64s, and soft-broken lines with nothing linkifiable take 42ms. So it needs many children AND many of them linkifying. For the second, replaying just the SCHEME_RE calls against the same growing prefixes with no markdown-it involved accounts for 11.7s of the 15s.
Ordinary prose does it too. 'ping a@b.co ok\n'.repeat(20000) is 293KB and takes 31s.
Caveat
linkify is off by default, so this only reaches apps that turn it on.
Impact
Availability only. A few hundred KB of fairly ordinary markdown pins one core for tens of seconds, and because it's quadratic it degrades quickly with size. Nothing is read, written or executed.
{
"affected": [
{
"package": {
"ecosystem": "npm",
"name": "markdown-it"
},
"ranges": [
{
"events": [
{
"introduced": "15.0.0"
},
{
"fixed": "15.0.1"
}
],
"type": "ECOSYSTEM"
}
],
"versions": [
"15.0.0"
]
},
{
"package": {
"ecosystem": "npm",
"name": "markdown-it"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "14.3.1"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [],
"database_specific": {
"cwe_ids": [
"CWE-400",
"CWE-407"
],
"github_reviewed": true,
"github_reviewed_at": "2026-09-29T17:57:56Z",
"nvd_published_at": null,
"severity": "MODERATE"
},
"details": "## Summary\n\nTwo independent quadratic paths in the `linkify: true` handling. Both are in markdown-it\u0027s own code rather than in linkify-it, which stays linear on both payloads.\n\n`src/rules_core/linkify.ts` calls `arrayReplaceAt` once per linkified text token, and that rebuilds the whole `children` array each time. A paragraph of N soft-broken lines is one inline token with about 2N children, so you get N rebuilds over a 2N array. Schema-less emails are what reach it. A `http://` link gets consumed by the inline rule first and never arrives as a text token, so those stay linear.\n\n`src/rules_inline/linkify.ts` runs `state.pending.match(SCHEME_RE)` at every `://` in the source. `state.pending` only gets truncated once a link is actually produced, so an unregistered scheme leaves it growing and every `://` rescans the lot.\n\n## Proof of concept\n\nClean install of 15.0.0 from npm, `new MarkdownIt({ linkify: true }).render(payload)`, Node 24.\n\n| N | 5,000 | 10,000 | 20,000 | 40,000 |\n|---|---|---|---|---|\n| `\u0027a@b.co\\n\u0027.repeat(N)`, 34KB to 273KB | 1.1s | 4.7s | 19.6s | 89s |\n| `\u0027a://\u0027.repeat(N)`, 20KB to 156KB | 0.25s | 0.76s | 3.1s | 15.0s |\n\nDoubling the input roughly quadruples the time in both. With `linkify: false` the same inputs run in 25 to 131ms and stay flat.\n\nControls for the first one, all at N=20000: putting every email in a single text token (`\u0027a@b.co \u0027`) takes 0.59s, one email per paragraph (`\u0027a@b.co\\n\\n\u0027`) takes 0.64s, and soft-broken lines with nothing linkifiable take 42ms. So it needs many children AND many of them linkifying. For the second, replaying just the `SCHEME_RE` calls against the same growing prefixes with no markdown-it involved accounts for 11.7s of the 15s.\n\nOrdinary prose does it too. `\u0027ping a@b.co ok\\n\u0027.repeat(20000)` is 293KB and takes 31s.\n\n## Caveat\n\n`linkify` is off by default, so this only reaches apps that turn it on.\n\n## Impact\n\nAvailability only. A few hundred KB of fairly ordinary markdown pins one core for tens of seconds, and because it\u0027s quadratic it degrades quickly with size. Nothing is read, written or executed.",
"id": "GHSA-253c-mchw-3w2r",
"modified": "2026-09-29T17:57:56Z",
"published": "2026-09-29T17:57:56Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/markdown-it/markdown-it/security/advisories/GHSA-253c-mchw-3w2r"
},
{
"type": "WEB",
"url": "https://github.com/markdown-it/markdown-it/commit/09fa07118dda4c953f058848f53dae88395618ca"
},
{
"type": "WEB",
"url": "https://github.com/markdown-it/markdown-it/commit/aaadcfa6d817b3c5f89afb49d97c7b797a6dd4fd"
},
{
"type": "WEB",
"url": "https://github.com/markdown-it/markdown-it/commit/ad70f6b7cff64bee10e42a774147112480ca0d49"
},
{
"type": "PACKAGE",
"url": "https://github.com/markdown-it/markdown-it"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:4.0/AV:N/AC:L/AT:P/PR:N/UI:N/VC:N/VI:N/VA:L/SC:N/SI:N/SA:N",
"type": "CVSS_V4"
}
],
"summary": "markdown-it linkify: true has two quadratic paths, so a few hundred KB of markdown blocks the event loop for tens of seconds"
}
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 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.