GHSA-F67J-2JQW-JPQ7
Vulnerability from github – Published: 2026-09-28 21:31 – Updated: 2026-09-28 21:31A Denial of Service (DoS) vulnerability exists in @angular/platform-server's DOM emulation parser (domino). When processing untrusted user input containing an incomplete DOCTYPE declaration ending with whitespace before EOF (such as <!DOCTYPE html), the HTML parser enters an infinite synchronous loop, pegging CPU utilization at 100% and completely freezing the Node.js server process.
Technical Description
In Angular Server-Side Rendering (SSR), @angular/platform-server uses domino to parse and sanitize HTML bound through template bindings (such as [innerHTML]) or manipulated via DOM APIs.
In Domino's HTML parser (lib/HTMLParser.js), tokenizer states that specify fixed lookahead—such as after_doctype_name_state (lookahead = 6)—rely on the state handler function to explicitly advance the character index pointer (nextchar). While branches for whitespace, >, and keyword matching advance nextchar, the EOF branch (case -1: // EOF) emitted doctype and EOF tokens without advancing nextchar or transitioning out of the state:
case -1: // EOF
forcequirks();
emitDoctype();
emitEOF();
break;
Because nextchar remained unchanged pointing to the EOF marker character (\uFFFF), the scanner loop (while (nextchar < numchars)) repeatedly re-invoked after_doctype_name_state with codepoint = EOF indefinitely. In Node.js's single-threaded runtime, this synchronous loop starves the event loop entirely.
Impact & Reachability
- Reachability: The vulnerability is reachable in any Angular SSR application where untrusted user input is bound to
[innerHTML], interpolated into markup, or sanitized on the server. - Impact: Successful exploitation allows an unauthenticated remote attacker to cause an immediate Denial of Service (DoS) by sending a payload containing an incomplete DOCTYPE (e.g.,
<!DOCTYPE html). The Node.js SSR process locks up at 100% CPU and ceases responding to all concurrent and subsequent HTTP requests.
Proof of Concept:
import { Component } from '@angular/core';
@Component({
selector: 'app-root',
standalone: true,
template: `<div [innerHTML]="payload"></div>`,
})
export class AppComponent {
// Attacker-controlled input containing an incomplete DOCTYPE ending with whitespace
payload = '<!DOCTYPE html ';
}
Workarounds
- Avoid binding untrusted user input directly to
[innerHTML]in server-rendered templates; use standard text interpolation ({{ userInput }}) or[textContent]when raw HTML rendering is not required. - Validate or sanitize user input before passing it to
[innerHTML]on the server by stripping or rejecting strings matching/^<!DOCTYPE/i.
{
"affected": [
{
"package": {
"ecosystem": "npm",
"name": "@angular/platform-server"
},
"ranges": [
{
"events": [
{
"introduced": "22.0.0"
},
{
"fixed": "22.1.6"
}
],
"type": "ECOSYSTEM"
}
]
},
{
"package": {
"ecosystem": "npm",
"name": "@angular/platform-server"
},
"ranges": [
{
"events": [
{
"introduced": "21.0.0"
},
{
"fixed": "21.2.23"
}
],
"type": "ECOSYSTEM"
}
]
},
{
"package": {
"ecosystem": "npm",
"name": "@angular/platform-server"
},
"ranges": [
{
"events": [
{
"introduced": "20.0.0"
},
{
"fixed": "20.3.31"
}
],
"type": "ECOSYSTEM"
}
]
},
{
"package": {
"ecosystem": "npm",
"name": "@angular/platform-server"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"last_affected": "19.2.25"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-101895"
],
"database_specific": {
"cwe_ids": [
"CWE-400",
"CWE-835"
],
"github_reviewed": true,
"github_reviewed_at": "2026-09-28T21:31:21Z",
"nvd_published_at": null,
"severity": "HIGH"
},
"details": "A Denial of Service (DoS) vulnerability exists in `@angular/platform-server`\u0027s DOM emulation parser (`domino`). When processing untrusted user input containing an incomplete DOCTYPE declaration ending with whitespace before EOF (such as `\u003c!DOCTYPE html `), the HTML parser enters an infinite synchronous loop, pegging CPU utilization at 100% and completely freezing the Node.js server process.\n\n### Technical Description\nIn Angular Server-Side Rendering (SSR), `@angular/platform-server` uses `domino` to parse and sanitize HTML bound through template bindings (such as `[innerHTML]`) or manipulated via DOM APIs.\n\nIn Domino\u0027s HTML parser (`lib/HTMLParser.js`), tokenizer states that specify fixed lookahead\u2014such as `after_doctype_name_state` (`lookahead = 6`)\u2014rely on the state handler function to explicitly advance the character index pointer (`nextchar`). While branches for whitespace, `\u003e`, and keyword matching advance `nextchar`, the EOF branch (`case -1: // EOF`) emitted doctype and EOF tokens without advancing `nextchar` or transitioning out of the state:\n\n```javascript\ncase -1: // EOF\n forcequirks();\n emitDoctype();\n emitEOF();\n break;\n```\n\nBecause `nextchar` remained unchanged pointing to the EOF marker character (`\\uFFFF`), the scanner loop (`while (nextchar \u003c numchars)`) repeatedly re-invoked `after_doctype_name_state` with `codepoint = EOF` indefinitely. In Node.js\u0027s single-threaded runtime, this synchronous loop starves the event loop entirely.\n\n### Impact \u0026 Reachability\n* **Reachability**: The vulnerability is reachable in any Angular SSR application where untrusted user input is bound to `[innerHTML]`, interpolated into markup, or sanitized on the server.\n* **Impact**: Successful exploitation allows an unauthenticated remote attacker to cause an immediate Denial of Service (DoS) by sending a payload containing an incomplete DOCTYPE (e.g., `\u003c!DOCTYPE html `). The Node.js SSR process locks up at 100% CPU and ceases responding to all concurrent and subsequent HTTP requests.\n\n**Proof of Concept:**\n```ts\nimport { Component } from \u0027@angular/core\u0027;\n\n@Component({\n selector: \u0027app-root\u0027,\n standalone: true,\n template: `\u003cdiv [innerHTML]=\"payload\"\u003e\u003c/div\u003e`,\n})\nexport class AppComponent {\n // Attacker-controlled input containing an incomplete DOCTYPE ending with whitespace\n payload = \u0027\u003c!DOCTYPE html \u0027;\n}\n```\n\n### Workarounds\n* Avoid binding untrusted user input directly to `[innerHTML]` in server-rendered templates; use standard text interpolation (`{{ userInput }}`) or `[textContent]` when raw HTML rendering is not required.\n* Validate or sanitize user input before passing it to `[innerHTML]` on the server by stripping or rejecting strings matching `/^\u003c!DOCTYPE/i`.",
"id": "GHSA-f67j-2jqw-jpq7",
"modified": "2026-09-28T21:31:21Z",
"published": "2026-09-28T21:31:21Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/angular/angular/security/advisories/GHSA-f67j-2jqw-jpq7"
},
{
"type": "PACKAGE",
"url": "https://github.com/angular/angular"
}
],
"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": "Angular SSR: Denial of Service (DoS) via Infinite Loop on Malformed DOCTYPE"
}
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.