Hacker News Reader

Comment by MobiusHorizons | Hacker News Reader

What is the purpose of compiling this to web assembly? What web assembly runtimes are there where there is not already an easily accessible (substantially faster) js execution environment? I know wasmtime exists and is not tied to a js execution engine like basically every other web assembly implementation, but the uses of wasmtime are not restricted from dependencies like v8 or jsc. Usually web assembly is used for providing sandboxing something a js execution environment is already designed to provide, and is only used when the code that requires sandboxing is native code not javascript. It sounds like a good way to waste a lot of performance for some additional sandboxing, but I can't imagine why you would ever design a system that way if you could choose a different (already available and higher performance) sandbox.

View on HN

I want to build features - both client- and server-side - where users can provide JavaScript code that I then execute safely.

Just having a WebAssembly engine available isn't enough for this - something has to take that user-provided string of JavaScript and execute it within a safe sandbox.

Generally that means you need a JavaScript interpreter that has itself been compiled to WebAssembly. I've experimented with QuickJS itself for that in the past - demo here: https://tools.simonwillison.net/quickjs - but MicroQuickJS may be interesting as a smaller alternative.

If there's a better option than that I'd love to hear about it!

GraalVM supports running javascript in a sandbox with a bunch of convenient options for running untrusted code.

https://www.graalvm.org/latest/security-guide/sandboxing/

Oh that looks neat! It appears to have the memory limits I want (engine.MaxIsolateMemory) and a robust CPU limit: sandbox.MaxCPUTime

One catch: the sandboxing feature isn't in the "community edition", so only available under the non-open-source (but still sometimes free, I think?) Oracle GraalVM.

This is generally the purpose of JavaScript execution environments like v8 or jsc (or quickjs although I understand not trusting that as a sandbox to the same degree). They are specifically intended for executing untrusted scripts (eg web browsers). Web assembly’s sandboxing comes from js sandboxing, since it was originally a feature of the same programs for the same reasons. Wrapping one sandbox in another is what I’m surprised by.

Running v8 itself as a sandbox is non-trivial, at least embedded in a Python or Node.js or similar application.

The web is littered with libraries that half do that and then have a note in the README that says "do not rely on this as a secure sandbox".

Is it though? I have not personally used these libraries, but a cursory google search reveals several options: - cloudflare/STPyV8: [0] From cloudflare, intended for executing untrusted code. - Pythonmonkey: [1] Embeds spidermonkey. Not clearly security focused, but sandboxing untrusted code is literally the point of browser js engines.

It's a little less clear how you would do this from node, but the v8 embedding instructions should work https://v8.dev/docs/embed even if nodejs is already a copy of v8.

[0]: https://github.com/cloudflare/stpyv8 [1]: https://docs.pythonmonkey.io

... whoa, I don't know how I missed it but I hadn't seen STPyV8 before.

I'd seen PyV8 and ruled it out as very unmaintained.

One of my requirements for a sandbox is that it needs to me maintained by a team of professionals who are running a multi-million dollar business on it. Cloudflare certainly count! I wonder what they use STPyV8 for themselves?

... on closer inspection it doesn't seem to have the feature I care most about: the ability to constrain memory usage (see comment here https://github.com/cloudflare/stpyv8/blob/57e881c7fbe178c598...) - and there's no built-in support for time limits either, you have to cancel tasks from a separate thread: https://github.com/cloudflare/stpyv8/issues/112

PythonMonkey looks promising too: the documentation says "MVP as of September 2024" so clearly it's intended to be stable for production use at this point.

I’m sure you are aware the sandbox that requires maintaining is v8 itself. Of course there are ways for the wrapper to break the sandbox by providing too much in thr global context, but short of that, which the application code could easily do as well, I don’t see why a wrapper should require significant resources to maintain beyond consuming regular updates from upstream. Is there some other reason you hold such a high bar for what is basically just python glue code for the underlying v8 embed api?

None of those v8 solutions provide what I need:

1. The ability to restrict the amount of memory that the sandboxed code can use

2. The ability to set a reliable time limit on execution after which the code will be terminated

My third requirement is that a company with real money on the line and a professional security team is actively maintaining the library. I don't want to be the first person to find out about any exploits!

Gotcha I hadn’t factored those capabilities into the concept of sandbox, but I can see why they would be important features.

I will admit I don’t really understand why the library that wraps v8 requires a security team in your view, given that v8 itself definitely has one. I’m trying to understand what you see as the dangerous piece of such code likely to lead to exploits. I’m probably missing something, but I fail to see where the complexity lies.

The biggest one is I don't want someone submitting malicious (or just poorly designed) code that crashes my server - hence the focus on memory and CPU limits.

I also need to limit filesystem access - don't want them stealing private files from elsewhere on the system, or filling the disk with garbage data (again causing a crash).

Network access restrictions are important too - I don't want my server becoming part of some DDoS attack, or an attacker using it to hit supposedly safe internal endpoints (SSRF).

As I noted in another comment Figma has used QuickJS to run JS inside Wasm ever since a security vulnerability was discovered in their previous implementation.

In a browser environment it's much easier to sandbox Wasm successfully than to sandbox JS.

That’s very interesting! Have they documented the reasoning for that approach? I would have expected iframes to be both simpler and faster sandboxing mechanism especially in compute bound cases. Maybe the communication overhead is too high in their workload?

EDIT: found this from your other comment: https://www.figma.com/blog/an-update-on-plugin-security/ they do not address any alternatives considered.

Reply on news.ycombinator.com