Maybe I'm misunderstanding the video, but it looks to me as if the situation is:

You are root inside a sandbox. As root-in-the-sandbox, you create a symlink and this gives you the ability to escape the sandbox.

(Whether this is interesting or not depends on whether anyone actually tries to use the sandbox facility in such a way as to give root-in-the-sandbox privileges to untrusted people or code. I don't know enough about OpenBSD to answer that.)

OpenBSD doesn't do different user accounts inside vs outside sandboxes; if you're root in the sandbox, you're root on the system.

Also I tried the Dirtyfrag exploit under Bubblewrap for GNU/Linux. It lasted, but finally I got root with a simple 'su'.

unveil was designed and intended to effectively sandbox root when combined with sufficiently strict pledge permissions. I don't think this exploit would have effected any existing OpenBSD services, but sometimes services need to keep around processes with higher privileges than the network-facing process, yet you still want to sandbox them as much as possible. For example, sshd uses a special auth process, and that process needs higher privileges to be able to access the password database. On OpenBSD this auth process doesn't need root, but there may be similar cases where you want to use unveil with a root process for defense-in-depth. Suffice it to say, it would be foolish to only use unveil with such processes.

The bug here actually involved the intersection of unveil and pledge. IIUC, it was more a pledge bug that accidentally allowed bypassing unveil checks.

So what? You're still root. You're relying on a sandbox to plug a few voids while you still effectively held keys to the kingdom before said voids were plugged.

I hear this excuse daily from developers who insist on running all their docker containers as root "because we have to".

If you're relying on a sandbox as your first line of defense you've already lost the war.

I think the idea is to not run programs as root in the sandbox.

The parents tone wasn't warranted, but bugs like this could be more serious if combined with privilege escalation bugs in the sandbox.

Ideally, sandboxes should be like Vegas - what happens in the sandbox stays in the sandbox.

(I'm just speaking hypothetically here, I'm not knowledgeable about OpenBSD or it's sandboxes)

>Your "evidence" for him to reconsider is a sandbox "bypass" that requires you to be root to set up the environment

Can you help figure out where does it say unveil does not really work when root is involved?

You left a snarky comment, then paraded around a positively lame example as some sort of trophy.

Here's what I can figure out: you need root to set up the environment just so. It's a don't-care. The end.

So, a break out of chroot in a chroot jailed app would be a non-issue because I need root to set it up?

If you need root to set up the escape, then yes that is relatively uninteresting. Like, we know chroot can't contain root.

Thanks. It was not evident from the example whether root inside of the sandbox is necessary - I assumed creating arbitrary symlinks doesn't require any particular capabilities, and there's nothing special about the locations.

Though it's not clear to me now:

- why was this patched then?

- is the point about root that non-root wouldn't have access to passwd anyway?

OpenBSD doesn't have separate user accounts for sandboxes. These sandboxes are not linux-style containers, they're narrowed views of the full install.

If you're root inside the sandbox, you're root outside it. This exploit requires you to already be root.

But the issue of root and accessing outside of the sandbox is orthogonal, no? Even if you're logged in as XYZ, accessing XYZ's contents outside of the sandbox is still a breach and a problem. Or does this issue require actual root to manifest?

This path was special cased used to allow restricted applications to access time zone files, which are needed for time functions. Not any symlink will do, it has to be the specific one shown in the example exploit, or one of a small handful of others that were special cased for similar reasons. The place these symlinks live are owned by root. This is the same root user outside the sandbox as inside it.

So, yes, you need to have root on the box to set up this exploit.

I see, thank you for your time and patience spent to explain this. So there's no elevation, no general escape, and this got patched because it could possibly be used as a set-up-use-later backdoor style thing (such as dropping a setuid root binary somewhere in the OS). Yeah, not a thing I would use as an argument that it's a terribly insecure system.

>Here's what I can figure out: you need root to set up the environment just so.

I guess you just don't understand what unveil does.

Your arrogance is continued proof you could never comprehend the work that goes into building, releasing, and maintaining an entire OS, and your contributions will forever be limited to snarky negativity on message boards.

If you think their code sucks to the point people should think twice about using it, I suggest you stop using OpenSSH immediately.

Please be sure to let us know when your better, more secure replacement is ready.