NsenterWriter is an arbitrary shell-execution / path traversal attack waiting to happen

[smarterclayton]

EDIT: This is not a vulnerability right now, but it is a potential one.

While fixing #16969 we noticed that NsenterWriter has the potential for arbitrary shell-execution if a caller doesn't validate their paths. We need to ensure that is impossible by invoking filepath.Clean on the passed input, and then using %q to verify that the shell we generate cannot be escaped. However, we may need to go even further (since bash + escaping == epic disaster).

Docker 1.10 should have a mount propagation setting which allows --shared mount, at which point this code should die and be totally removed. We should leave the containerized code in experimental and put big security warnings around it until we have that.

I'd also like to figure out how we can have much stronger review policies around places where we execute bash code in the node, by requiring it go through a strict review process, and having specific helpers. I'd also like to put some automation in place to ensure we do not execute shell except through whitelisted paths.

@kubernetes/rh-platform-management @kubernetes/goog-node @liggitt

[smarterclayton]

A quick review indicated this is the only potential spot for this in the codebase right now but that could change.

[pmorie]

Touchpoint: containerized exec #13691

Also, if we have to support containerizing the kubelet on pre-1.10, we will need to keep this around :-/

@smarterclayton, do you think that will be a need?

[timstclair]

I think in this specific case you can just use dd of= instead of cat > to avoid needing the shell.

[fejta-bot]

Issues go stale after 30d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

Prevent issues from auto-closing with an /lifecycle frozen comment.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or @fejta.
/lifecycle stale

[fejta-bot]

Stale issues rot after 30d of inactivity.
Mark the issue as fresh with /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or @fejta.
/lifecycle rotten
/remove-lifecycle stale

[fejta-bot]

Rotten issues close after 30d of inactivity.
Reopen the issue with /reopen.
Mark the issue as fresh with /remove-lifecycle rotten.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/close

[tallclair]

/reopen
/assign

[k8s-ci-robot]

@tallclair: you can't re-open an issue/PR unless you authored it or you are assigned to it.

In response to this:

/reopen
/assign

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.