28
Your Container Is Not a Sandbox (emirb.github.io)
submitted 6 months ago by to c/security@programming.dev
7 comments fedilink hide all child comments

all 7 comments

sorted by: hot top controversial new old
[–] 3 points 6 months ago (6 children)

For a proper sandbox (but also not a VM) on Linux, use Sydbox (or syd-oci). Virtual machines are an obvious choice, like said in the article, but a sandbox offers less overhead. The other issue with a VM is that, while it does isolate the guest from direct access to host resources, it doesn't stop the guest from doing whatever it wants (in the guest OS). The compromised guest could still attack the host or other network attached devices. Virt guests should still be configured with least privilege using a MAC (like SELinux or SMACK), or/and a sandbox policy (like with syd-oci).

  • permalink
  • fedilink
  • source
  • hideshow 6 child comments
    •
  • [–] 2 points 6 months ago* (3 children)

    That's some excessive text linking in the README I've not seen before. More blue than white, and three word-links one after another.

    Quite the contrast to the "only" three footnotes.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] 1 point 6 months ago (2 children)

    Is that a bad thing? I personally like links to where I can learn more.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 2 points 6 months ago*

    I don't. Many of the terms are easily searchable, and it's frustrating to click on one of them expecting to see syd-specific documentation about a topic or usecase only to see a generic post about a login shell (one of the links). It's trivial to highlight something and then right click and "Search DuckDuckGo for "highlighted term"" (firefox right click menu) nowadays. A search for "Login Shell Linux" nets that link they put in their documentation as literally the first result.

    I wish they only actually linked syd's internal documentation, maybe to stuff like the LWN articles explaining some of their design decisions

    Actually some of the links are easter eggs and they are pretty funny. Those can stay ig.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 1 point 6 months ago

    I find the text harder to read, the more significant links and terms or aspects harder to identify, and the consecutive-words-links unable to identify without hovering.

    I get the idea, but I think it can be done better, and if it's like this, between less and more, I prefer less - even if that means fewer references.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 2 points 6 months ago (1 child)

    Syd's architecture is similar to gvisor, mentioned in the article. It has similar tradeoffs, although I suspect it is more performant (can't find benchmarks) since it is written in rust, rather than go.

    Gvisor has some significant performance hits: https://gvisor.dev/docs/architecture_guide/performance/ . microvm/cloud hypervisor/ other vm solutions, with kvm, are around a 95% performance.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 1 point 6 months ago

    The advantage of an application kernel is that it reduces the access a running application has to exploit the kernel. Sure, with a VM the guest runs its own kernel. But the KVM hypervisor is still in the host's kernelspace. Implementing syscalls in userspace while using a safer subset of the kernel's syscalls helps prevent certain attacks. The performance hit is real of course. But syd has different goals than gVisor because it prevents apps from running unless they are given the permissions to do so.

  • permalink
  • fedilink
  • source
  • parent