Conversation
I presented an overview of the Ultraviolet project at @linuxplumbersconf today. This is a proposed architecture and reference implementation for generalised code integrity in Linux. The goal of the presentation was to seek early feedback from the community and to invite collaboration. The slides are here: 🖥️ https://lpc.events/event/19/contributions/2222/
2
4
7

@jmorris @linuxplumbersconf What is container-specific about the proposed solution? Assuredly there's no reason that such images could not be booted on the metal? When folks talk about immutable image with measured boot and HW roots of trust, I think about @yoctoproject and embedded Linux.

1
0
0
@alison @yoctoproject @linuxplumbersconf the idea is that we want to use the container boundary as a well defined security abstraction, and then we can simplify the design of the system overall, i.e., the base system is essentially fixed, and there is a well defined “app model” at the container boundary. The type of container actually implemented is not specified — it can be a systemd portable service, for example. It would be possible to not use a standard container runtime, and build/use some lightweight name spacing scheme. We do want to avoid workloads which install directly and do whatever they want all over the system, as it is infeasible to secure at scale in a broadly adoptable manner.
0
0
0