The layers
1
The core
A small, trusted core handles the essentials: scheduling work across your processor, managing memory, letting parts of the system talk to each other, and enforcing who is allowed to do what.
2
Drivers and services
The parts that talk to hardware — disks, input, and more — and system services run outside the trusted core, in protected user space. Each gets only the narrow access it needs.
3
Applications
Your programs run further out still, each in its own protected space, able to touch only what you’ve granted them.
Why the core stays small
When a driver or service runs outside the trusted core, a fault in it stays contained — it can be stopped and restarted without bringing the whole system down. On many traditional systems a single driver runs with full trust, so one bad one can take everything with it. OMOS is deliberately built the other way.Everything talks through controlled channels
Parts of the system don’t reach into each other directly. They communicate over controlled channels, and a part can only send a message to something it has been given a handle to. That’s the same idea as the security model — access is something you hold and can pass on, never something that’s simply assumed.The same shape everywhere
This structure is identical on every supported chip. The details of talking to an Intel processor, an ARM one, or a RISC-V one differ underneath, but the shape of the system above that — the core, the protected drivers and services, your apps — is the same. See Platform support for what that covers today.Higher layers described in the vision — a graphical shell, networking, a distributed layer that pools machines — build on this foundation and are on the roadmap.