I remember implementing the original fork() call in the first Posix implementation. It consumed so much CPU time and resources (on machines with precious little of either) that I optimized it. It was often (usually) followed by exec() which combined meant, start this parallel process. The downside was, fork copied the entire parent process state, just to instantly throw it away on the next call.
So I made fork() do a lightweight copy (no pages, no tables, just mark the new process as a child of the older one) and upon exec 'forget' to complete the fork. Else the child process would fault on the incomplete copy as it executed and complete it then.
After 30 years in the industry, "trivial", "basic", and "simple" are words I've learned should never appear in the description for anything I expect to use. Ah, the author is the delusional sort, I'm sure the experience of using this thing will be perfectly smooth and well thought out.
Nice writeup. The step after forking paused sandboxes would be to fork live running sandboxes with the RAM, device state and disk all being forked. Get the same execution point with all processes still running rather than having to resume. You then also have to take care of kernel RNG
state, boot ID, clocks, cached session keys etc. before the forked version starts meaningful work.
I wrote up about our sandboxes. Everyone and their dog is offering them, but I think the fork operation is somewhat interesting at least for our own use - being able to fire copies of sandbox with a single API call has been convenient for our agentic use (reusing all earlier setup steps has saved lots of time and tokens).
Disclaimer: I know there's some more effort 'snapshotting' and other concepts elsewhere too, but from our point of view making it first-class citizen in both UI and API felt sensible and it has been useful.
You can do this at the container layer level with Dagger. Forking and time travel were my primary aims when building an agent sandbox on top of Dagger.
if you reuse a nonce with different data but the same key eg in AES-GCM, for example, so if you fork after creating an ephemeral key but before use thats quite likely, if the nonces are sequential.
I remember implementing the original fork() call in the first Posix implementation. It consumed so much CPU time and resources (on machines with precious little of either) that I optimized it. It was often (usually) followed by exec() which combined meant, start this parallel process. The downside was, fork copied the entire parent process state, just to instantly throw it away on the next call.
So I made fork() do a lightweight copy (no pages, no tables, just mark the new process as a child of the older one) and upon exec 'forget' to complete the fork. Else the child process would fault on the incomplete copy as it executed and complete it then.
Disclosure: I am working an open-source KVM-based VMM for agent sandboxes, https://github.com/instavm/tarit
right?
We know those work. But it would be crazy awesome to have a unbooted AND booted/execution image to finally sidestep the "worked on my machine" issue.
Disclaimer: I know there's some more effort 'snapshotting' and other concepts elsewhere too, but from our point of view making it first-class citizen in both UI and API felt sensible and it has been useful.
EDIT: I see you're only cloning filesystem contents, never mind.