<IdleBot_5e50c57d>
Intuitively, building CrosVM in a container using Nix should not be that different
manveru has quit [*.net *.split]
spacekookie has quit [*.net *.split]
awordnot has joined #spectrum
manveru has joined #spectrum
spacekookie has joined #spectrum
awordnot has quit [Read error: Connection reset by peer]
awordnot has joined #spectrum
<qyliss>
It's not the build that fails. It's a runtime failure.
<IdleBot_5e50c57d>
Well, given that I construct a container when running CrosVM anyway…
<IdleBot_5e50c57d>
What is the current failure?
<qyliss>
same one as before because I started doing Nixpkgs things and didn't come back to it to do socat
<IdleBot_5e50c57d>
Ah OK. That should be easy to fix once you approach it correctly. What else is needed: KVM inside the container (obviously), /var/empty (it is really unhappy otherwise), preferably chronos user and group existing inside (seems optional but less spam easier debugging)
<qyliss>
Yikes on the user
<qyliss>
I've never had it complain about that when I just run it on my NixOS system
<IdleBot_5e50c57d>
Presumably you know its tuntap needs better than me anyway. 9p… you still do not use 9p, right?
<qyliss>
not the built in one
<IdleBot_5e50c57d>
Re: 9p — ah cool, for rust-9p, say, there should be no problems
<IdleBot_5e50c57d>
Re: chronos — are you sure it does not complain? As it runs anyway, it can be immediately scrolled out by kernel log
<IdleBot_5e50c57d>
Or maybe it is because I do not let CrosVM be a EUID-0 process at any time, not even the EUID-0-in-NS