# `Tank.Runtime.Rootfs`
[🔗](https://github.com/oshlabs/tank/blob/v0.2.0/lib/tank/runtime/rootfs.ex#L1)

Host-side container rootfs bring-up, run during the `Linx.Process` `:ready`
checkpoint. Builds the container's filesystem inside its mount namespace and
pivots into it, leaving the workload ready to `execve`.

Every step runs in the child's mount namespace via `in: {:pid, host_pid}`.
The sequence mirrors the proven Linx M2 bring-up:

  1. **make `/` rec-private** — sever mount propagation so nothing leaks back
     to the host (a child's mount ns is a shared peer of the host's).
  2. **bind `rootfs` → `rootfs`, make it private** — `pivot_root` requires the
     new root to be a private mount point.
  3. **`/proc`** — mounted pidns-aware (Linx forks into the container's PID
     namespace, so `/proc` shows the container's pids).
  4. **`/dev`** — a fresh tmpfs with the standard device nodes bind-mounted
     from the host (`mknod` is barred in containers; binds are the way).
  5. **`/sys`** — recursive bind of the host's sysfs.
  6. **`pivot_root`** into `rootfs`, then detach the old root.

All mounts are set up *under* `rootfs` and come along with the pivot.

# `setup`

```elixir
@spec setup(pos_integer(), Path.t(), [{Path.t(), Path.t()}]) :: :ok | {:error, term()}
```

Bring up `rootfs` for the container parked at `host_pid`. On success the
child's mount namespace has `rootfs` as `/`, with `/proc`, `/dev`, and `/sys`
populated. Returns the first error encountered.

`etc_files` is a list of `{host_path, in_rootfs_path}` to bind into the rootfs
before the pivot — per-pod files (e.g. `/etc/resolv.conf`, `/etc/hosts`) that
must not mutate the shared, content-addressed image rootfs.

---

*Consult [api-reference.md](api-reference.md) for complete listing*