Rust library for easy implementation of FUSE filesystems in userspace.
This library does not just provide bindings, it is actually an improved rewrite of the original FUSE C library to fully take advantage of Rust's architecture.
A working FUSE filesystem consists of three parts:
- The kernel driver that registers as a filesystem and forwards operations into a communication channel to a userspace process that handles them.
- The userspace library (libfuse) that helps the userspace process to establish and run communication with the kernel driver.
- The userspace implementation that actually processes the filesystem operations.
The kernel driver is provided by the FUSE project, the userspace implementation needs to be provided by the developer. This Rust library provides a replacement for the libfuse userspace library between these two. This way, a developer can fully take advantage of the Rust type interface and runtime features when building a FUSE filesystem in Rust.
Except for a single setup (mount) function call and a final teardown (umount) function call to libfuse, everything runs in Rust.
To run a program that mounts a FUSE filesystem, the target system needs FUSE (OSXFUSE on macOS) to be properly installed (i.e. kernel driver and libraries. Some platforms may also require userland utils like fusermount
). A default installation of package fuse
on Linux, fusefs-libs
on FreeBSD, or OSXFUSE
on macOS is usually sufficient.
To build, the host system needs FUSE libraries and headers installed. On Linux, the header package is usually called libfuse-dev
. On FreeBSD and macOS, fusefs-libs
/OSXFUSE
installs everything that's needed. The build process also requires pkg-config
to locate headers and libraries.
Put this in your Cargo.toml
:
[dependencies]
fuse = "0.3"
and in your crate root:
extern crate fuse;
To create a new filesystem, implement the trait Filesystem
. Filesystem operations from the kernel are dispatched to the methods of the Filesystem
trait. Most methods get a reply
parameter that must be used to eventually answer the request. All methods have default implementations that reply with neutral answers, so if you implement no method at all, you still get a mountable filesystem that does nothing.
To actually mount the filesystem, pass an object that implements Filesystem
and the path of an (existing) mountpoint to the mount
function. mount
will not return until the filesystem is unmounted.
To mount a filesystem and keep running other code, use spawn_mount
instead of mount
. spawn_mount
spawns a background thread to handle filesystem operations while the filesystem is mounted. It returns a handle that should be stored to reference the mounted filesystem. If the handle is dropped, the filesystem is unmounted.
To unmount a filesystem, use any arbitrary unmount/eject method of your OS.
See the examples directory for some basic examples.
There's still a lot of stuff to be done. Feel free to contribute.
- Interrupting a filesystem operation isn't handled yet.
- An additional more high level API would be nice. It should provide pathnames instead inode numbers and automatically handle concurrency and interruption (like the FUSE C library's high level API).
In general, see the list of issues on GitHub and search the source files for comments containing "TODO" or "FIXME" to see what's still missing.
Developed and tested on macOS with OSXFUSE and on Linux with FUSE, using stable, beta and nightly Rust versions (see Travis CI for details).
FreeBSD also works, but is not tested in CI.