| b.liu | e958203 | 2025-04-17 19:18:16 +0800 | [diff] [blame] | 1 | .. SPDX-License-Identifier: GPL-2.0 |
| 2 | |
| 3 | =================================================== |
| 4 | virtiofs: virtio-fs host<->guest shared file system |
| 5 | =================================================== |
| 6 | |
| 7 | - Copyright (C) 2019 Red Hat, Inc. |
| 8 | |
| 9 | Introduction |
| 10 | ============ |
| 11 | The virtiofs file system for Linux implements a driver for the paravirtualized |
| 12 | VIRTIO "virtio-fs" device for guest<->host file system sharing. It allows a |
| 13 | guest to mount a directory that has been exported on the host. |
| 14 | |
| 15 | Guests often require access to files residing on the host or remote systems. |
| 16 | Use cases include making files available to new guests during installation, |
| 17 | booting from a root file system located on the host, persistent storage for |
| 18 | stateless or ephemeral guests, and sharing a directory between guests. |
| 19 | |
| 20 | Although it is possible to use existing network file systems for some of these |
| 21 | tasks, they require configuration steps that are hard to automate and they |
| 22 | expose the storage network to the guest. The virtio-fs device was designed to |
| 23 | solve these problems by providing file system access without networking. |
| 24 | |
| 25 | Furthermore the virtio-fs device takes advantage of the co-location of the |
| 26 | guest and host to increase performance and provide semantics that are not |
| 27 | possible with network file systems. |
| 28 | |
| 29 | Usage |
| 30 | ===== |
| 31 | Mount file system with tag ``myfs`` on ``/mnt``: |
| 32 | |
| 33 | .. code-block:: sh |
| 34 | |
| 35 | guest# mount -t virtiofs myfs /mnt |
| 36 | |
| 37 | Please see https://virtio-fs.gitlab.io/ for details on how to configure QEMU |
| 38 | and the virtiofsd daemon. |
| 39 | |
| 40 | Internals |
| 41 | ========= |
| 42 | Since the virtio-fs device uses the FUSE protocol for file system requests, the |
| 43 | virtiofs file system for Linux is integrated closely with the FUSE file system |
| 44 | client. The guest acts as the FUSE client while the host acts as the FUSE |
| 45 | server. The /dev/fuse interface between the kernel and userspace is replaced |
| 46 | with the virtio-fs device interface. |
| 47 | |
| 48 | FUSE requests are placed into a virtqueue and processed by the host. The |
| 49 | response portion of the buffer is filled in by the host and the guest handles |
| 50 | the request completion. |
| 51 | |
| 52 | Mapping /dev/fuse to virtqueues requires solving differences in semantics |
| 53 | between /dev/fuse and virtqueues. Each time the /dev/fuse device is read, the |
| 54 | FUSE client may choose which request to transfer, making it possible to |
| 55 | prioritize certain requests over others. Virtqueues have queue semantics and |
| 56 | it is not possible to change the order of requests that have been enqueued. |
| 57 | This is especially important if the virtqueue becomes full since it is then |
| 58 | impossible to add high priority requests. In order to address this difference, |
| 59 | the virtio-fs device uses a "hiprio" virtqueue specifically for requests that |
| 60 | have priority over normal requests. |