axdevice is a reusable, OS-agnostic device abstraction layer designed for virtual machines. It allows dynamic device configuration and MMIO emulation in no_std
environments, making it suitable for hypervisors or operating systems targeting RISC-V or AArch64.
- 📦 Componentized: Designed as a modular crate to be integrated into any OS or hypervisor.
- 🧩 Flexible device abstraction: Supports dynamic device registration and MMIO handling.
- 🛠️ No
std
required: Usesalloc
andcore
only, suitable for bare-metal development. - 🧵 Thread-safe: Devices are stored using
Arc
, ready for multicore use. - 🧱 Easily extensible: Just plug in device types via
axdevice_base::BaseDeviceOps
.
config.rs
: DefinesAxVmDeviceConfig
, a wrapper for device configuration input.device.rs
: DefinesAxVmDevices
, manages and dispatches MMIO to registered devices.
+-------------------+
| axvmconfig | <- defines EmulatedDeviceConfig
+-------------------+
|
v
+------------------+ uses +-----------------------+
| axdevice +-------------->+ axdevice_base::trait |
| (this crate) | +-----------------------+
+------------------+ ^
| |
v |
+------------------+ |
| axaddrspace | -- GuestPhysAddr ----+
+------------------+
[1] Load VM device config (Vec<EmulatedDeviceConfig>)
↓
[2] Create AxVmDeviceConfig
↓
[3] Pass into AxVmDevices::new()
↓
[4] MMIO access triggers handle_mmio_{read,write}
↓
[5] Device selected by GuestPhysAddr
↓
[6] Forwarded to BaseDeviceOps::handle_{read,write}()
use axdevice::{AxVmDeviceConfig, AxVmDevices};
// Step 1: Load configuration (e.g. from .toml or hypervisor setup)
let config = AxVmDeviceConfig::new(vec![/* EmulatedDeviceConfig */]);
// Step 2: Initialize devices
let devices = AxVmDevices::new(config);
// Step 3: Emulate MMIO access
let _ = devices.handle_mmio_read(0x1000_0000, 4);
devices.handle_mmio_write(0x1000_0000, 4, 0xdead_beef);
axvmconfig
axaddrspace
axdevice_base
log
,alloc
,cfg-if
,axerrno