-
Notifications
You must be signed in to change notification settings - Fork 0
Driver Model
This page describes the driver bring-up order and safety boundaries.
Driver work starts with the devices needed for QEMU on macOS:
- serial console;
- UEFI handoff data;
- PCI discovery;
- VirtIO block;
- VirtIO network;
- ACPI table parsing;
- NVMe;
- Intel NICs;
- server NICs;
- ARM/NVIDIA platform devices.
| Phase | Required drivers | Acceptance test |
|---|---|---|
| QEMU boot | serial, UEFI handoff | deterministic boot log appears |
| QEMU storage | PCI, VirtIO block | read a known test block from image |
| QEMU network | VirtIO net | exchange ICMP or UDP test packet |
| Intel Desktop | NVMe, first PCIe NIC | boot from NVMe and run latency test |
| Intel Xeon | server NIC, MSI-X, multi-queue NVMe, IOMMU | queues are NUMA-local and protected |
| ARM/NVIDIA | AArch64 timer, GIC, SMMU, platform discovery | boot, route interrupts, protect DMA |
Serial comes first because it is the boot and panic visibility path. VirtIO block and VirtIO net follow because they unlock disk-backed boot and control-plane networking.
Intel Desktop needs NVMe and PCIe NIC support. Intel Xeon needs NUMA-aware server NIC and multi-queue NVMe behavior. ARM/NVIDIA N1X-compatible systems need AArch64 boot, GIC, SMMU, and platform discovery.
DMA must be protected by IOMMU or SMMU where available. Device ownership must match AI Cell and service permissions.
Drivers must fail closed when DMA protection cannot be configured for a device that can access AI Cell memory.
This page defines the GitHub Wiki navigation sidebar.
- Architecture
- AI Cells
- CPU AI Runtime
- App Agents
- Memory System
- Networking
- Scheduler and Core Isolation
- Filesystem and Storage
- Driver Model
- Security Model
- Build System
- Build System
- Project Tracker
- Implementation Plan
- QEMU Full OS Core Workdown
- QEMU 100 Completion Plan
- Example Apps
- Codex Work Packages
- Testing and Benchmarking