developer-cloud

RISC-V Developer Cloud

Welcome to the central repository for the RISC-V Developer Cloud management software, hardware bring-up specifications, and CI integration architecture.

About RISE

RISE is a collaborative, industry-led initiative under the Linux Foundation that accelerates open-source software development for the RISC-V architecture.

1. Vision & Purpose

The core mission of the RISC-V Developer Cloud is to democratize access to physical RISC-V hardware, accelerating software maturity, toolchain readiness, kernel enablement, and upstreaming across the entire ecosystem.

Moving Beyond QEMU: Software emulation via QEMU is widely available, but physical silicon is essential for validating real-world hardware performance, cache hierarchies, memory controller behavior, hardware vector extensions (RVV), peripheral driver stability, and true hardware regression testing.
Empowering Open-Source Maintainers: Providing reliable, automated bare-metal infrastructure enables distribution maintainers (Debian, Ubuntu, RHEL, Gentoo) and upstream developers to build, test, and benchmark daily mainline kernels and application toolchains directly on real RISC-V hardware.

2. Current Rollout Status

Deployment Location: Physical hardware is hosted and co-located at the Oregon State University (NORAM/USA), Scaleway (EMEA/France), and SOPIC (APAC/China).
Current Phase: Phase 1 (Labgrid Pilot MVP).
Active Hardware Cluster: The pilot is built around an initial allocation of 20 Alibaba Zhihe A210 boards donated by Alibaba.
Active Focus Areas:
- Configuring U-Boot TFTP boot and DHCP network boot pipelines.
- Integrating GitHub Actions self-hosted runners (board-farm-controller) for automated PR validation.

3. Architecture Roadmap & Phased Execution

The rollout is structured into two macro phases:

Phase 1 (Labgrid Pilot): Delivers automated hardware access for kernel testing, CI runners, and SSH debugging using a lightweight Labgrid setup.
Phase 2 (OpenStack Cloud Production): Scales management into a production multi-tenant bare-metal cloud leveraging OpenStack Ironic, Redfish APIs, and OpenBMC controllers.

Phase 1 (MVP)                             Phase 2 (Production Cloud)
+-----------------------------------+     +-----------------------------------------+
| GitHub Actions Workflow           |     | OpenStack Nova / Ironic Bare-Metal Cloud|
|           │                       |     |                    │                    |
|           ▼                       |     |                    ▼                    |
| Labgrid (Coordinator + Exporter)  | ──► | OpenBMC (D-Bus / Redfish on Exporter)   |
|           │                       |     |                    │                    |
|           ▼                       |     |                    ▼                    |
| DUT Serial/Power + U-Boot TFTP    |     | UEFI PXE Netboot + IPA eMMC Flashing    |
+-----------------------------------+     +-----------------------------------------+

📁 Architectural Specs (`docs/`)

Management Software Proposal: Overall project scope, phases, roadmap, and use cases.
Labgrid DHCP & TFTP Research: Labgrid 4-role process model, U-Boot TFTP vs PXE boot flows, and network isolation rules.
OpenBMC RISC-V BMC Plan: Implementation plan for OpenBMC multi-host Redfish controllers on x86 exporter hardware.
TFTP Boot & GHA Flashing Design: GitHub Actions workflows, TFTP boot, and exclusive container pool testing designs.
Build-Centric TFTP Storage & Release Lifecycle: Immutable per-build TFTP directory layout, Hetzner cross-compilation CI pipeline, 90-day retention policy, and monthly eMMC release cycle.