Skip to content

CARV-ICS-FORTH/qemu-riscv64

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

RISC-V QEMU in Kubernetes

This project provides an easy method for deploying RISC-V QEMU instances in Kubernetes. Using Helm, you can quickly configure and deploy a RISC-V VM and access its console through a web interface. The Helm chart is compatible with Knot. Pre-compiled configurations support RVV 1.0 and RVV 0.7.1.

Deploying

When runnning Kubernetes in Docker Desktop, you can just run the following and visit http://localhost:8080:

helm install myvm ./chart/qemu-riscv64

To run a second instance, specify a different port:

helm install myvm ./chart/qemu-riscv64 \
    --set service.port=8081

If running Kubernetes with minikube, the above commands apply, but you also need to enable service forwarding to localhost with minikube tunnel.

The table below lists all available options (the current version is set in the Makefile):

Variable Default Notes
images.launcher carvicsforth/qemu-riscv64-launcher:<version> Default launcher, supporting RVV 1.0
images.launcherRVV071 carvicsforth/qemu-riscv64-launcher-rvv-0.7.1:<version> Custom launcher, supporting RVV 0.7.1
images.console carvicsforth/qemu-riscv64-console:<version>
images.data carvicsforth/qemu-riscv64-ubuntu:22.04-1 Use for custom files, kernel, and BIOS
service.type LoadBalancer
service.port 8080
cpu.features Can be overriden by the data image
cpu.count 2 Also sets resource request in deployment
memory 2048 Also sets resource request in deployment
append Kernel options. Can be overriden by the data image

Available data images:

Image Notes
carvicsforth/qemu-riscv64-ubuntu:22.04-1 Standard Ubuntu image (login: ubuntu/ubuntu)
carvicsforth/qemu-riscv64-buildroot:00709af-1 Custom built image, uses RVV 1.0 (login: root/)
carvicsforth/qemu-riscv64-buildroot-rvv-0.7.1:954aeb7-1 Custom built image, uses RVV 0.7.1 (login: root/)

To images set the CPU specifications they need, as well as custom kernel options. If the image name contains rvv-0.7.1, the chart automatically switches to the respective launcher. To run the custom Buildroot image:

helm install myvm ./chart/qemu-riscv64 \
    --set images.data="carvicsforth/qemu-riscv64-buildroot:00709af-1"

Architecture

We use two basic containers for each VM:

  • The launcher contains the QEMU binary. A script starts QEMU automatically when it runs, using environmental variables to define the number of CPUs, the memory, and other parameters. QEMU is configured to serve its console via telnet at local port 10023. Networking is handled by passt, so the VM gets the same IP as the pod and can communicate with other VMs in the same Kubernetes environment.
  • The console is based on gotty. Gotty runs a local command (in this case telnet), captures input and output streams, exposes them via websockets, and then offers a web-based terminal emulator (xterm.js) for the user to interact with what has been run.

The root filesystem used by QEMU is also packaged up in a container image (which we call data), using a similar method to KubeVirt. Before each VM starts, the data container copies its contents to an ephemeral volume that is shared with the launcher. As data contains only VM-specific files (it can optionally also hold a custom kernel and BIOS), we copy over a BusyBox binary before it starts, so it can run cp to populate the shared volume.

Building images

Container images are available for both amd64 and arm64 architectures.

To build your own locally, set the REGISTRY variable and run:

REGISTRY=<Docker Hub username> make container

The image tag for the launcher and console images is controlled via the VERSION variable.

Acknowledgements

This project has received funding from the European Union’s Horizon Europe research and innovation programme through project RISER ("RISC-V for Cloud Services", GA-101092993) and from the Key Digital Technologies Joint Undertaking through project REBECCA ("Reconfigurable Heterogeneous Highly Parallel Processing Platform for safe and secure AI", GA-101097224). KDT JU projects are jointly funded by the European Commission and the involved state members (including the Greek General Secretariat for Research and Innovation).