nemu-local-test.md

• NEMU build
  • Cross building
• Testing
  • OVMF Logs
  • Direct kernel boot
    • Kernel
    • Initramfs image
    • Command line
  • Cloud image
    • Command line
  • Hotplug
    • CPU hotplug
    • Memory hotplug
    • PCI hotplug
  • Minimal CI
• ACPI
  • Tables dump

NEMU build

Configure script: configure.sh

git clone https://github.com/intel/nemu
cd nemu
git checkout -b topic/virt-x86 origin/topic/virt-x86
mkdir build-x86-64
cd build-x86-64
./configure.sh
make -j `nproc` 

Cross building

SRCDIR=/your/path/to/nemu/
make docker-image-debian-arm64-corss 
docker run --rm -it -v $SRCDIR:$SRCDIR -w $SRCDIR -u $(id -u) qemu:debian-arm64-cross

Testing

We can test the NEMU virt machine type through either direct kernel boot (a.k.a. nofw) or through a regular cloud or server image as long as it supports EFI boot.

OVMF logs

Add:

-device sysbus-debugcon,iobase=0x402,chardev=debugcon -chardev file,path=/tmp/debug-log,id=debugcon

to the qemu command line.

Direct kernel boot

Kernel

I use a 4.18.0 kernel with a custom .config

Initramfs image

See the busybox initramfs build instructions, and then configure the above kernel build to point to your local rootfs.

Command line

#!/bin/bash
# -*- mode: shell-script; indent-tabs-mode: nil; sh-basic-offset: 4; -*-
# ex: ts=8 sw=4 sts=4 et filetype=sh


sudo /home/samuel/devlp/hypervisor/nemu/build-x86-64/x86_64-softmmu/qemu-system-x86_64 \
     -nographic \
     -nodefaults \
     -L . \
     -net none \
     -machine virt,accel=kvm,kernel_irqchip,nofw \
     -smp sockets=1,cpus=4,cores=2,maxcpus=8 -cpu host \
     -m 1G,slots=3,maxmem=4G \
     -kernel /home/samuel/devlp/kernels/nfc/arch/x86/boot/compressed/vmlinux.bin -append 'console=hvc0 single iommu=false root=/dev/ram0' \
     -device virtio-serial-pci,id=virtio-serial0 -device virtconsole,chardev=charconsole0,id=console0 -chardev stdio,id=charconsole0 \
     -monitor telnet:127.0.0.1:55555,server,nowait

Cloud image

Warning Most cloud or server images kernels do not have HW reduced ACPI enabled so they won't support CPU and memory hotplug with our HW reduced ACPI virt machine type.

We have tested the latest Clear Linux cloud image with our custom OVMF binary.

Command line

#!/bin/bash
# -*- mode: shell-script; indent-tabs-mode: nil; sh-basic-offset: 4; -*-
# ex: ts=8 sw=4 sts=4 et filetype=sh

sudo /home/samuel/devlp/hypervisor/nemu/build-x86_64/x86_64_virt-softmmu/qemu-system-x86_64_virt \
     -bios $HOME/devlp/hypervisor/edk2/Build/OvmfX64/DEBUG_GCC5/FV/OVMF.fd \
     -nographic \
     -nodefaults \
     -L . \
     -net none \
     -machine virt,accel=kvm,kernel_irqchip \
     -smp sockets=1,cpus=4,cores=2,maxcpus=8 -cpu host \
     -m 1G,slots=3,maxmem=4G \
     -kernel /home/samuel/devlp/kernels/linux-nemu/arch/x86/boot/bzImage -append 'console=hvc0 iommu=false root=/dev/vda3 rw rootfstype=ext4 data=ordered rcupdate.rcu_expedited=1 tsc=reliable no_timer_check reboot=t noapictimer' \
     -device virtio-serial-pci,id=virtio-serial0 -device virtconsole,chardev=charconsole0,id=console0 -chardev stdio,id=charconsole0 \
     -drive file=/home/samuel/devlp/hypervisor/nemu/clear-cloud/clear-24570-kvm.img,if=virtio,format=raw \
     -monitor telnet:127.0.0.1:55555,server,nowait

Hotplug

Hotplug is done through the QEMU monitoring socket (-monitor telnet:127.0.0.1:55555,server,nowait from the above command lines). The socket is reachable through port 23 (telnet):

# telnet 127.0.0.1 55555

CPU hotplug

Plug

(qemu) device_add host-x86_64-cpu,id=core4,socket-id=1,core-id=1,thread-id=0 

Unplug

(qemu) device_del core4

Memory hotplug

Plug

(qemu) object_add memory-backend-ram,id=mem1,size=1G
(qemu) device_add pc-dimm,id=dimm1,memdev=mem1

Unplug

(qemu) device_del dimm1

PCI hotplug

Plug

(qemu) device_add virtio-net-pci,id=virtio-net1

Unplug

(qemu) device_del virtio-net1

Minimal CI

# cd tools/CI/
# sudo -E ./minimal_ci.sh -cloudinit ~/devlp/hypervisor/nemu/tools/CI/cloud-init/ -hypervisor ~/devlp/hypervisor/nemu/build-x86-64/x86_64-softmmu/qemu-system-x86_64 -workloads ~/workloads/

ACPI

Tables dump

We'll take the DSDT as an example, but this can be applied to all ACPI tables:

1. Install the acpica-tools package from your distro
2. From the guest command line:

cat /sys/firmware/acpi/tables/DSDT > dsdt.dat

3. Copy dsdt.dat back to the host, for example:

scp dsdt.dat hostusername@host-tun-ip:~/

4. From the host, disassemble the table

iasl -d dsdt.dat
iasl -tc dsdt.dsl