Hello, in this tutorial, I'll show you how to run xfstests on a custom kernel. I needed to do this because I made a change to the fs subsystem and needed to understand how to run the tests after that change. Since it was a difficult task, I decided to repeat it a few times to understand how to do it and be able to teach you step by step. Also, I'd like to remind everyone not to make the same mistake I did by not properly testing a change made to the kernel. Don't just test build and boot.
First, let’s set up our environment by installing some tools we’ll need along the way. Just like a good hitchhiker of the galaxy, we always need to prepare our backpack before setting off—take everything that might be necessary. Here they are:
sudo dnf install build-essential ncurses-devel flex bison openssl-devel libssl-devel dkms elfutils-libelf-devel libudev-devel pciutils-devel python3-devel qemu-system-x86 debootstrap rsync
use your distro's package manager (e.g., apt).
.config in the root of the kernel source tree). For our case, use make menuconfig; this will open a simple but powerful menu to configure your kernel.make menuconfig
Inside menuconfig, you need to ensure some key variables are set. You can use the search function by pressing / and typing the variable name, and then by pressing 1 on your keyboard, you will navigate to the variable.. Make sure each of these is enabled with Y (built directly into the kernel, not as a module).
CONFIG_XFS_FSCONFIG_E1000ECONFIG_VIRTIO_NETCONFIG_VIRTIO_PCICONFIG_NET_FAILOVERCONFIG_E1000After saving the configuration, compile the kernel and headers. We gonna installs the kernel header files to a temporary directory. These headers are needed to compile the xfstests tool inside the VM.
make -j$(nproc)
make headers_install INSTALL_HDR_PATH=/tmp/kernel-headers
xfstests.First, we create a 20 GB file named disk.img using dd. This will be the home for our operating system.
dd if=/dev/zero of=disk.img bs=1M count=20480
Now we have to format this disk with the ext4 filesystem. This turns our empty file into a real, usable partition, ready to hold files and folders.
mkfs.ext4 disk.img
We use xfs_io to create a 10 GB file that will be our primary test disk.
xfs_io -f -c "falloc 0 10g" test.img
Since we are testing the XFS filesystem, we format this disk with mkfs.xfs.
mkfs.xfs test.img
xfs_io -f -c "falloc 0 10g" scratch.img
We have successfully created a file that represents a partition, so let’s mount it. For this step, I suggest creating a directory to serve as the mount point.
mkdir -p qemu-rootfs
sudo mount disk.img qemu-rootfs
Normally Debian is used for testing the kernel, so that will be the distro we will be using on our VM for testing our kernel. We will use a tool called debootstrap to create a Debian root filesystem (rootfs) within disk.img
sudo debootstrap --arch=amd64 bookworm qemu-rootfs http://deb.debian.org/debian/
Right now, the Debian filesystem is fully set up and includes the usual tools you’d expect from a regular Debian installation (browse the qemu-rootfs folder to see them). To work inside this environment, we can use a command called chroot (change root), which lets us switch into that filesystem. Keep in mind that running it requires sudo access
sudo chroot qemu-rootfs /bin/bash
Now you can navigate through the directories using ls and check the kernel version with uname -r, but the most important thing we need to do now is set a password for the root user. This will allow us to log in as root with our own password.
passwd
This command will prompt you to create a new password for the root user, which you will need to log in to the VM later. Type exit to leave the chroot environment.
These commands copy the compiled kernel headers into the VM's filesystem and unmount it .
sudo rsync -av /tmp/kernel-headers qemu-rootfs/usr/src/
sudo umount qemu-rootfs
Start the VM using your custom-compiled kernel and the prepared disks.
qemu-system-x86_64 \
-kernel <path_to_ur_kernel>arch/x86/boot/bzImage \
-hda disk.img \
-hdb test.img \
-hdc scratch.img \
-netdev user,id=mynet0 \
-device e1000,netdev=mynet0 \
-append "root=/dev/sda rw console=ttyS0" \
-nographic \
-smp 2 \
-m 2G
After the VM boots, log in as the root user with the password you just created.
Before we can install the tools we need, we must configure the network for our VM. This involves manually setting up the IP address.
List all network interfaces to find the name of the virtual network device (for example ens3):
ip link show
Enable the network interface
ip link set ens3 up
Obtain an IP address via DHCP
dhclient ens3
Set the DNS server so your VM can resolve domain names
echo "nameserver 8.8.8.8" > /etc/resolv.conf
After these steps, your VM will have network connectivity and you can proceed to install the required tools.
apt-get update
apt-get install -y pkg-config autoconf libtool
apt-get install -y acl attr automake bc dbench dump e2fsprogs fio gawk \
gcc git indent libacl1-dev libaio-dev libcap-dev libgdbm-dev libtool \
libtool-bin liburing-dev libuuid1 lvm2 make psmisc python3 quota sed \
uuid-dev uuid-runtime xfsprogs xfslibs-dev sqlite3 libgdbm-compat-dev
These commands install all dependencies for compiling xfstests.
xfstests:git clone https://git.kernel.org/pub/scm/fs/xfs/xfstests-dev.git
cd xfstests-dev
make
These commands download and compile the test suite.
mkdir -p /mnt/test
mkdir -p /mnt/scratch
mount /dev/sdb /mnt/test
cp local.config.example local.config
Next, edit the local.config file with vim:
export TEST_DEV=/dev/sdb
export TEST_DIR=/mnt/test
export SCRATCH_DEV=/dev/sdc
export SCRATCH_MNT=/mnt/scratch
export FSTYP=xfs
This configuration file tells xfstests which devices to use for testing.
With everything configured, you can now run the tests to validate your changes.
./check -g auto
./check xfs/001
Licensed under CC-BY-SA 4.0