About

Welcome to the Void Handbook! Please be sure to read the "About This Handbook" section to learn how to use this documentation effectively. A local copy of this handbook, in several formats, can be installed via the void-docs package and accessed with the void-docs(1) utility.

Void is an independent, rolling release Linux distribution, developed from scratch rather than as a fork, with a focus on stability over bleeding-edge. In addition, there are several features that make Void unique:

  • The XBPS package manager, which is extremely fast, developed in-house, and performs checks when installing updates to ensure that libraries are not changed to incompatible versions which can break dependencies.
  • The musl libc, which focuses on standards compliance and correctness, has first class support. This allows us to build certain components for musl systems statically, which would not be practical on glibc systems.
  • runit is used for init(8) and service supervision. This allows Void to support musl as a second libc choice, which would not be possible with systemd. A side effect of this decision is a core system with clean and efficient operation, and a small code base.

Void is developed in the spare time of a handful of developers, and is generally considered stable enough for daily use. We do this for fun and hope that our work will be useful to others.

The name "Void" comes from the C literal void. It was chosen rather randomly, and is void of any meaning.

History

Knowledge of the ancients, grepped from the Git logs themselves:

  • 2008-09-26: first Git import of void-packages
  • 2009-08-17: first Git import of xbps
  • 2011-06-25: first systemd commit in void-packages
  • 2013-03-01: first musl toolchains added
  • 2014-07-14: begin switching to LibreSSL
  • 2014-07-28: switch from systemd to runit
  • 2015-07-09: full aarch64 support with linux4.1
  • 2018-07-06: first use of Terraform for GitHub permissions, for increased transparency
  • 2021-03-05: begin switching to OpenSSL

About This Handbook

This handbook is not an extensive guide on how to use and configure common Linux software. The purpose of this document is to explain how to install, configure, and maintain Void Linux systems, and to highlight the differences between common Linux distributions and Void.

To search for a particular term within the Handbook, select the 'magnifying glass' icon, or press 's'.

Those looking for tips and tricks on how to configure a Linux system in general should consult upstream software documentation. Additionally, the Arch Wiki provides a fairly comprehensive outline of common Linux software configuration, and a variety of internet search engines are available for further assistance.

Reading The Manuals

While this handbook does not provide a large amount of copy and paste configuration instructions, it does provide links to the man pages for the referenced software wherever possible.

To learn how to use the man(1) man page viewer, run the command man man. It can be configured by editing /etc/man.conf; read man.conf(5) for details.

Void uses the mandoc toolset for man pages. mandoc was formerly known as "mdocml", and is provided by the mdocml package.

Example Commands

Examples in this guide may have snippets of commands to be run in your shell. When you see these, any line beginning with $ is run as your normal user. Lines beginning with # are run as root. After either of these lines, there may be example output from the command.

Placeholders

Some examples include text with placeholders. Placeholders indicate where you should substitute the appropriate information. For example:

# ln -s /etc/sv/<service_name> /var/service/

This means you need to substitute the text <service_name> with the actual service name.

InfraDocs

InfraDocs is the meta-manual for the Void project systems management.

Installation

This section includes general information about the process of installing Void. For specific guides, see the "Advanced Installation" section.

Base system requirements

Void can be installed on very minimalist hardware, though we recommend the following minimums for most installations:

ArchitectureCPURAMStorage
x86_64-glibcx86_6496MB700MB
x86_64-muslx86_6496MB600MB
i686-glibcPentium 4 (SSE2)96MB700MB

Note that xfce image installations require more resources.

Void is not available for the i386, i486, or i586 architectures.

Before installing musl Void, please read the "musl" section of this Handbook, so that you are aware of software incompatibilities.

It is highly recommended to have a network connection available during install to download updates, but this is not required. ISO images contain installation data on-disk and can be installed without network connectivity.

Downloading installation media

The most recent live images and rootfs tarballs can be downloaded from https://repo-default.voidlinux.org/live/current/. They can also be downloaded from other mirrors. Previous releases can be found under https://repo-default.voidlinux.org/live/, organized by date.

Verifying images

Each image release's directory contains two files used to verify the image(s) you download. First, there is a sha256sum.txt file containing image checksums to verify the integrity of the downloaded images. Second is the sha256sum.sig file, used to verify the authenticity of the checksums.

It is necessary to verify both the image's integrity and authenticity. It is, therefore, recommended that you download both files.

Verifying image integrity

You can verify the integrity of a downloaded file using sha256sum(1) with the sha256sum.txt file downloaded above. The following command will check the integrity of only the image(s) you have downloaded:

$ sha256sum -c --ignore-missing sha256sum.txt
void-live-x86_64-musl-20170220.iso: OK

This verifies that the image is not corrupt.

Verifying digital signature

Prior to using any image you're strongly encouraged to validate the signatures on the image to ensure they haven't been tampered with.

Current images are signed using a minisign key that is specific to the release. If you're on Void already, you can obtain the keys from the void-release-keys package, which will be downloaded using your existing XBPS trust relationship with your mirror and package signatures. You will also need a copy of minisign(1); on Void, this is provided by the minisign package.

The minisign executable is usually provided by a package of the same name, and can also be installed on Windows, even without WSL or MinGW.

If you are not currently using Void Linux, it will also be necessary to obtain the appropriate signing key from our Git repository here.

Once you've obtained the key, you can verify your image with the sha256sum.sig and sha256sum.txt files. First, you need to verify the authenticity of the sha256sum.txt file.

The following example demonstrates the verification of the sha256sum.txt file for the 20230628 images with minisign:

$ minisign -V -p /usr/share/void-release-keys/void-release-20230628.pub -x sha256sum.sig -m sha256sum.txt
Signature and comment signature verified
Trusted comment: This key is only valid for images with date 20230628.

Finally, you need to verify that the checksum for your image matches the one in the sha256sum.txt file. This can be done with the sha256(1) utility from the outils package, as demonstrated below for the 20230628 x86_64 base image:

$ sha256 -C sha256sum.txt void-live-x86_64-20230628-base.iso 
(SHA256) void-live-x86_64-20230628-base.iso: OK

Alternatively, if the sha256 utility isn't available to you, you can use sha256sum(1):

$ sha256sum -c sha256sum.txt --ignore-missing
void-live-x86_64-20230628-base.iso: OK

If neither program is available to you, you can compute the SHA256 hash of the file and compare it to the value contained in sha256sum.txt.

If the verification process does not produce the expected "OK" status, do not use it! Please alert the Void Linux team of where you got the image and how you verified it, and we will follow up on it.

Live Installers

Void provides live installer images containing a base set of utilities, an installer program, and package files to install a new Void system. These live images are also useful for repairing a system that is not able to boot or function properly.

There are x86_64 images for both glibc and musl based systems. There are also images for i686, but only glibc is supported for this architecture. Live installers are not provided for other architectures. Users of other architectures will need to use rootfs tarballs, or perform an installation manually.

Installer images

Void releases two types of images: base images and xfce images. Linux beginners are encouraged to try one of the more full-featured xfce images, but more advanced users may often prefer to start from a base image to install only the packages they need.

Base images

The base images provide only a minimal set of packages to install a usable Void system. These base packages are only those needed to configure a new machine, update the system, and install additional packages from repositories.

Xfce image

The xfce image includes a full desktop environment, web browser, and basic applications configured for that environment. The only difference from the base images is the additional packages and services installed.

The following software is included:

  • Window manager: xfwm4
  • File manager: Thunar
  • Web Browser: Firefox
  • Terminal: xfce4-terminal
  • Plain text editor: Mousepad
  • Image viewer: Ristretto
  • Other: Bulk rename, Orage Globaltime, Orage Calendar, Task Manager, Parole Media Player, Audio Mixer, MIME type editor, Application finder

The install process for the xfce image is the same as the base images, except that you must select the Local source when installing. If you select Network instead, the installer will download and install the latest version of the base system, without any additional packages included on the live image.

Accessibility support

All Void installer images support the console screenreader espeakup and the console braille display driver brltty. These services can be enabled at boot by pressing s in the bootloader menu to enable accessibility support. On UEFI-based systems, GRUB is the bootloader, and it will play a two-tone chime when the menu is available. On BIOS-based systems and UEFI systems in legacy/compatibility mode, SYSLINUX is the bootloader, and no chime is played. SYSLINUX also requires pressing the enter key after pressing s. The hotkey r will also boot with accessibility support, but will load the live ISO into RAM.

After booting into the installer image with accessibility support enabled, if there are multiple soundcards detected, a short audio menu allows for the selection of the soundcard for the screenreader. Press enter when the beep for the desired soundcard is heard to select it.

If the Local installation source is selected in the installer, espeakup and brltty will also be installed and enabled on the installed system if enabled in the live environment.

The xfce image also supports the graphical screenreader orca. This can be enabled by pressing Win + R and entering orca -r. Orca will also be available on the installed system if the Local installation source is selected.

Kernel Command-line Parameters

Void installer images support several kernel command-line arguments that can change the behavior of the live system. See the void-mklive README for a full list.

Prepare Installation Media

After downloading a live image, it must be written to bootable media, such as a USB drive, SD card, or CD/DVD.

Create a bootable USB drive or SD card on Linux

Identify the Device

Before writing the image, identify the device you'll write it to. You can do this using fdisk(8). After connecting the storage device, identify the device path by running:

# fdisk -l
Disk /dev/sda: 7.5 GiB, 8036286464 bytes, 15695872 sectors
Disk model: Your USB Device's Model
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes

In the example above, the output shows the USB device as /dev/sda. On Linux, the path to the device will typically be in the form of /dev/sdX (where X is a letter) for USB devices, /dev/mmcblkX for SD cards, or other variations depending on the device. You can use the model and size (7.5GiB above, after the path) to identify the device if you're not sure what path it will have.

Once you've identified the device you'll use, ensure it's not mounted by unmounting it with umount(8):

# umount /dev/sdX
umount: /dev/sdX: not mounted.

Write the live image

The dd(1) command can be used to copy a live image to a storage device. Using dd, write the live image to the device:

Warning: this will destroy any data currently on the referenced device. Exercise caution.

# dd bs=4M if=/path/to/void-live-ARCH-DATE-VARIANT.iso of=/dev/sdX
90+0 records in
90+0 records out
377487360 bytes (377 MB, 360 MiB) copied, 0.461442 s, 818 MB/s

dd won't print anything until it's completed (or if it failed), so, depending on the device, this can take a few minutes or longer. You can enable printing by adding status=progress to the command if using GNU coreutils dd.

Finally, ensure all data is flushed before disconnecting the device:

$ sync

The number of records, amount copied, and rates will all vary depending on the device and the live image you chose.

Burning to a CD or DVD

Any disk burning application should be capable of writing the .iso file to a CD or DVD. The following free software applications are available (cross-platform support may vary):

It should be noted that, with a CD or DVD, live sessions will be less responsive than with a USB stick or hard drive.

Partitioning Notes

Partitioning for a modern Linux distribution is generally very simple, however the introduction of GPT and UEFI booting does bring new complexity to the process. When creating your new partition table you will need a partition for the root filesystem, along with a swap partition and possibly another partition or two to facilitate booting, if required.

Note that if the disk has already been initialized, the top of the cfdisk screen will show the partition layout already present: Label: dos for the MBR scheme, Label: gpt for the GPT scheme. If you just want to erase the partition table before starting the installer, use wipefs(8). Otherwise, you can run cfdisk(8) manually with the -z option to start with an uninitialized disk layout; cfdisk will prompt you for the label type before continuing to the main screen.

The following sections will detail the options for partition configuration.

BIOS system notes

It is recommended that you create an MBR partition table if you are using a BIOS boot system. This will limit the number of partitions you create to four.

It is possible to use a GPT partition table on a BIOS system, but GRUB will require a special partition to boot properly. This partition should be at the beginning of your disk and have a size of 1MB, with type BIOS boot (GUID 21686148-6449-6E6F-744E-656564454649). Don't create any filesystem in it. GRUB should then install itself successfully.

UEFI system notes

UEFI users are recommended to create a GPT partition table. UEFI booting with GRUB also requires a special partition of the type EFI System with a vfat filesystem mounted at /boot/efi. A reasonable size for this partition could be between 200MB and 1GB. With this partition setup during the live image installation, the installer should successfully set up the bootloader automatically.

Swap partitions

A swap partition is not strictly required, but recommended for systems with low RAM. If you want to use hibernation, you will need a swap partition. The following table has recommendations for swap partition size.

System RAMRecommended swap spaceSwap space if using hibernation
< 2GB2x the amount of RAM3x the amount of RAM
2-8GBEqual to amount of RAM2x the amount of RAM
8-64GBAt least 4GB1.5x the amount of RAM
64GBAt least 4GBHibernation not recommended

Boot partition (optional)

On most modern systems, a separate /boot partition is no longer necessary to boot properly. If you choose to use one, note that Void does not remove old kernels after updates by default and also that the kernel tends to increase in size with each new version, so plan accordingly (e.g. /boot with one Linux 5.x x86_64 kernel and GRUB occupies about 60MB).

Other partitions

It is fine to install your system with only a large root partition, but you may create other partitions if you want. One helpful addition could be a separate partition for your /home directory. This way if you need to reinstall Void (or another distribution) you can save the data and configuration files in your home directory for your new system.

Installation Guide

Once you have downloaded a Void image to install and prepared your install media, you are ready to install Void Linux.

Before you begin installation, you should determine whether your machine boots using BIOS or UEFI. This will affect how you plan partitions. See Partitioning Notes for more detail.

The following features are not supported by the installer script:

Booting

Boot your machine from the install media you created. If you have enough RAM, there is an option on the boot screen to load the entire image into ram, which will take some time but speed up the rest of the install process.

Once the live image has booted, log in as root with password voidlinux and run:

# void-installer

The following sections will detail each screen of the installer.

Keyboard

Select the keymap for your keyboard; standard "qwerty" keyboards will generally use the "us" keymap.

Network

Select your primary network interface. If you do not choose to use DHCP, you will be prompted to provide an IP address, gateway, and DNS servers.

If you choose a wireless network interface, you will be prompted to provide the SSID, encryption type (wpa or wep), and password. If void-installer fails to connect to your network, you may need to exit the installer and configure it manually using wpa_supplicant and dhcpcd before continuing.

Source

To install packages provided on the install image, select Local. Otherwise, you may select Network to download the latest packages from the Void repository.

Warning: If you are installing the desktop environment from the xfce image, you MUST choose Local for the source!

Hostname

Select a hostname for your computer (that is all lowercase, with no spaces.)

Locale

Select your default locale settings. This option is for glibc only, as musl does not currently support locales.

Timezone

Select your timezone based on standard timezone options.

Root password

Enter and confirm your root password for the new installation. The password will not be shown on screen.

User account

Choose a login (default void) and a descriptive name for that login. Then enter and confirm the password for the new user. You will then be prompted to verify the groups for this new user. They are added to the wheel group by default and will have sudo access. Default groups and their descriptions are listed here.

Login names have some restrictions, as described in useradd(8).

Bootloader

Select the disk to install a bootloader on when Void is installed. You may select none to skip this step and install a bootloader manually after completing the installation process. If installing a bootloader, you will also be asked whether or not you want a graphical terminal for the GRUB menu.

Partition

Next, you will need to partition your disks. Void does not provide a preset partition scheme, so you will need to create your partitions manually with cfdisk(8). You will be prompted with a list of disks. Select the disk you want to partition and the installer will launch cfdisk for that disk. Remember you must write the partition table to the drive before you exit the partition editor.

If using UEFI, it is recommended you select GPT for the partition table and create a partition (typically between 200MB-1GB) of type EFI System, which will be mounted at /boot/efi.

If using BIOS, it is recommended you select MBR for the partition table. Advanced users may use GPT but will need to create a special BIOS partition for GRUB to boot.

See the Partitioning Notes for more details about partitioning your disk.

Filesystems

Create the filesystems for each partition you have created. For each partition you will be prompted to choose a filesystem type, whether you want to create a new filesystem on the partition, and a mount point, if applicable. When you are finished, select Done to return to the main menu.

If using UEFI, create a vfat filesystem and mount it at /boot/efi.

Review settings

It is a good idea to review your settings before proceeding. Use the right arrow key to select the settings button and hit <enter>. All your selections will be shown for review.

Install

Selecting Install from the menu will start the installer. The installer will create all the filesystems selected, and install the base system packages. It will then generate an initramfs and install a GRUB2 bootloader to the bootable partition.

These steps will all run automatically, and after the installation is completed successfully, you can reboot into your new Void Linux install!

Post installation

After booting into your Void installation for the first time, perform a system update.

Advanced Installation Guides

This section contains guides for more specific or complex use-cases.

Section Contents

Installation via chroot (x86/x86_64/aarch64)

This guide details the process of manually installing Void via a chroot on an x86, x86_64 or aarch64 architecture. It is assumed that you have a familiarity with Linux, but not necessarily with installing a Linux system via a chroot. This guide can be used to create a "typical" setup, using a single partition on a single SATA/IDE/USB disk. Each step may be modified to create less typical setups, such as full disk encryption.

Void provides two options for bootstrapping the new installation. The XBPS method uses the XBPS Package Manager running on a host operating system to install the base system. The ROOTFS method installs the base system by unpacking a ROOTFS tarball.

The XBPS method requires that the host operating system have XBPS installed. This may be an existing installation of Void, an official live image, or any Linux installation running a statically linked XBPS.

The ROOTFS method requires only a host operating system that can enter a Linux chroot and that has both tar(1) and xz(1) installed. This method may be preferable if you wish to install Void using a different Linux distribution.

Prepare Filesystems

Partition your disks and format them using mke2fs(8), mkfs.xfs(8), mkfs.btrfs(8) or whatever tools are necessary for your filesystem(s) of choice.

mkfs.vfat(8) is also available to create FAT32 partitions. However, due to restrictions associated with FAT filesystems, it should only be used when no other filesystem is suitable (such as for the EFI System Partition).

cfdisk(8) and fdisk(8) are available on the live images for partitioning, but you may wish to use gdisk(8) (from the package gptfdisk) or parted(8) instead.

For a UEFI booting system, make sure to create an EFI System Partition (ESP). The ESP should have the partition type "EFI System" (code EF00) and be formatted as FAT32 using mkfs.vfat(8).

If you're unsure what partitions to create, create a 1GB partition of type "EFI System" (code EF00), then create a second partition of type "Linux Filesystem" (code 8300) using the remainder of the drive.

Format these partitions as FAT32 and ext4, respectively:

# mkfs.vfat /dev/sda1
# mkfs.ext4 /dev/sda2

Create a New Root and Mount Filesystems

This guide will assume the new root filesystem is mounted on /mnt. You may wish to mount it elsewhere.

If using UEFI, mount the EFI System Partition as /mnt/boot/efi.

For example, if /dev/sda2 is to be mounted as / and dev/sda1 is the EFI System Partition:

# mount /dev/sda2 /mnt/
# mkdir -p /mnt/boot/efi/
# mount /dev/sda1 /mnt/boot/efi/

Initialize swap space, if desired, using mkswap(8).

Base Installation

Follow only one of the two following subsections.

If on aarch64, it will be necessary to install a kernel package in addition to base-system. For example, linux is a kernel package that points to the latest stable kernel packaged by Void.

The XBPS Method

Select a mirror and use the appropriate URL for the type of system you wish to install. For simplicity, save this URL to a shell variable. A glibc installation, for example, would use:

# REPO=https://repo-default.voidlinux.org/current

XBPS also needs to know what architecture is being installed. Available options are x86_64, x86_64-musl, i686 for PC architecture computers and aarch64. For example:

# ARCH=x86_64

This architecture must be compatible with your current operating system, but does not need to be the same. If your host is running an x86_64 operating system, any of the three architectures can be installed (whether the host is musl or glibc), but an i686 host can only install i686 distributions.

Copy the RSA keys from the installation medium to the target root directory:

# mkdir -p /mnt/var/db/xbps/keys
# cp /var/db/xbps/keys/* /mnt/var/db/xbps/keys/

Use xbps-install(1) to bootstrap the installation by installing the base-system metapackage:

# XBPS_ARCH=$ARCH xbps-install -S -r /mnt -R "$REPO" base-system

The ROOTFS Method

Download a ROOTFS tarball matching your architecture.

Unpack the tarball into the newly configured filesystems:

# tar xvf void-<...>-ROOTFS.tar.xz -C /mnt

Configuration

With the exception of the section "Install base-system (ROOTFS method only)", the remainder of this guide is common to both the XBPS and ROOTFS installation methods.

Entering the Chroot

xchroot(1) (from xtools) can be used to set up and enter the chroot. Alternatively, this can be done manually.

# xchroot /mnt /bin/bash

Install base-system (ROOTFS method only)

ROOTFS images generally contain out of date software, due to being a snapshot of the time when they were built, and do not come with a complete base-system. Update the package manager and install base-system:

[xchroot /mnt] # xbps-install -Su xbps
[xchroot /mnt] # xbps-install -u
[xchroot /mnt] # xbps-install base-system
[xchroot /mnt] # xbps-remove base-container-full

Installation Configuration

Specify the hostname in /etc/hostname. Go through the options in /etc/rc.conf. If installing a glibc distribution, edit /etc/default/libc-locales, uncommenting desired locales.

nvi(1) is available in the chroot, but you may wish to install your preferred text editor at this time.

For glibc builds, generate locale files with:

[xchroot /mnt] # xbps-reconfigure -f glibc-locales

Set a Root Password

Configure at least one super user account. Other user accounts can be configured later, but there should either be a root password, or a new user account with sudo(8) privileges.

To set a root password, run:

[xchroot /mnt] # passwd

Configure fstab

The fstab(5) file can be automatically generated from currently mounted filesystems by copying the file /proc/mounts:

[xchroot /mnt] # cp /proc/mounts /etc/fstab

Remove lines in /etc/fstab that refer to proc, sys, devtmpfs and pts.

Replace references to /dev/sdXX, /dev/nvmeXnYpZ, etc. with their respective UUID, which can be found by running blkid(8). Referring to filesystems by their UUID guarantees they will be found even if they are assigned a different name at a later time. In some situations, such as booting from USB, this is absolutely essential. In other situations, disks will always have the same name unless drives are physically added or removed. Therefore, this step may not be strictly necessary, but is almost always recommended.

Change the last zero of the entry for / to 1, and the last zero of every other line to 2. These values configure the behaviour of fsck(8).

For example, the partition scheme used throughout previous examples yields the following fstab:

/dev/sda1       /boot/efi   vfat    rw,relatime,[...]       0 0
/dev/sda2       /           ext4    rw,relatime             0 0

The information from blkid results in the following /etc/fstab:

UUID=6914[...]  /boot/efi   vfat    rw,relatime,[...]       0 2
UUID=dc1b[...]  /           ext4    rw,relatime             0 1

Note: The output of /proc/mounts will have a single space between each field. The columns are aligned here for readability.

Add an entry to mount /tmp in RAM:

tmpfs           /tmp        tmpfs   defaults,nosuid,nodev   0 0

If using swap space, add an entry for any swap partitions:

UUID=1cb4[...]  swap        swap    rw,noatime,discard      0 0

Enable services

Services can be enabled after booting the new system, but you may need to enable some of them (e.g.: dhcpcd, sshd) now in order to access it.

Installing GRUB

Use grub-install to install GRUB onto your boot disk.

On a BIOS computer, install the package grub, then run grub-install /dev/sdX, where /dev/sdX is the drive (not partition) that you wish to install GRUB to. For example:

[xchroot /mnt] # xbps-install grub
[xchroot /mnt] # grub-install /dev/sda

On a UEFI computer, install either grub-x86_64-efi, grub-i386-efi or grub-arm64-efi, depending on your architecture, then run grub-install, optionally specifying a bootloader label (this label may be used by your computer's firmware when manually selecting a boot device):

[xchroot /mnt] # xbps-install grub-x86_64-efi
[xchroot /mnt] # grub-install --target=x86_64-efi --efi-directory=/boot/efi --bootloader-id="Void"

Troubleshooting GRUB installation

It may be necessary to mount the efivarfs filesystem.

[xchroot /mnt] # mount -t efivarfs none /sys/firmware/efi/efivars

If EFI variables are still not available, add the option --no-nvram to the grub-install command.

Installing on removable media or non-compliant UEFI systems

Unfortunately, not all systems have a fully standards compliant UEFI implementation. In some cases, it is necessary to "trick" the firmware into booting by using the default fallback location for the bootloader instead of a custom one. In that case, or if installing onto a removable disk (such as USB), add the option --removable to the grub-install command.

Alternatively, use mkdir(1) to create the /boot/efi/EFI/boot directory and copy the installed GRUB executable, usually located in /boot/efi/EFI/Void/grubx64.efi (its location can be found using efibootmgr(8)), into the new folder:

[xchroot /mnt] # mkdir -p /boot/efi/EFI/boot
[xchroot /mnt] # cp /boot/efi/EFI/Void/grubx64.efi /boot/efi/EFI/boot/bootx64.efi

Finalization

Use xbps-reconfigure(1) to ensure all installed packages are configured properly:

[xchroot /mnt] # xbps-reconfigure -fa

This will make dracut(8) generate an initramfs, and will make GRUB generate a working configuration.

At this point, the installation is complete. Exit the chroot and reboot your computer:

[xchroot /mnt] # exit
# umount -R /mnt
# shutdown -r now

After booting into your Void installation for the first time, perform a system update.

Full Disk Encryption

Warning: Your drive's block device and other information may be different, so make sure it is correct.

Partitioning

Boot a live image and login.

Create a single physical partition on the disk using cfdisk, marking it as bootable. For an MBR system, the partition layout should look like the following.

# fdisk -l /dev/sda
Disk /dev/sda: 48 GiB, 51539607552 bytes, 100663296 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: dos
Disk identifier: 0x4d532059

Device     Boot Start       End   Sectors Size Id Type
/dev/sda1  *     2048 100663295 100661248  48G 83 Linux

UEFI systems will need the disk to have a GPT disklabel and an EFI system partition. The required size for this may vary depending on needs, but 100M should be enough for most cases. For an EFI system, the partition layout should look like the following.

# fdisk -l /dev/sda
Disk /dev/sda: 48 GiB, 51539607552 bytes, 100663296 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: EE4F2A1A-8E7F-48CA-B3D0-BD7A01F6D8A0

Device      Start       End   Sectors  Size Type
/dev/sda1    2048    264191    262144  128M EFI System
/dev/sda2  264192 100663262 100399071 47.9G Linux filesystem

Encrypted volume configuration

Cryptsetup defaults to LUKS2, yet GRUB releases before 2.06 only had support for LUKS1.

LUKS2 is only partially supported by GRUB; specifically, only the PBKDF2 key derivation function is implemented, which is not the default KDF used with LUKS2, that being Argon2i (GRUB Bug 59409). LUKS encrypted partitions using Argon2i (as well as the other KDF) can not be decrypted. For that reason, this guide only recommends LUKS1 be used.

Keep in mind the encrypted volume will be /dev/sda2 on EFI systems, since /dev/sda1 is taken up by the EFI partition.

# cryptsetup luksFormat --type luks1 /dev/sda1

WARNING!
========
This will overwrite data on /dev/sda1 irrevocably.

Are you sure? (Type uppercase yes): YES
Enter passphrase:
Verify passphrase:

Once the volume is created, it needs to be opened. Replace voidvm with an appropriate name. Again, this will be /dev/sda2 on EFI systems.

# cryptsetup luksOpen /dev/sda1 voidvm
Enter passphrase for /dev/sda1:

Once the LUKS container is opened, create the LVM volume group using that partition.

# vgcreate voidvm /dev/mapper/voidvm
  Volume group "voidvm" successfully created

There should now be an empty volume group named voidvm.

Next, logical volumes need to be created for the volume group. For this example, I chose 10G for /, 2G for swap, and will assign the rest to /home.

# lvcreate --name root -L 10G voidvm
  Logical volume "root" created.
# lvcreate --name swap -L 2G voidvm
  Logical volume "swap" created.
# lvcreate --name home -l 100%FREE voidvm
  Logical volume "home" created.

Next, create the filesystems. The example below uses XFS as a personal preference of the author. Any filesystem supported by GRUB will work.

# mkfs.xfs -L root /dev/voidvm/root
meta-data=/dev/voidvm/root       isize=512    agcount=4, agsize=655360 blks
...
# mkfs.xfs -L home /dev/voidvm/home
meta-data=/dev/voidvm/home       isize=512    agcount=4, agsize=2359040 blks
...
# mkswap /dev/voidvm/swap
Setting up swapspace version 1, size = 2 GiB (2147479552 bytes)

System installation

Next, setup the chroot and install the base system.

# mount /dev/voidvm/root /mnt
# mkdir -p /mnt/home
# mount /dev/voidvm/home /mnt/home

On a UEFI system, the EFI system partition also needs to be mounted.

# mkfs.vfat /dev/sda1
# mkdir -p /mnt/boot/efi
# mount /dev/sda1 /mnt/boot/efi

Copy the RSA keys from the installation medium to the target root directory:

# mkdir -p /mnt/var/db/xbps/keys
# cp /var/db/xbps/keys/* /mnt/var/db/xbps/keys/

Before we enter the chroot to finish up configuration, we do the actual install. Do not forget to use the appropriate repository URL for the type of system you wish to install.

# xbps-install -Sy -R https://repo-default.voidlinux.org/current -r /mnt base-system lvm2 cryptsetup grub
[*] Updating `https://repo-default.voidlinux.org/current/x86_64-repodata' ...
x86_64-repodata: 1661KB [avg rate: 2257KB/s]
130 packages will be downloaded:
...

UEFI systems will have a slightly different package selection. The installation command for a UEFI system will be as follows.

# xbps-install -Sy -R https://repo-default.voidlinux.org/current -r /mnt base-system cryptsetup grub-x86_64-efi lvm2

When it's done, we can enter the chroot with xchroot(1) (from xtools) and finish up the configuration. Alternatively, entering the chroot can be done manually.

# xchroot /mnt
[xchroot /mnt] # chown root:root /
[xchroot /mnt] # chmod 755 /
[xchroot /mnt] # passwd root
[xchroot /mnt] # echo voidvm > /etc/hostname

and, for glibc systems only:

[xchroot /mnt] # echo "LANG=en_US.UTF-8" > /etc/locale.conf
[xchroot /mnt] # echo "en_US.UTF-8 UTF-8" >> /etc/default/libc-locales
[xchroot /mnt] # xbps-reconfigure -f glibc-locales

Filesystem configuration

The next step is editing /etc/fstab, which will depend on how you configured and named your filesystems. For this example, the file should look like this:

# <file system>   <dir> <type>  <options>             <dump>  <pass>
tmpfs             /tmp  tmpfs   defaults,nosuid,nodev 0       0
/dev/voidvm/root  /     xfs     defaults              0       0
/dev/voidvm/home  /home xfs     defaults              0       0
/dev/voidvm/swap  swap  swap    defaults              0       0

UEFI systems will also have an entry for the EFI system partition.

/dev/sda1	/boot/efi	vfat	defaults	0	0

GRUB configuration

Next, configure GRUB to be able to unlock the filesystem. Add the following line to /etc/default/grub:

GRUB_ENABLE_CRYPTODISK=y

Next, the kernel needs to be configured to find the encrypted device. First, find the UUID of the device.

[xchroot /mnt] # blkid -o value -s UUID /dev/sda1
135f3c06-26a0-437f-a05e-287b036440a4

Edit the GRUB_CMDLINE_LINUX_DEFAULT= line in /etc/default/grub and add rd.lvm.vg=voidvm rd.luks.uuid=<UUID> to it. Make sure the UUID matches the one for the sda1 device found in the output of the blkid(8) command above. This will be /dev/sda2 on EFI systems.

LUKS key setup

And now to avoid having to enter the password twice on boot, a key will be configured to automatically unlock the encrypted volume on boot. First, generate a random key.

[xchroot /mnt] # dd bs=1 count=64 if=/dev/urandom of=/boot/volume.key
64+0 records in
64+0 records out
64 bytes copied, 0.000662757 s, 96.6 kB/s

Next, add the key to the encrypted volume.

[xchroot /mnt] # cryptsetup luksAddKey /dev/sda1 /boot/volume.key
Enter any existing passphrase:

Change the permissions to protect the generated key.

[xchroot /mnt] # chmod 000 /boot/volume.key
[xchroot /mnt] # chmod -R g-rwx,o-rwx /boot

This keyfile also needs to be added to /etc/crypttab. Again, this will be /dev/sda2 on EFI systems.

voidvm   /dev/sda1   /boot/volume.key   luks

And then the keyfile and crypttab need to be included in the initramfs. Create a new file at /etc/dracut.conf.d/10-crypt.conf with the following line:

install_items+=" /boot/volume.key /etc/crypttab "

Complete system installation

Next, install the boot loader to the disk.

[xchroot /mnt] # grub-install /dev/sda

Ensure an initramfs is generated:

[xchroot /mnt] # xbps-reconfigure -fa

Exit the chroot, unmount the filesystems, and reboot the system.

[xchroot /mnt] # exit
# umount -R /mnt
# reboot

Installing Void on a ZFS Root

Because the Void installer does not support ZFS, it is necessary to install via chroot. Aside from a few caveats regarding bootloader and initramfs support, installing Void on a ZFS root filesystem is not significantly different from any other advanced installation. ZFSBootMenu is a bootloader designed from the ground up to support booting Linux distributions directly from a ZFS pool. However, it is also possible to use traditional bootloaders with a ZFS root.

ZFSBootMenu

Although it will boot (and can be run atop) a wide variety of distributions, ZFSBootMenu officially considers Void a first-class distribution. ZFSBootMenu supports native ZFS encryption, offers a convenient recovery environment that can be used to clone prior snapshots or perform advanced manipulation in a pre-boot environment, and will support booting from any pool that is importable by modern ZFS drivers. The ZFSBootMenu documentation offers, among other content, several step-by-step guides for installing a Void system from scratch. The UEFI guide describes the procedure of bootstrapping a Void system for modern systems. For legacy BIOS systems, the syslinux guide provides comparable instructions.

Traditional bootloaders

For those that wish to forego ZFSBootMenu, it is possible to bootstrap a Void system with another bootloader. To avoid unnecessary complexity, systems that use bootloaders other than ZFSBootMenu should plan to use a separate /boot that is located on an ext4 or xfs filesystem.

Installation media

Installing Void to a ZFS root requires an installation medium with ZFS drivers. It is possible to build a custom image from the official void-mklive repository by providing the command-line option -p zfs to the mklive.sh script. However, for x86_64 systems, it may be more convenient to fetch a pre-built hrmpf image. These images, maintained by a Void team member, are extensions of the standard Void live images that include pre-compiled ZFS modules in addition to other useful tools.

Partition disks

After booting a live image with ZFS support, partition your disks. The considerations in the partitioning guide apply to ZFS installations as well, except that

  • The boot partition should be considered necessary unless you intend to use gummiboot, which expects that your EFI system partition will be mounted at /boot. (This alternative configuration will not be discussed here.)
  • Aside from any EFI system partition, GRUB BIOS boot partition, swap or boot partitions, the remainder of the disk should typically be a single partition with type code BF00 that will be dedicated to a single ZFS pool. There is no benefit to creating separate ZFS pools on a single disk.

As needed, format the EFI system partition using mkfs.vfat(8) and the the boot partition using mke2fs(8) or mkfs.xfs(8). Initialize any swap space using mkswap(8).

It is possible to put Linux swap space on a ZFS zvol, although there may be a risk of deadlocking the kernel when under high memory pressure. This guide takes no position on the matter of swap space on a zvol. However, if you wish to use suspension-to-disk (hibernation), note that the kernel is not capable of resuming from memory images stored on a zvol. You will need a dedicated swap partition to use hibernation. Apart from this caveat, there are no special considerations required to resume a suspended image when using a ZFS root.

Create a ZFS pool

Create a ZFS pool on the partition created for it using zpool(8). For example, to create a pool on /dev/disk/by-id/wwn-0x5000c500deadbeef-part3:

# zpool create -f -o ashift=12 \
    -O compression=lz4 \
    -O acltype=posixacl \
    -O xattr=sa \
    -O relatime=on \
    -o autotrim=on \
    -m none zroot /dev/disk/by-id/wwn-0x5000c500deadbeef-part3

Adjust the pool (-o) and filesystem (-O) options as desired, and replace the partition identifier wwn-0x5000c500deadbeef-part3 with that of the actual partition to be used.

When adding disks or partitions to ZFS pools, it is generally advisable to refer to them by the symbolic links created in /dev/disk/by-id or (on UEFI systems) /dev/disk/by-partuuid so that ZFS will identify the right partitions even if disk naming should change at some point. Using traditional device nodes like /dev/sda3 may cause intermittent import failures.

Next, export and re-import the pool with a temporary, alternate root path:

# zpool export zroot
# zpool import -N -R /mnt zroot

Create initial filesystems

The filesystem layout on your ZFS pool is flexible. However, it is customary to put operating system root filesystems ("boot environments") under a ROOT parent:

# zfs create -o mountpoint=none zroot/ROOT
# zfs create -o mountpoint=/ -o canmount=noauto zroot/ROOT/void

Setting canmount=noauto on filesystems with mountpoint=/ is useful because it permits the creation of multiple boot environments (which may be clones of a common Void installation or contain completely separate distributions) without fear that ZFS auto-mounting will attempt to mount one over another.

To separate user data from the operating system, create a filesystem to store home directories:

# zfs create -o mountpoint=/home zroot/home

Other filesystems may be created as desired.

Mount the ZFS hierarchy

All ZFS filesystems should be mounted under the /mnt alternate root established by the earlier re-import. Mount the manual-only root filesystem before allowing ZFS to automatically mount everything else:

# zfs mount zroot/ROOT/void
# zfs mount -a

At this point, the entire ZFS hierarchy should be mounted and ready for installation. To improve boot-time import speed, it is useful to record the current pool configuration in a cache file that Void will use to avoid walking the entire device hierarchy to identify importable pools:

# mkdir -p /mnt/etc/zfs
# zpool set cachefile=/mnt/etc/zfs/zpool.cache zroot

Mount non-ZFS filesystems at the appropriate places. For example, if /dev/sda2 holds an ext4 filesystem that should be mounted at /boot and /dev/sda1 is the EFI system partition:

# mkdir -p /mnt/boot
# mount /dev/sda2 /mnt/boot
# mkdir -p /mnt/boot/efi
# mount /dev/sda1 /mnnt/boot/efi

Installation

At this point, ordinary installation can proceed from the "Base Installation" section. of the standard chroot installation guide. However, before following the "Finalization" instructions, make sure that the zfs package has been installed and dracut is configured to identify a ZFS root filesystem:

[xchroot /mnt] # mkdir -p /etc/dracut.conf.d
[xchroot /mnt] # cat > /etc/dracut.conf.d/zol.conf <<EOF
nofsck="yes"
add_dracutmodules+=" zfs "
omit_dracutmodules+=" btrfs resume "
EOF
[xchroot /mnt] # xbps-install zfs

Finally, follow the "Finalization" instructions and reboot into your new system.

ARM Devices

Void Linux provides packages and images for several ARM devices. Installing Void on such devices can be done in several ways:

  • Pre-built images: images that can be flashed directly onto an SD card or other storage medium, but which give you a limited partition layout, and require manual expansion if you wish to increase the size of the partitions;
  • Tarball installation: PLATFORMFS and ROOTFS tarballs that can be extracted to a previously prepared partition scheme; and
  • Chroot installation: follows most of the steps outlined in the chroot guide.

This guide also outlines configuration steps that are mostly specific to such devices.

Since most of the commands in this guide will be run on external storage, it is important to run sync(1) before removing the device.

Installation

If you are installing Void Linux on one of the ARM devices covered in the "Supported platforms" page, make sure to read its section thoroughly.

Pre-built images

The pre-built images provided are prepared for 2GB SD cards. After downloading and verifying an image, it can be uncompressed with unxz(1) and written to the relevant media with cat(1), pv(1), or dd(1). For example, to flash it onto an SD card located at /dev/mmcblk0:

$ unxz -k <image>.img.xz
# dd if=<image>.img of=/dev/mmcblk0 bs=4M status=progress

After flashing, the root partition can optionally be expanded to fit the storage device with cfdisk(8), fdisk(8), or another partitioning tool, and the filesystem can be resized to fit the expanded partition with resize2fs(8).

Custom partition layout

Customizing an installation - for example, with a custom partition layout - requires a more involved process. Two available options are:

To prepare the storage for these installation methods, it is necessary to partition the storage medium and then mount the partitions at the correct mount points.

The usual partitioning scheme for ARM devices requires at least two partitions, on a drive formatted with an MS-DOS partition table:

  • one formatted as FAT32 with partition type 0c, which will be mounted on /boot;
  • one that can be formatted as any file system that Linux can boot from, such as ext4, which will be mounted on /. If you're using an SD card, you can create the ext4 file system with the ^has_journal option - this disables journaling, which might increase the drive's life, at the cost of a higher chance of data loss.

There are a variety of tools available for partitioning, e.g. cfdisk(8).

To access the newly created file systems, it is necessary to mount them. This guide will assume that the second partition will be mounted on /mnt, but you may mount it elsewhere. To mount these filesystems, you can use the commands below, replacing the device names with the appropriate ones for your setup:

# mount /dev/mmcblk0p2 /mnt
# mkdir /mnt/boot
# mount /dev/mmcblk0p1 /mnt/boot

Tarball installation

First, download and verify a PLATFORMFS or ROOTFS tarball for your desired platform and prepare your storage medium. Then, unpack the tarball onto the file system using tar(1):

# tar xvfp <image>.tar.xz -C /mnt

Chroot installation

It is also possible to perform a chroot installation using the appropriate architecture and base packages, some of which are listed in the "Supported Platforms" section. Make sure to prepare your storage medium properly for the device.

If doing this from a computer with an incompatible archtecture (such as x86_64), install binfmt-support, enable the binfmt-support service, and install the relevant QEMU user emulator (like qemu-user-aarch64 for aarch64 or qemu-user-arm for 32-bit ARM) before installing. If binfmt-support was installed after the QEMU user emulator, use xbps-reconfigure -f qemu-user-<arch> to enable the relevant binfmts.

Configuration

Some additional configuration steps need to be followed to guarantee a working system. Configuring a graphical session should work as normal.

Logging in

For the pre-built images and tarball installations, the root user password is voidlinux.

fstab

The /boot partition should be added to /etc/fstab, with an entry similar to the one below. It is possible to boot without that entry, but updating the kernel package in that situation can lead to breakage, such as being unable to find kernel modules, which are essential for functionality such as wireless connectivity. If you aren't using an SD card, replace /dev/mmcblk0p1 with the appropriate device path.

/dev/mmcblk0p1 /boot vfat defaults 0 0

System time

Several of the ARM devices supported by Void Linux don't have battery powered real time clocks (RTCs), which means they won't keep track of time once powered off. This issue can present itself as HTTPS errors when browsing the Web or using the package manager. It is possible to set the time manually using the date(1) utility. In order to fix this issue for subsequent boots, install and enable an NTP client. Furthermore, it is possible to install the fake-hwclock package, which provides the fake-hwclock service. fake-hwclock(8) periodically stores the current time in a configuration file and restores it at boot, leading to a better initial approximation of the current time, even without a network connection.

Warning: Images from before 2020-03-16 might have an issue where the installation of the chrony package, the default NTP daemon, is incomplete, and the system will be missing the chrony user. This can be checked in the output of the getent(1) command, which will be empty if it doesn't exist:

$ getent group chrony
chrony:x:997

In order to fix this, it is necessary to reconfigure the chrony package using xbps-reconfigure(1).

Graphical session

The xf86-video-fbturbo package ships a modified version of the DDX Xorg driver found in the xf86-video-fbdev package, which is optimized for ARM devices. This can be used for devices which lack more specific drivers.

Supported Platforms

Raspberry Pi

The rpi-kernel packages for all Raspberry Pi variants are built from the Raspberry Pi Foundation's kernel tree, which should enable all special functionality that isn't available with mainline kernels. The RPi kernel packages also have their own header packages, rpi-kernel-headers. These packages should be installed if you want to use any DKMS packages. Void ships rpi-base meta-packages that install the relevant rpi-kernel and rpi-firmware packages. Together, these packages enable Wi-Fi and Bluetooth functionality.

The command line parameters passed to the kernel are in the /boot/cmdline.txt file. Some of the relevant parameters are documented in the official documentation.

Supported Models

ModelArchitecture
1 A, 1 B, 1 A+, 1 B+, Zero, Zero W, Zero WHarmv6l
2 Barmv7l
3 B, 3 A+, 3 B+, Zero 2W, 4 B, 400, CM4, 5aarch64

It is possible to run the armv7l images on an RPi 3, as the RPi 3's CPU supports both the Armv8 and Armv7 instruction sets. The difference between these images is that the armv7l image provides a 32-bit system while the aarch64 image provides a 64-bit system.

Raspberry Pi 5 Kernel

The rpi5-kernel and rpi5-kernel-headers packages provide a kernel and headers optimized for the Raspberry Pi 5 with 16KB pages. To switch from the generic rpi-kernel, install rpi5-kernel. This will remove rpi-kernel and replace it with rpi5-kernel.

Note: not all software is compatible with kernels that have larger page-sizes. View any known issues and report any compatibility problems found in the tracking issue.

Enabling hardware RNG device

By default, the HWRNG device is not used by the system, which may result in the random devices taking long to seed on boot. This can be annoying if you want to start sshd and expect to be able to connect immediately.

In order to fix this, install the rng-tools package and enable the rngd service, which uses the /dev/hwrng device to seed /dev/random.

Graphical session

The mesa-dri package contains drivers for all the Raspberry Pi variants, and can be used with the modesetting Xorg driver or Wayland.

Hardware

More configuration information can be found in the Raspberry Pi Foundation's official documentation. The raspi-config utility isn't available for Void Linux, so editing the /boot/config.txt file is usually required.

Audio

To enable the soundchip, add dtparam=audio=on to /boot/config.txt.

Serial

To enable serial console logins, enable the agetty-ttyAMA0 service. See securetty(5) for interfaces that allow root login. For configuration of the serial port at startup, refer to the kernel command line in /boot/cmdline.txt - in particular, the console=ttyAMA0,115200 parameter.

I2C

To enable I2C, add device_tree_param=i2c_arm=on to /boot/config.txt, and bcm2708.vc_i2c_override=1 to /boot/cmdline.txt. Then create a modules-load(8) .conf file with the following content:

i2c-dev

Finally, install the i2c-tools package and use i2cdetect(8) to verify your configuration. It should show:

$ i2cdetect -l
i2c-1i2c          bcm2835 I2C adapter                 I2C adapter

Memory cgroup

The kernel from the rpi-kernel package disables the memory cgroup by default.

This breaks workloads which use containers. Therefore, if you want to use containers on your Raspberry Pi, you need to enable memory cgroups by adding cgroup_enable=memory to /boot/cmdline.txt.

musl

musl is a libc implementation which strives to be lightweight, fast, simple, and correct.

Void officially supports musl by using it in its codebase for all target platforms (although binary packages are not available for i686). Additionally, all compatible packages in our official repositories are available with musl-linked binaries in addition to their glibc counterparts.

Currently, there are nonfree and debug sub-repositories for musl, but no multilib sub-repo.

Incompatible software

musl practices very strict and minimal standard compliance. Many commonly used platform-specific extensions are not present. Because of this, it is common for software to need modification to compile and/or function properly. Void developers work to patch such software and hopefully get portability/correctness changes accepted into the upstream projects.

Proprietary software usually supports only glibc systems, though sometimes such applications are available as flatpaks and can be run on a musl system. In particular, the proprietary NVIDIA drivers do not support musl, which should be taken into account when evaluating hardware compatibility.

glibc chroot

Software requiring glibc can be run in a glibc chroot.

Configuration

This section and its subsections provide information about configuring your Void system.

Package Documentation

The most common media for documentation in Void Linux are manual pages.

Many packages contain documentation in other formats, like HTML. This documentation can usually be found in a /usr/share/doc/<package> directory.

More extensive documentation may be split into separate *-doc packages, such as julia-doc. This is often the case for programming languages, databases and big software libraries.

In addition to documentation provided by upstream projects, packages may also contain description of initial setup or usage specific to Void, provided by distribution contributors. It will be located in /usr/share/doc/<package>/README.voidlinux.

Manual Pages

Many Void packages come with manual ('man') pages. The default installation includes the mandoc manpage toolset, via the mdocml package.

The man(1) command can be used to show man pages:

$ man chroot

Every man page belongs to a particular section:

  • 1: User commands (Programs)
  • 2: System calls
  • 3: Library calls
  • 4: Special files (devices)
  • 5: File formats and configuration files
  • 6: Games
  • 7: Overview, conventions, and miscellaneous
  • 8: System management commands

Refer to man-pages(7) for details.

There are some man pages which have the same name, but are used in different contexts, and are thus in a different section. You can specify which one to use by including the section number in the call to man:

$ man 1 printf

man can be configured via man.conf(5).

The mandoc toolset contains apropos(1), which can be used to search for manual pages. apropos uses a database that can be generated and updated with the makewhatis(8) command:

# makewhatis
$ apropos chroot
chroot(1) - run command or interactive shell with special root directory
xbps-uchroot(1) - XBPS utility to chroot and bind mount with Linux namespaces
xbps-uunshare(1) - XBPS utility to chroot and bind mount with Linux user namespaces
chroot(2) - change root directory

The mdocml package provides a cron job to update the database daily, /etc/cron.daily/makewhatis. You will need to install and enable a cron daemon for this functionality to be activated.

Development and POSIX manuals are not installed by default, but are available via the man-pages-devel and man-pages-posix packages.

Localized manual pages

It is also possible to use localized man pages from packages which provide their own as well as those provided by the manpages-* packages. However, this can require some configuration.

With mdocml

If mdocml is being used and the settings should be applied for all users, it is necessary to add the relevant paths to man.conf(5). For example, German speakers would add these two lines to their configuration file:

/usr/share/man/de
/usr/share/man/de.UTF-8

Alternatively, each user can export the MANPATH variable in their environment, as explained in man(1).

Firmware

Void provides a number of firmware packages in the repositories. Some firmware is only available if you have enabled the nonfree repository.

Microcode

Microcode is loaded onto the CPU or GPU at boot by the BIOS, but can be replaced later by the OS itself. An update to microcode can allow a CPU's or GPU's behavior to be modified to work around certain yet to be discovered bugs, without the need to replace the hardware.

Intel

Install the Intel microcode package, intel-ucode. This package is in the nonfree repo, which has to be enabled. After installing this package, it is necessary to regenerate your initramfs. For subsequent updates, the microcode will be added to the initramfs automatically.

AMD

Install the AMD package, linux-firmware-amd, which contains microcode for both AMD CPUs and GPUs. AMD CPUs and GPUs will automatically load the microcode, no further configuration required.

Verification

The /proc/cpuinfo file has some information under microcode that can be used to verify the microcode update.

Removing firmware

By default, linuxX.Y packages and the base-system package install a number of firmware packages. It is not necessary to remove unused firmware packages, but if you wish to do so, you can configure XBPS to ignore those packages, then remove them.

Locales and Translations

For a list of currently enabled locales, run

$ locale -a

Enabling locales

To enable a certain locale, un-comment or add the relevant lines in /etc/default/libc-locales and force-reconfigure the glibc-locales package.

Setting the system locale

Set LANG=xxxx in /etc/locale.conf.

Application locale

Some programs have their translations in a separate package that must be installed in order to use them. You can search for the desired language (e.g. "german" or "portuguese") in the package repositories and install the packages relevant to the applications you use. An especially relevant case is when installing individual packages from the LibreOffice suite, such as libreoffice-writer, which require installing at least one of the libreoffice-i18n-* packages to work properly. This isn't necessary when installing the libreoffice meta-package, since doing so will install the most common translation packages.

Users and Groups

The useradd(8), userdel(8) and usermod(8) commands are used to add, delete and modify users respectively. The passwd(1) command is used to change passwords.

The groupadd(8), groupdel(8) and groupmod(8) commands are used to add, delete and modify groups respectively. The groups(1) command lists all groups a user belongs to.

Default shell

The default shell for a user can be changed with chsh(1):

$ chsh -s <shell> <user_name>

<shell> must be the path to the shell as specified by /etc/shells or the output of chsh -l, which provides a list of installed shells.

sudo

sudo(8) is installed by default, but might not be configured appropriately for your needs. It is only necessary to configure sudo if you wish to use it.

Use visudo(8) as root to edit the sudoers(5) file.

To create a superuser, uncomment the line

#%wheel ALL=(ALL) ALL

and add users to the wheel group.

Default Groups

Void Linux defines a number of groups by default.

GroupDescription
rootComplete access to the system.
binUnused - present for historical reasons.
sysUnused - present for historical reasons.
kmemAbility to read from /dev/mem and /dev/port.
wheelElevated privileges for specific system administration tasks.
ttyAccess to TTY-like devices: /dev/tty*, /dev/pts*, /dev/vcs*.
tapeAccess to tape devices.
daemonSystem daemons that need to write to files on disk.
floppyAccess to floppy drives.
diskRaw access to /dev/sd* and /dev/loop*.
lpAccess to printers.
dialoutAccess to serial ports.
audioAccess to audio devices.
videoAccess to video devices.
utmpAbility to write to /var/run/utmp, /var/log/wtmp and /var/log/btmp.
admUnused - present for historical reasons. This group was traditionally used for system monitoring, such as viewing files in /var/log.
cdromAccess to CD devices.
opticalAccess to DVD/CD-RW devices.
mailUsed by some mail packages, e.g. dma.
storageAccess to removable storage devices.
scannerAbility to access scanners.
networkUsed by some networking-related packages, e.g. connman, NetworkManager, wicd.
kvmAbility to use KVM for virtual machines, e.g. via QEMU.
inputAccess to input devices: /dev/mouse*, /dev/event*.
plugdevAccess to pluggable devices.
nogroupSystem daemons that don't need to own any files.
usbmonAccess to /dev/usbmon*.
usersOrdinary users.
xbuilderTo use xbps-uchroot(1) with xbps-src.

Services and Daemons - runit

Void uses the runit(8) supervision suite to run system services and daemons.

Some advantages of using runit include:

  • a small code base, making it easier to audit for bugs and security issues.
  • each service is given a clean process state, regardless of how the service was started or restarted: it will be started with the same environment, resource limits, open file descriptors, and controlling terminals.
  • a reliable logging facility for services, where the log service stays up as long as the relevant service is running and possibly writing to the log.

If you don't need a program to be running constantly, but would like it to run at regular intervals, you might like to consider using a cron daemon.

Section Contents

Service Directories

Each service managed by runit has an associated service directory.

A service directory requires only one file: an executable named run, which is expected to exec a process in the foreground.

Optionally, a service directory may contain:

  • an executable named check, which will be run to check whether the service is up and available; it's considered available if check exits with 0.
  • an executable named finish, which will be run on shutdown/process stop.
  • a conf file; this can contain environment variables to be sourced and referenced in run.
  • a directory named log; a pipe will be opened from the output of the run process in the service directory to the input of the run process in the log directory.

When a new service is created, a supervise folder will be automatically created on the first run.

Configuring Services

Most services can take configuration options set by a conf file in the service directory. This allows service customization without modifying the service directory provided by the relevant package.

Check the service file for how to pass configuration parameters. A few services have a field like OPTS="--value ..." in their conf file.

To make more complex customizations, you should edit the service.

Editing Services

To edit a service, first copy its service directory to a different directory name. Otherwise, xbps-install(1) can overwrite the service directory. Then, edit the new service file as needed. Finally, the old service should be stopped and disabled, and the new one should be started.

Managing Services

Runsvdirs

A runsvdir is a directory in /etc/runit/runsvdir containing enabled services in the form of symlinks to service directories. On a running system, the current runsvdir is accessible via the /var/service symlink.

The runit-void package comes with two runsvdirs, single and default:

Additional runsvdirs can be created in /etc/runit/runsvdir/.

See runsvdir(8) and runsvchdir(8) for further information.

Booting A Different runsvdir

To boot a runsvdir other than default, the name of the desired runsvdir can be added to the kernel command-line. As an example, adding single to the kernel command line will boot the single runsvdir.

Basic Usage

To start, stop, restart or get the status of a service:

# sv up <services>
# sv down <services>
# sv restart <services>
# sv status <services>

The <services> placeholder can be:

  • Service names (service directory names) inside the /var/service/ directory.
  • The full paths to the services.

For example, the following commands show the status of a specific service and of all enabled services:

# sv status dhcpcd
# sv status /var/service/*

See sv(8) for further information.

Enabling Services

Void Linux provides service directories for most daemons in /etc/sv/.

To enable a service on a booted system, create a symlink to the service directory in /var/service/:

# ln -s /etc/sv/<service> /var/service/

If the system is not currently running, the service can be linked directly into the default runsvdir:

# ln -s /etc/sv/<service> /etc/runit/runsvdir/default/

This will automatically start the service. Once a service is linked it will always start on boot and restart if it stops, unless administratively downed.

To prevent a service from starting at boot while allowing runit to manage it, create a file named down in its service directory:

# touch /etc/sv/<service>/down

The down file mechanism also makes it possible to disable services that are enabled by default, such as the agetty(8) services for ttys 1 to 6. This way, package updates which affect these services (in this case, the runit-void package) won't re-enable them.

Disabling Services

To disable a service, remove the symlink from the running runsvdir:

# rm /var/service/<service>

Or, for example, from the default runsvdir, if either the specific runsvdir, or the system, is not currently running:

# rm /etc/runit/runsvdir/default/<service>

Testing Services

To check if a service is working correctly when started by the service supervisor, run it once before fully enabling it:

# touch /etc/sv/<service>/down
# ln -s /etc/sv/<service> /var/service/
# sv once <service>

If everything works, remove the down file to enable the service.

Per-User Services

Sometimes it can be nice to have user-specific runit services. For example, you might want to open an ssh tunnel as the current user, run a virtual machine, or regularly run daemons on your behalf.

runsvdir

The most basic way to do this is to create a system-level service that runs runsvdir(8) as your user, in order to start and monitor the services in a personal services directory. This does have limitations and downsides, though, as per-user services are started at boot and do not have access to things like the user's graphical session or D-Bus session bus.

For example, you could create a service called /etc/sv/runsvdir-<username> with the following run script, which should be executable:

#!/bin/sh

export USER="<username>"
export HOME="/home/<username>"

groups="$(id -Gn "$USER" | tr ' ' ':')"
svdir="$HOME/service"

exec chpst -u "$USER:$groups" runsvdir "$svdir"

In this example chpst(8) is used to start a new runsvdir(8) process as the specified user. chpst(8) does not read groups on its own, but expects the user to list all required groups separated by a :. The id and tr pipe is used to create a list of all the user's groups in a way chpst(8) understands it. Note that we export $USER and $HOME because some user services may not work without them.

The user can then create new services or symlinks to them in the /home/<username>/service directory. To control the services using the sv(8) command, the user can specify the services by path, or by name if the SVDIR environment variable is set to the user's services directory. This is shown in the following examples:

$ sv status ~/service/*
run: /home/duncan/service/gpg-agent: (pid 901) 33102s
run: /home/duncan/service/ssh-agent: (pid 900) 33102s
$ SVDIR=~/service sv restart gpg-agent
ok: run: gpg-agent: (pid 19818) 0s

It may be convenient to export the SVDIR=~/service variable in your shell profile.

turnstile

Turnstile supports running per-user services that start with the user session using either runit or dinit(8).

If using the runit service backend, user services should be placed in ~/.config/service/.

To ensure that a subset of services are started before login can proceed, these services can be listed in ~/.config/service/turnstile-ready/conf, for example:

core_services="dbus foo"

The turnstile-ready service is created by turnstile on first login.

To give user services access to important environment variables, chpst(8)'s envdir functionality can be used. Inside user services, the convenience variable TURNSTILE_ENV_DIR can be used to refer to this directory.

To make a service aware of these variables, wrap the exec line with chpst -e "$TURNSTILE_ENV_DIR":

exec chpst -e "$TURNSTILE_ENV_DIR" foo

The helper script turnstile-update-runit-env can be used to update variables in this shared envdir:

$ turnstile-update-runit-env DISPLAY XAUTHORITY FOO=bar BAZ=

To run the D-Bus session bus using a turnstile-managed user service:

$ mkdir -p ~/.config/service/dbus
$ ln -s /usr/share/examples/turnstile/dbus.run ~/.config/service/dbus/run

Logging

Syslog

The default installation comes with no syslog daemon. However, there are syslog implementations available in the Void repositories.

Socklog

socklog(8) is a syslog implementation from the author of runit(8). Use socklog if you're not sure which syslog implementation to use. To enable it, install the socklog-void package and enable the socklog-unix and nanoklogd services. Ensure no other syslog daemon is running.

The logs are saved in sub-directories of /var/log/socklog/, and svlogtail can be used to access them conveniently.

The ability to read logs is limited to root and users who are part of the socklog group.

Other syslog daemons

The Void repositories also include packages for rsyslog and metalog.

rc.conf, rc.local and rc.shutdown

The files /etc/rc.conf, /etc/rc.local and /etc/rc.shutdown can be used to configure certain parts of your Void system. rc.conf is often configured by void-installer.

rc.conf

Sourced in runit stages 1 and 3. This file can be used to set variables, including the following:

KEYMAP

Specifies which keymap to use for the Linux console. Available keymaps are listed in /usr/share/kbd/keymaps. For example:

KEYMAP=fr

For further details, refer to loadkeys(1).

HARDWARECLOCK

Specifies whether the hardware clock is set to UTC or local time.

By default this is set to utc. However, Windows sets the hardware clock to local time, so if you are dual-booting with Windows, you need to either configure Windows to use UTC, or set this variable to localtime.

For further details, refer to hwclock(8).

FONT

Specifies which font to use for the Linux console. Available fonts are listed in /usr/share/kbd/consolefonts. For example:

FONT=eurlatgr

For further details, refer to setfont(8).

rc.local

Sourced in runit stage 2. A shell script which can be used to specify configuration to be done prior to login.

rc.shutdown

Sourced in runit stage 3. A shell script which can be used to specify tasks to be done during shutdown.

Cron

cron is a daemon for running programs at regular intervals. The programs and intervals are specified in a crontab file, which can be edited with crontab(1). Running crontab -e as the superuser will edit the system crontab; otherwise, it will edit the crontab for the current user.

By default, a cron daemon is not installed. However, multiple cron implementations are available, including cronie, dcron, fcron and more.

Once you have chosen and installed an implementation, enable the corresponding service. There is also a generic crond service which is maintained by the alternatives system, but there is no real benefit in using it and just makes your setup harder to follow.

As an alternative to the standard cron implementations, you can use snooze(1) together with the snooze-hourly, snooze-daily, snooze-weekly and snooze-monthly services, which are provided by the snooze package for this purpose. Each of these services execute scripts in the respective /etc/cron.* directories.

Solid State Drives

Post installation, you will need to enable TRIM for solid state drives. You can check which devices allow TRIM by running:

$ lsblk --discard

If the DISC-GRAN (discard granularity) and DISC-MAX (discard maximum bytes) columns are non-zero, that means the block device has TRIM support. If your solid state drive partition does not show TRIM support, please verify that you chose a file system with TRIM support (ext4, Btrfs, F2FS, etc.). Note that F2FS requires kernel 4.19 or above to support TRIM.

To run TRIM one-shot, you can run fstrim(8) manually. For example, if your / directory is on an SSD:

# fstrim /

To automate running TRIM, use cron or add the discard option to /etc/fstab.

Periodic TRIM with cron

Add the following lines to /etc/cron.weekly/fstrim:

#!/bin/sh

fstrim /

Finally, make the script executable:

# chmod u+x /etc/cron.weekly/fstrim

Continuous TRIM with fstab discard

You can use either continuous or periodic TRIM, but usage of continuous TRIM is discouraged if you have an SSD that doesn't handle NCQ correctly. Refer to the kernel blacklist.

Edit /etc/fstab and add the discard option to block devices that need TRIM.

For example, if /dev/sda1 was an SSD partition, formatted as ext4, and mounted at /:

/dev/sda1  /           ext4  defaults,discard   0  1

LVM

To enable TRIM for LVM's commands (lvremove, lvreduce, etc.), open /etc/lvm/lvm.conf, uncomment the issue_discards option, and set it to 1:

issue_discards=1

LUKS

Warning: Before enabling discard for your LUKS partition, please be aware of the security implications.

To open an encrypted LUKS device and allow discards to pass through, open the device with the --allow-discards option:

# cryptsetup luksOpen --allow-discards /dev/sdaX luks

Non-root devices

Edit /etc/crypttab and set the discard option for devices on the SSD. For example, if you have a LUKS device with the name externaldrive1, device /dev/sdb2, and password none:

externaldrive1  /dev/sdb2   none    luks,discard

Root devices

If your root device is on LUKS, add rd.luks.allow-discards to CMDLINE_LINUX_DEFAULT. In the case of GRUB, edit /etc/default/grub:

GRUB_CMDLINE_LINUX_DEFAULT="rd.luks.allow-discards"

Then update GRUB:

# update-grub

Verifying configuration

To verify that you have configured TRIM correctly for LUKS, reboot and run:

# dmsetup table /dev/mapper/crypt_dev --showkeys

If this command output contains the string allow_discards, you have successfully enabled TRIM on your LUKS device.

ZFS

Before running trim on a ZFS pool, ensure that all devices in the pool support it:

# zpool get all | grep trim

If the pool allows autotrim (set off by default), you can trim the pool periodically or automatically. To one-shot trim yourpoolname:

# zpool trim yourpoolname

Periodic TRIM

Add the following lines to /etc/cron.daily/ztrim:

#!/bin/sh
zpool trim yourpoolname

Finally, make the script executable:

# chmod u+x /etc/cron.daily/ztrim

Autotrim

To set autotrim for yourpoolname, run:

# zpool set autotrim=on yourpoolname

Security

There are several ways you can make your installation more secure. This section explores some of them.

Section Contents

AppArmor

AppArmor is a mandatory access control mechanism (like SELinux). It can constrain programs based on pre-defined or generated policy definitions.

Void ships with some default profiles for several services, such as dhcpcd and wpa_supplicant. Container runtimes such as LXC and podman integrate with AppArmor for better security for container payloads.

To use AppArmor on a system, one must:

  1. Install the apparmor package.
  2. Set apparmor=1 security=apparmor on the kernel commandline.

To accomplish the second step, consult the documentation on how to modify the kernel cmdline.

The APPARMOR variable in /etc/default/apparmor controls how profiles will be loaded at boot, the value is set to complain by default and corresponds to AppArmor modes (disable, complain, enforce).

AppArmor tools aa-genprof(8) and aa-logprof(8) require either configured syslog or a running auditd(8) service.

Date and Time

To view your system's current date and time information, as well as make direct changes to it, use date(1).

Timezone

The default system timezone can be set by linking the timezone file to /etc/localtime:

# ln -sf /usr/share/zoneinfo/<timezone> /etc/localtime

Note: If the variable TIMEZONE is set in /etc/rc.conf, it should be removed or commented out, as this will override what has been set with ln on reboot.

To change the timezone on a per user basis, the TZ variable can be exported from your shell's profile:

export TZ=<timezone>

Note that setting the timezone does not set the time (or date); instead, it simply specifies an offset from UTC, as described in timezone(3).

Hardware clock

By default, the hardware clock in Void is stored as UTC. Windows does not use UTC by default, and if you are dual-booting, this will conflict with Void. You can either change Windows to use UTC, or change Void Linux to use localtime by setting the HARDWARECLOCK variable in /etc/rc.conf:

export HARDWARECLOCK=localtime

For more details, see hwclock(8).

NTP

To maintain accuracy of your system's clock, you can use the Network Time Protocol (NTP).

Void provides packages for the following NTP daemons: NTP, OpenNTPD, Chrony and ntpd-rs.

Once you have installed an NTP daemon, you can enable the service for it, either through its own service or the ntpd service managed by xbps-alternatives(1).

NTP

NTP is the official reference implementation of the Network Time Protocol.

The ntp package provides NTP and the isc-ntpd service.

For further information, visit the NTP site.

OpenNTPD

OpenNTPD focuses on providing a secure, lean NTP implementation which "just works" with reasonable accuracy for a majority of use-cases.

The openntpd package provides OpenNTPD and the openntpd service.

For further information, visit the OpenNTPD site.

Chrony

Chrony is designed to work well in a variety of conditions; it can synchronize faster and with greater accuracy than NTP.

The chrony package provides Chrony and the chronyd service.

The Chrony site provides a brief overview of its advantages over NTP, as well as a detailed feature comparison between Chrony, NTP and OpenNTPD.

ntpd-rs

ntpd-rs is a full-featured NTP server and client implementation, including NTS support.

The ntpd-rs package provides ntpd-rs and the ntpd-rs service.

For further information and migration guides from other implementations, visit the ntpd-rs docs.

Kernel

Kernel series

Void Linux provides many kernel series in the default repository. These are named linux<x>.<y>: for example, linux4.19. You can query for all available kernel series by running:

$ xbps-query --regex -Rs '^linux[0-9.]+-[0-9._]+'

The linux meta package, installed by default, depends on one of the kernel packages, usually the package containing the latest mainline kernel that works with all DKMS modules. Newer kernels might be available in the repository, but are not necessarily considered stable enough to be the default; use these at your own risk. If you wish to use a more recent kernel and have DKMS modules that you need to build, install the relevant linux<x>.<y>-headers package, then use xbps-reconfigure(1) to reconfigure the linux<x>.<y> package you installed. This will build the DKMS modules.

Removing old kernels

When updating the kernel, old versions are left behind in case it is necessary to roll back to an older version. Over time, old kernel versions can accumulate, consuming disk space and increasing the time taken by DKMS module updates. Furthermore, if /boot is a separate partition and fills up with old kernels, updating can fail or result in incomplete initramfs filesystems to be generated and result in kernel panics if they are being booted. Thus, it may be advisable to clean old kernels from time to time.

Removing old kernels is done using the vkpurge(8) utility. vkpurge comes pre-installed on every Void Linux system. This utility runs the necessary hooks when removing old kernels. Note that vkpurge does not remove kernel packages, only particular kernels.

Removing the default kernel series

If you've installed a kernel package for a series other than the default, and want to remove the default kernel packages, you should install the linux-base package or mark it as a manual package in case it is already installed. After this procedure, you can remove the default kernel packages with xbps-remove(1). It might be necessary to add linux and linux-headers to an ignorepkg entry in xbps.d(5), since base packages can depend on them.

Switching to another kernel series

If you'd like to use the linux-lts or linux-mainline kernel series instead of the default linux, first install the desired series metapackage (and the linux-lts-headers or linux-mainline-headers metapackage if needed). Then you can add linux and linux-headers to an ignorepkg entry in xbps.d(5) and uninstall those packages.

Changing the default initramfs generator

By default, Void Linux uses dracut to prepare initramfs images for installed kernels. Alternatives such as mkinitcpio are available. Each initramfs generator registers an XBPS alternative in the initramfs group to link its kernel hooks to be used when creating or removing initramfs images for a given kernel.

To replace dracut with, e.g., mkinitcpio, install the mkinitcpio package; confirm that mkinitcpio appears in the list of available alternatives by running

$ xbps-alternatives -l -g initramfs

Issue the command

# xbps-alternatives -s mkinitcpio

to replace the dracut kernel hooks with those provided by mkinitcpio. With subsequent kernel updates (or updates to DKMS packages that trigger initramfs regeneration), mkinitcpio will be used instead of dracut to prepare initramfs images. To force images to regenerate, reconfigure your kernel packages by invoking

# xbps-reconfigure -f linux<x>.<y>

for each linux<x>.<y> package that is currently installed.

cmdline

The kernel, the initial RAM disk (initrd) and some system programs can be configured at boot by kernel command line arguments. The parameters understood by the kernel are explained in the kernel-parameters documentation and by bootparam(7). Parameters understood by dracut can be found in dracut.cmdline(7).

Once the system is booted, the current kernel command line parameters can be found in the /proc/cmdline file. Some system programs can change their behavior based on the parameters passed in the command line, which is what happens when booting a different runsvdir, for example.

There are different ways of setting these parameters, some of which are explained below.

GRUB

Kernel command line arguments can be added through the GRUB bootloader by editing /etc/default/grub, changing the GRUB_CMDLINE_LINUX_DEFAULT variable and then running update-grub.

dracut

Dracut offers a kernel_cmdline configuration option and --kernel-cmdline command-line option that will encode command-line arguments directly in the initramfs image. When dracut is used to create a UEFI executable, arguments set with these options will be passed to the kernel. However, when an ordinary initramfs is produced, these options will not be passed to the kernel at boot. Instead, they will be written to a configuration file in /etc/cmdline.d within the image. While dracut parses this configuration to control its own boot-time behavior, the kernel itself will not be aware of anything set via this mechanism.

After modifying a dracut configuration, regenerate the initramfs to ensure that it includes the changes.

Kernel hardening

Void Linux ships with some kernel security options enabled by default. This was originally provided by kernel command line arguments slub_debug=P page_poison=1, but since kernel series 5.3, these have been replaced with init_on_alloc and init_on_free (see this commit).

Void's kernels come with the init_on_alloc option enabled by default where available (i.e. linux5.4 and greater). In most cases you should usually not disable it, as it has a fairly minimal impact on performance (within 1%). The init_on_free option is more expensive (around 5% on average) and needs to be enabled manually by passing init_on_free=1 on the kernel command line. If you need to disable init_on_alloc, you can do that similarly by passing init_on_alloc=0.

There is a chance that your existing system still has the old options enabled. They still work in newer kernels, but have a performance impact more in line with init_on_free=1. On older hardware this can be quite noticeable. If you are running a kernel series older than 5.4, you can keep them (or add them) for extra security at the cost of speed; otherwise you should remove them.

Kernel modules

Kernel modules are typically drivers for devices or filesystems.

Loading kernel modules during boot

Normally the kernel automatically loads required modules, but sometimes it may be necessary to explicitly specify modules to be loaded during boot.

To load kernel modules during boot, a .conf file like /etc/modules-load.d/virtio.conf needs to be created with the contents:

# load virtio-net
virtio-net

Blacklisting kernel modules

Blacklisting kernel modules is a method for preventing modules from being loaded by the kernel. There are two different methods for blacklisting kernel modules, one for modules loaded by the initramfs and one for modules loaded after the initramfs process is done. Modules loaded by the initramfs have to be blacklisted in the initramfs configuration.

To blacklist modules loaded after the initramfs process, create a .conf file, like /etc/modprobe.d/radeon.conf, with the contents:

blacklist radeon

Blacklisting modules in the initramfs

After making the necessary changes to the configuration files, the initramfs needs to be regenerated for the changes to take effect on the next boot.

dracut

Dracut can be configured to not include kernel modules through a configuration file. To blacklist modules from being included in a dracut initramfs, create a .conf file, like /etc/dracut.conf.d/radeon.conf, with the contents:

omit_drivers+=" radeon "
mkinitcpio

To blacklist modules from being included in a mkinitcpio initramfs a .conf file needs to be created like /etc/modprobe.d/radeon.conf with the contents:

blacklist radeon

Kernel hooks

Void Linux provides directories for kernel hooks in /etc/kernel.d/{pre-install,post-install,pre-remove,post-remove}.

These hooks are used to update the boot menus for bootloaders like grub, gummiboot and lilo.

Install hooks

The {pre,post}-install hooks are executed by xbps-reconfigure(1) when configuring a Linux kernel, such as building its initramfs. This happens when a kernel series is installed for the first time or updated, but can also be run manually:

# xbps-reconfigure --force linux<x>.<y>

If run manually, they serve to apply initramfs configuration changes to the next boot.

Remove hooks

The {pre,post}-remove hooks are executed by vkpurge(8) when removing old kernels.

Dynamic Kernel Module Support (DKMS)

There are kernel modules that are not part of the Linux source tree that are built at install time using DKMS and kernel hooks. The available modules can be listed by searching for dkms in the package repositories.

DKMS build logs are available in /var/lib/dkms/.

Power Management

acpid

The acpid service for acpid(8) is installed and, if Void was installed from a live image using the local source, will be enabled by default. ACPI events are handled by /etc/acpi/handler.sh, which uses zzz(8) for suspend-to-RAM events.

elogind

The elogind service is provided by the elogind package. By default, elogind(8) listens for, and processes, ACPI events related to lid-switch activation and the power, suspend and hibernate keys. This will conflict with the acpid service if it is installed and enabled. Either disable acpid when enabling elogind, or configure elogind to ignore ACPI events in logind.conf(5). There are several configuration options, all starting with the keyword Handle, that should be set to ignore to avoid interfering with acpid.

To run loginctl poweroff and loginctl reboot without root privileges, polkit must be installed.

Power Saving - tlp

Laptop battery life can be extended by using tlp(8). To use it, install the tlp package, and enable the tlp service. Refer to the TLP documentation for details.

Network

Network configuration in Void Linux can be done in several ways. The default installation comes with the dhcpcd(8) service enabled.

Interface Names

Newer versions of udev(7) no longer use the traditional Linux naming scheme for interfaces (eth0, eth1, wlan0, ...).

This behavior can be reverted by adding net.ifnames=0 to the kernel cmdline.

Static Configuration

A simple way to configure a static network at boot is to add the necessary ip(8) commands to the /etc/rc.local file:

ip link set dev eth0 up
ip addr add 192.168.1.2/24 brd + dev eth0
ip route add default via 192.168.1.1

Bridge Interfaces

To configure bridge interfaces at boot, the /etc/rc.local file can be used to run ip(8) commands to add the bridge br0 and set it as the master for the eth0 interface as example:

ip link add name br0 type bridge
ip link set eth0 master br0
ip link set eth0 up

dhcpcd

To run dhcpcd(8) on all interfaces, enable the dhcpcd service.

To run dhcpcd only on a specific interface, copy the dhcpcd-eth0 service and modify it to match your interface:

$ ip link show
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: enp3s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP mode DEFAULT group default qlen 1000
        link/ether ff:ff:ff:ff:ff:ff brd ff:ff:ff:f
# cp -R /etc/sv/dhcpcd-eth0 /etc/sv/dhcpcd-enp3s0
# sed -i 's/eth0/enp3s0/' /etc/sv/dhcpcd-enp3s0/run
# ln -s /etc/sv/dhcpcd-enp3s0 /var/service/

For more information on configuring dhcpcd, refer to dhcpcd.conf(5)

Wireless

Before using wireless networking, use rfkill(8) to check whether the relevant interfaces are soft- or hard-blocked.

Void provides several ways to connect to wireless networks:

Firewalls

iptables

By default, the iptables package is installed on the base system. It provides iptables(8)/ip6tables(8). The related services use the /etc/iptables/iptables.rules and /etc/iptables/ip6tables.rules ruleset files, which must be created by the system administrator.

Two example rulesets are provided in the /etc/iptables directory: empty.rules and simple_firewall.rules.

Applying the rules at boot

To apply iptables rules at runit stage 1, install the runit-iptables package. This adds a core-service which restores the iptables.rules and ip6tables.rules rulesets.

Alternatively, to apply these rules at stage 2, add the following to /etc/rc.local:

if [ -e /etc/iptables/iptables.rules ]; then
  iptables-restore /etc/iptables/iptables.rules
fi

if [ -e /etc/iptables/ip6tables.rules ]; then
  ip6tables-restore /etc/iptables/ip6tables.rules
fi

After rebooting, check the active firewall rules:

# iptables -L
# ip6tables -L

Applying the rules at runtime

iptables comes with two runit services, iptables and ip6tables, to quickly flush or restore the iptables.rules and ip6tables.rules rulesets. Once these services are enabled, you can flush the rulesets by downing the relevant service, e.g.:

# sv down iptables

and restore them by upping the relevant service, e.g.:

# sv up ip6tables

nftables

nftables replaces iptables, ip6tables, arptables and ebtables (collectively referred to as xtables). The nftables wiki describes the main differences from the iptables toolset.

To use nftables, install the nftables package, which provides nft(8). It also provides iptables-translate(8)/ip6tables-translate(8) and iptables-restore-translate(8)/ip6tables-restore-translate(8), which convert iptables rules to nftables rules.

Applying the rules at boot

To apply nftables rules at runit stage 1, install the runit-nftables package. This adds a core-service which restores the ruleset in /etc/nftables.conf.

Applying the rules at runtime

The nftables package provides the nftables service, which uses rules from /etc/nftables.conf. Once you enable the nftables service, to load the rules, run:

# sv up nftables

To flush the rules, run:

# sv down nftables

wpa_supplicant

The wpa_supplicant package is installed by default on the base system. It includes utilities to configure wireless interfaces and handle wireless security protocols. To use wpa_supplicant, you will need to enable the wpa_supplicant service.

wpa_supplicant(8) is a daemon that manages wireless interfaces based on wpa_supplicant.conf(5) configuration files. An extensive overview of configuration options, including examples, can be found in /usr/share/examples/wpa_supplicant/wpa_supplicant.conf.

wpa_passphrase(8) helps create pre-shared keys for use in configuration files. wpa_cli(8) provides a CLI for managing the wpa_supplicant daemon.

WPA-PSK

To use WPA-PSK, generate a pre-shared key with wpa_passphrase(8) and append the output to the relevant wpa_supplicant.conf file:

# wpa_passphrase <MYSSID> <passphrase> >> /etc/wpa_supplicant/wpa_supplicant-<device_name>.conf

WPA-EAP

WPA-EAP is often used for institutional logins, notably eduroam. This does not use PSK, but a password hash can be generated like this:

$ echo -n <passphrase> | iconv -t utf16le | openssl md4

WEP

For WEP configuration, add the following lines to your device's wpa-supplicant.conf:

network={
    ssid="MYSSID"
    key_mgmt=NONE
    wep_key0="YOUR AP WEP KEY"
    wep_tx_keyidx=0
    auth_alg=SHARED
}

The wpa_supplicant service

The wpa_supplicant service checks the following options in /etc/sv/wpa_supplicant/conf:

  • OPTS: Options to be passed to the service. Overrides any other options.
  • CONF_FILE: Path to file to be used for configuration. Defaults to /etc/wpa_supplicant/wpa_supplicant.conf.
  • WPA_INTERFACE: Interface to be matched. May contain a wildcard; defaults to all interfaces.
  • DRIVER: Driver to use. See wpa_supplicant -h for available drivers.

If no conf file is found, the service searches for the following files in /etc/wpa_supplicant:

  • wpa_supplicant-<interface>.conf: If found, these files are bound to the named interface.
  • wpa_supplicant.conf: If found, this file is loaded and binds to all other interfaces found.

Once you are satisfied with your configuration, enable the wpa_supplicant service.

Using wpa_cli

When using wpa_cli to manage wpa_supplicant from the command line, be sure to specify which network interface to use via the -i option, e.g.:

# wpa_cli -i wlp2s0

Not doing so can result in various wpa_cli commands (for example, scan and scan_results) not producing the expected output.

IWD

IWD (iNet Wireless Daemon) is a wireless daemon for Linux that aims to replace WPA supplicant.

Installation

Install the iwd package and enable the dbus and iwd services.

Usage

The command-line client iwctl(1) can be used to add, remove, and configure network connections. Commands can be passed as arguments; when run without arguments, it provides an interactive session. To list available commands, run iwctl help, or run iwctl and enter help at the interactive prompt.

By default, only the root user and those in the wheel group have permission to operate iwctl.

Configuration

Daemon configuration

The main configuration file is located in /etc/iwd/main.conf. If it does not exist, you may create it. It is documented in iwd.config(5).

Network configuration

Network configuration, including examples, is documented in iwd.network(5). IWD stores information on known networks, and reads information on pre-provisioned networks from network configuration files located in /var/lib/iwd; IWD monitors the directory for changes. Network configuration filenames consist of the encoding of the SSID followed by .open, .psk, or .8021x as determined by the security type.

As an example, a basic configuration file for a WPA2/PSK secured network would be called <ssid>.psk, and it would contain the plain text password:

[Security]
Passphrase=<password>

Troubleshooting

By default, IWD will create and destroy the wireless interfaces (e.g. wlan0) that it manages. This can interfere with udevd, which may attempt to rename the interface using its rules for persistent network interface names. The following messages may be printed to your screen as a symptom of this interference:

[   39.441723] udevd[1100]: Error changing net interface name wlan0 to wlp59s0: Device or resource busy
[   39.442472] udevd[1100]: could not rename interface '3' from 'wlan0' to 'wlp59s0': Device or resource busy

A simple fix is to prevent IWD from manipulating the network interfaces in this way by adding UseDefaultInterface=true to the [General] section of /etc/iwd/main.conf.

An alternative approach is to disable the use of persistent network interface names by udevd. This may be accomplished either by adding net.ifnames=0 to your kernel cmdline or by creating a symbolic link to /dev/null at /etc/udev/rules.d/80-net-name-slot.rules to mask the renaming rule. This alternative approach will affect the naming of all network devices.

NetworkManager

NetworkManager(8) is a daemon that manages Ethernet, Wi-Fi, and mobile broadband network connections. Install the NetworkManager package for the basic NetworkManager utilities.

Starting NetworkManager

Before enabling the NetworkManager daemon, disable any other network management services, such as dhcpcd, wpa_supplicant, or wicd. These services all control network interface configuration, and will interfere with NetworkManager.

Also ensure that the dbus service is enabled and running. NetworkManager uses dbus to expose networking information and a control interface to clients, and will fail to start without it.

Finally, enable the NetworkManager service.

Configuring NetworkManager

Users of NetworkManager must belong to the network group.

The NetworkManager package includes a command line tool, nmcli(1), and a text-based user interface, nmtui(1), to control network settings.

There are many other front-ends to NetworkManager, including nm-applet (from network-manager-applet) or nm-tray for system trays, plasma-nm for KDE Plasma, and a built-in network configuration tool in GNOME.

Eduroam with NetworkManager

Eduroam is a roaming service providing international, secure Internet access at universities and other academic institutions. More information can be found here.

Dependencies

Install the python3-dbus package.

Installation

Download the correct eduroam_cat installer for your institution from here and execute it. It will provide a user interface guiding you through the process.

ConnMan

ConnMan(8) is a daemon that manages network connections, is designed to be slim and to use as few resources as possible. The connman package contains the basic utilities to run ConnMan.

Starting ConnMan

To enable the ConnMan daemon, first disable any other network managing services like dhcpcd, wpa_supplicant, or wicd. These services all control network interface configuration, and interfere with each other.

Finally, enable the connmand service.

Configuring ConnMan

ConnMan CLI

The connman package includes a command line tool, connmanctl(1) to control network settings. If you do not provide any commands, connmanctl starts as an interactive shell.

Establishing a connection to an access point using the connmanctl interactive shell might look as follows:

# connmanctl
> enable wifi
> agent on
> scan wifi
> services
> connect wifi_<uniqueid>
> exit

ConnMan Front-End Tools

There are many other front-ends to ConnMan, including connman-ui for system trays, connman-gtk for GTK, cmst for QT and connman-ncurses for ncurses based UI.

Preventing DNS overrides by ConnMan

Create /etc/sv/connmand/conf with the following content:

OPTS="--nodnsproxy"

Network Filesystems

NFS

Mounting an NFS Share

To mount an NFS share, start by installing the nfs-utils and sv-netmount packages.

Before mounting an NFS share, enable the statd, rpcbind, and netmount services. If the server supports nfs4, the statd service isn't necessary.

To mount an NFS share:

# mount -t <mount_type> <host>:/path/to/sourcedir /path/to/destdir

<mount_type> should be nfs4 if the server supports it, or nfs otherwise. <host> can be either the hostname or IP address of the server.

Mounting options can be found in mount.nfs(8), while unmounting options can be found in umount.nfs(8).

For example, to connect /volume on a server at 192.168.1.99 to an existing /mnt/volume directory on your local system:

# mount -t nfs 192.168.1.99:/volume /mnt/volume

To have the directory mounted when the system boots, add an entry to fstab(5):

192.168.1.99:/volume /mnt/volume nfs rw,hard 0 0

Refer to nfs(5) for information about the available mounting options.

Setting up a server (NFSv4, Kerberos disabled)

To run an NFS server, start by installing the nfs-utils package.

Edit /etc/exports to add a shared volume:

/storage/foo    *.local(rw,no_subtree_check,no_root_squash)

This line exports the /storage/foo directory to any host in the local domain, with read/write access. For information about the no_subtree_check and no_root_squash options, and available options more generally, refer to exports(5).

Finally, enable the rpcbind, statd, and nfs-server services.

This will start your NFS server. To check if the shares are working, use the showmount(8) utility to check the NFS server status:

# showmount -e localhost

You can use nfs.conf(5) to configure your server.

Session and Seat Management

Session and seat management is not necessary for every setup, but it can be used to safely create temporary runtime directories, provide access to hardware devices and multi-seat capabilities, and control system shutdown.

D-Bus

D-Bus is an IPC (inter-process communication) mechanism used by userspace software in Linux. D-Bus can provide a system bus and/or a session bus, the latter being specific to a user session.

  • To provide a system bus, you should enable the dbus service. This might require a system reboot to work properly.
  • To provide a session bus, you can start a given program (usually a window manager or interactive shell) with dbus-run-session(1). Most desktop environments, if launched through an adequate display manager, will launch a D-Bus session themselves. If a D-Bus session is active for the current session, the environment variable DBUS_SESSION_BUS_ADDRESS should be defined.

Note that some software assumes the presence of a system bus, while other software assumes the presence of a session bus.

elogind

elogind(8) manages user logins and system power, as a standalone version of systemd-logind. elogind provides necessary features for most desktop environments and Wayland compositors. It can also be one of the mechanisms for rootless Xorg.

Please read the "Power Management" section for things to consider before installing elogind.

To make use of its features, install the elogind package and make sure the system D-Bus is enabled. You might need to log out and in again.

If you're having any issues with elogind, enable its service, as waiting for a D-Bus activation can lead to issues.

turnstile

turnstile is an alternative session manager, and can be used with or without elogind.

If using turnstile with elogind, disable rundir (XDG_RUNTIME_DIR) management in /etc/turnstile/turnstiled.conf by setting manage_rundir to no. Note that while elogind and turnstile currently can coexist, this may change in the future.

If using turnstile without elogind, consider installing and enabling seatd for seat management and acpid for power management.

To use it, enable the turnstiled service and re-log in.

Turnstile can also manage per-user services (including running a D-Bus session bus), removing the need for wrapping graphical sessions with dbus-run-session.

seatd

seatd(1) is a minimal seat management daemon and an alternative to elogind primarily for wlroots compositors.

To use it, install the seatd package and enable its service. If you want non-root users to be able to access the seatd session, add them to the _seatd group.

Note that, unlike elogind, seatd doesn't do anything besides managing seats.

XDG_RUNTIME_DIR

XDG_RUNTIME_DIR is an environment variable defined by the XDG Base Directory Specification. Its value sets the path to the base directory where programs should store user-specific runtime files.

Install elogind or turnstile as your session manager to automatically set up XDG_RUNTIME_DIR.

Alternatively, manually set the environment variable through the shell. Make sure to create a dedicated user directory and set its permissions to 700. A good default location is /run/user/$(id -u).

Graphical Session

In order to configure a graphical session, you need:

You may also need:

Graphics Drivers

This section covers basic graphics setup depending on the hardware configuration of your system.

Section Contents

AMD or ATI

AMD GPU support requires the linux-firmware-amd package. If you have installed the linux or linux-lts packages, it will be installed as a dependency. If you installed a version-specific kernel package (e.g., linux5.4), it may be necessary to manually install linux-firmware-amd.

OpenGL

Install the Mesa DRI package, mesa-dri. This is already included in the xorg meta-package, but it is needed when installing Xorg via xorg-minimal or for running a Wayland compositor.

Vulkan

Install vulkan-loader, the Khronos Vulkan Loader. Then install one or both of the Mesa AMD Vulkan driver, mesa-vulkan-radeon; or the GPUOpen AMD Vulkan driver, amdvlk.

Xorg

Installing the xorg meta-package will pull in both xf86-video-amdgpu and, for older hardware, xf86-video-ati. If you install xorg-minimal, choose one of these Xorg driver packages to match your hardware. The amdgpu driver should support cards built on AMD's "Graphics Core Next 1.2" architecture, introduced circa 2012.

Video acceleration

Install the mesa-vaapi and mesa-vdpau packages.

Intel

Intel GPU support requires the linux-firmware-intel package. If you have installed the linux or linux-lts packages, it will be installed as a dependency. If you installed a version-specific kernel package (e.g., linux5.4), it may be necessary to manually install linux-firmware-intel.

OpenGL

OpenGL requires the Mesa DRI package, mesa-dri. This is provided by the xorg meta-package, but will need to be installed manually when using the xorg-minimal package or running a Wayland compositor.

Vulkan

Install the Khronos Vulkan Loader and the Mesa Intel Vulkan driver packages, respectively vulkan-loader and mesa-vulkan-intel.

Video acceleration

Install the intel-video-accel meta-package:

This will install all the Intel VA-API drivers. intel-media-driver will be used by default, but this choice can be overridden at runtime via the environment variable LIBVA_DRIVER_NAME:

Driver PackageSupported GPU GenExplicit selection
libva-intel-driverup to Coffee LakeLIBVA_DRIVER_NAME=i965
intel-media-driverfrom BroadwellLIBVA_DRIVER_NAME=iHD

Troubleshooting

The kernels packaged by Void are configured with CONFIG_INTEL_IOMMU_DEFAULT_ON=y, which can lead to issues with their graphics drivers, as reported by the kernel documentation. To fix this, it is necessary to disable IOMMU for the integrated GPU. This can be done by adding intel_iommu=igfx_off to your kernel cmdline. This problem is expected to happen on the Broadwell generation of internal GPUs. If you have another internal GPU and your issues are fixed by this kernel option, you should file a bug reporting the problem to kernel developers.

For newer Intel chipsets, the DDX drivers may interfere with correct operation. This is characterized by graphical acceleration not working and general graphical instability. If this is the case, try removing all xf86-video-* packages.

On some Intel chipsets, the DDX drivers may be required. If this is the case, install xf86-video-intel.

NVIDIA

nouveau (Open Source Driver)

This is a reverse engineered driver largely developed by the community, with some documentation provided by Nvidia. It tends to perform well on older hardware, and is required to use a large portion of the available Wayland compositors.

At the time of writing, graphics cards starting with second generation Maxwell (GTX 9xx) are unable to perform at their full potential with nouveau. This is because the linux-firmware collection is missing signed firmware blobs needed to reclock these cards past their boot frequencies.

To use nouveau with Wayland, you only need the mesa-dri package, which provides the accelerated OpenGL driver. On X11, you also need an appropriate Xorg driver. You can either install xf86-video-nouveau or use the universal modesetting driver bundled with Xorg (this is the only option on Tegra based ARM boards). The former can make use of GPU-specific 2D acceleration paths, which is primarily useful on older cards with specialized fixed function hardware (the modesetting driver will accelerate 2D using OpenGL via GLAMOR). When in doubt, it's a good idea to try xf86-video-nouveau first.

Note: xf86-video-nouveau is usually installed by default if you use the xorg metapackage. If you use xorg-minimal, you will need to install it manually, either directly or through xorg-video-drivers.

nvidia (Proprietary Driver)

The proprietary drivers are available in the nonfree repository.

Check if your graphics card belongs to the legacy branch. If it does not, install the nvidia package. Otherwise you should install the appropriate legacy driver, nvidia470 or nvidia390. The older legacy driver, nvidia340, is no longer available, and users are encouraged to switch to nouveau.

BrandTypeModelDriver Package
NVIDIAProprietary800+nvidia
NVIDIAProprietary600/700nvidia470
NVIDIAProprietary400/500 Seriesnvidia390

The proprietary driver integrates in the kernel through DKMS.

This driver offers better performance and power handling, and is recommended where performance is needed.

32-bit program support (glibc only)

In order to run 32-bit programs with driver support, you need to install additional packages.

If using the nouveau driver, install the mesa-dri-32bit package.

If using the nvidia driver, install the nvidia<x>-libs-32bit package. <x> represents the legacy driver version (470 or 390) or can be left empty for the main driver.

Reverting from nvidia to nouveau

Uninstalling nvidia

In order to revert to the nouveau driver, install the nouveau driver (if it was not installed already), then remove the nvidia, nvidia470, or nvidia390 package, as appropriate.

If you were using the obsolete nvidia340 driver, you might need to install the libglvnd package after removing the nvidia340 package.

Keeping both drivers

It is possible to use the nouveau driver while still having the nvidia driver installed. To do so, remove the blacklisting of nouveau in /etc/modprobe.d/nouveau_blacklist.conf, /usr/lib/modprobe.d/nvidia.conf, or /usr/lib/modprobe.d/nvidia-dkms.conf by commenting it out:

#blacklist nouveau

For Xorg, specify that it should load the nouveau driver rather than the nvidia driver by creating the file /etc/X11/xorg.conf.d/20-nouveau.conf with the following content:

Section "Device"
    Identifier "Nvidia card"
    Driver "nouveau"
EndSection

You may need to reboot your system for these changes to take effect.

NVIDIA Optimus

NVIDIA Optimus refers to a dual graphics configuration found on laptops consisting of an Intel integrated GPU and a discrete NVIDIA GPU.

There are different methods to take advantage of the NVIDIA GPU, which depend on the driver version supported by your hardware.

In order to determine the correct driver to install, it is not enough to look at the "Supported Products" list on NVIDIA's website, because they are not guaranteed to work in an Optimus configuration. So the only way is to try installing the latest nvidia, rebooting, and looking at the kernel log. If your device is not supported, you will see a message like this:

NVRM: The NVIDIA GPU xxxx:xx:xx.x (PCI ID: xxxx:xxxx)
NVRM: installed in this system is not supported by the xxx.xx
NVRM: NVIDIA Linux driver release.  Please see 'Appendix
NVRM: A - Supported NVIDIA GPU Products' in this release's
NVRM: README, available on the Linux driver download page
NVRM: at www.nvidia.com.

which means you have to uninstall nvidia and install the legacy nvidia390.

A summary of the methods supported by Void, which are mutually exclusive:

PRIME Render Offload

  • available on nvidia and nvidia470
  • allows to switch to the NVIDIA GPU on a per-application basis
  • more flexible but power saving capabilities depend on the hardware (pre-Turing devices are not shut down completely)

Offloading Graphics Display with RandR 1.4

  • available on nvidia, nvidia470, and nvidia390
  • allows to choose which GPU to use at the start of the X session
  • less flexible, but allows the user to completely shut down the NVIDIA GPU when not in use, thus saving power

Bumblebee

  • available on nvidia, nvidia470, and nvidia390
  • allows to switch to the NVIDIA GPU on a per-application basis
  • unofficial method, offers poor performance

Nouveau PRIME

  • uses the open source driver nouveau
  • allows to switch to the NVIDIA GPU on a per-application basis
  • nouveau is a reverse-engineered driver and offers poor performance

You can check the currently used GPU by searching for renderer string in the output of the glxinfo command. It is necessary to install the glxinfo package for this. For the first two alternatives below, it is also possible to verify that a process is using the NVIDIA GPU by checking the output of nvidia-smi.

PRIME Render Offload

In this method, GPU switching is done by setting environment variables when executing the application to be rendered on the NVIDIA GPU. The wrapper script prime-run is available from the nvidia package, and can be used as shown below:

$ prime-run <application>

For more information, see NVIDIA's README

Bumblebee

Enable the bumblebeed service and add the user to the bumblebee group. This requires a re-login to take effect.

Run the application to be rendered on the NVIDIA GPU with optirun:

$ optirun <application>

Nouveau PRIME

This method uses the open source nouveau driver. If the NVIDIA drivers are installed, it is necessary to configure the system to use nouveau.

Set DRI_PRIME=1 to run an application on the NVIDIA GPU:

$ DRI_PRIME=1 <application>

Xorg

This section details the manual installation and configuration of the Xorg display server and common related services and utilities. If you would just like to install a full desktop environment, it is recommended to try the xfce image.

Installation

Void provides a comprehensive xorg package which installs the server and all of the free video drivers, input drivers, fonts, and base applications. This package is a safe option, and should be adequate for most systems which don't require proprietary video drivers.

If you would like to select only the packages you need, the xorg-minimal package contains the base xorg server only. If you install only xorg-minimal, you will likely need to install a font package (like xorg-fonts), a terminal emulator (like xterm), and a window manager to have a usable graphics system.

Video Drivers

Void provides both open-source and proprietary (non-free) video drivers.

Open Source Drivers

Xorg can use two categories of open source drivers: DDX or modesetting.

DDX

The DDX drivers are installed with the xorg package by default, or may be installed individually if the xorg-minimal package was installed. They are provided by the xf86-video-* packages.

For advanced configuration, see the man page corresponding to the vendor name, like intel(4).

Modesetting

Modesetting requires the mesa-dri package, and no additional vendor-specific driver package.

Xorg defaults to DDX drivers if they are present, so in this case modesetting must be explicitly selected: see Forcing the modesetting driver.

For advanced configuration, see modesetting(4).

Proprietary Drivers

Void also provides proprietary NVIDIA drivers, which are available in the nonfree repository.

Input Drivers

A number of input drivers are available for Xorg. If xorg-minimal was installed and a device is not responding, or behaving unexpectedly, a different driver may correct the issue. These drivers can grab everything from power buttons to mice and keyboards. They are provided by the xf86-input-* packages.

Xorg Configuration

Although Xorg normally auto-detects drivers and configuration is not needed, a config for a specific keyboard driver may look something like a file /etc/X11/xorg.conf.d/30-keyboard.conf with the contents:

Section "InputClass"
  Identifier "keyboard-all"
  Driver "evdev"
  MatchIsKeyboard "on"
EndSection

Forcing the modesetting driver

Create the file /etc/X11/xorg.conf.d/10-modesetting.conf:

Section "Device"
    Identifier "GPU0"
    Driver "modesetting"
EndSection

and restart Xorg. Verify that the configuration has been picked up with:

$ grep -E -m1 '\(II\) modeset\([0-9]+\):' /var/log/Xorg.0.log

If there is a match, modesetting is being used.

Starting X Sessions

startx

The xinit package provides the startx(1) script as a frontend to xinit(1), which can be used to start X sessions from the console. For example, to start i3(1), edit ~/.xinitrc to contain exec /bin/i3 on the last line.

To start arbitrary programs together with an X session, add them in ~/.xinitrc before the last line. For example, to start xscreensaver(1) before starting i3, add xscreensaver & before the last line.

A ~/.xinitrc file which starts xscreensaver and i3 is shown below:

xscreensaver &
exec /bin/i3

Then call startx to start a session.

If a D-Bus session bus is required, you can manually start one.

Display Managers

Display managers (DMs) provide a graphical login UI. A number of DMs are available in the Void repositories, including gdm (the GNOME DM), sddm (the KDE DM) and lightdm. When setting up a display manager, be sure to test the service before enabling it.

Wayland

This section details the manual installation and configuration of Wayland compositors and related services and utilities.

Installation

Unlike Xorg, Wayland implementations combine the display server, the window manager and the compositor in a single application.

Desktop Environments

GNOME, KDE Plasma and Enlightenment have Wayland sessions. GNOME uses its Wayland session by default. When using these desktop environments, applications built with GTK+ will automatically choose the Wayland backend, while Qt5 and EFL applications might require setting some environment variables if used outside KDE or Enlightenment, respectively.

Standalone compositors

Void Linux currently packages the following Wayland compositors:

  • Weston: reference compositor for Wayland
  • Sway: an i3-compatible Wayland compositor
  • Wayfire: 3D Wayland compositor
  • Hikari: a stacking compositor with some tiling features
  • Cage: a Wayland kiosk
  • River: a dynamic tiling Wayland compositor
  • labwc: a window-stacking compositor, inspired by Openbox
  • Qtile: a dynamic tiling Wayland compositor (via qtile-wayland)

Some compositors do not depend on any fonts, which can cause many applications to not work. Install a font package to fix this.

Video drivers

Both GNOME and KDE Plasma have EGLStreams backends for Wayland, which means they can use the proprietary NVIDIA drivers. Most other Wayland compositors require drivers that implement the GBM interface. The main driver for this purpose is provided by the mesa-dri package. The "Graphics Drivers" section has more details regarding setting up graphics in different systems.

Seat management

Wayland compositors require some way of controlling the video display and accessing input devices. In Void systems, this requires a seat manager service, which can be either elogind or seatd. Enabling them is explained in the "Session and Seat Management" session.

Native applications

Qt5-based applications require installing the qt5-wayland package and setting the environment variable QT_QPA_PLATFORM=wayland-egl to enable their Wayland backend. Some KDE specific applications also require installing the kwayland package. EFL-based applications require setting the environment variable ELM_DISPLAY=wl, and can have issues without it, due to not supporting XWayland properly. SDL-based applications require setting the environment variable SDL_VIDEODRIVER=wayland. GTK+-based applications should use the Wayland backend automatically.

Media applications, such as mpv(1), vlc(1) and imv work natively on Wayland.

Web browsers

Mozilla Firefox ships with a Wayland backend which is disabled by default. To enable the Wayland backend, either set the environment variable MOZ_ENABLE_WAYLAND=1 before running firefox or use the provided firefox-wayland script.

Browsers based on GTK+ or Qt5, such as Midori and qutebrowser(1), should work on Wayland natively.

Running X applications inside Wayland

If an application doesn't support Wayland, it can still be used in a Wayland context. XWayland is an X server that bridges this gap for most Wayland compositors, and is installed as a dependency for most of them. Its package is xorg-server-xwayland. For Weston, the correct package is weston-xwayland.

Configuration

The Wayland library requires the XDG_RUNTIME_DIR environment variable to determine the directory for the Wayland socket.

It is also possible that some applications use the XDG_SESSION_TYPE environment variable in some way, which requires that you set it to wayland.

Fonts

A number of fonts and font collections are available from XBPS. dejavu-fonts-ttf or xorg-fonts are a good baseline if you're unsure of what to pick. noto-fonts-ttf contains fonts for many languages and scripts. noto-fonts-cjk extends this with fonts for Chinese, Japanese, and Korean, and noto-fonts-emoji provides emojis. nerd-fonts provides a number of fonts with special characters like custom icons included.

Fonts not available from XBPS can be manually installed to either /usr/share/fonts (system-wide) or ~/.local/share/fonts (per-user).

To customize font display in your graphical session, you can use configurations provided in /usr/share/fontconfig/conf.avail/. To do so, create a symlink to the relevant .conf file in /etc/fonts/conf.d/, then use xbps-reconfigure(1) to reconfigure the fontconfig package.

For example, to disable use of bitmap fonts:

# ln -s /usr/share/fontconfig/conf.avail/70-no-bitmaps.conf /etc/fonts/conf.d/
# xbps-reconfigure -f fontconfig

Use fc-conflist(1) to list which configurations are in effect.

Icons

GTK

By default, GTK-based applications try to use the Adwaita icon theme for application icons. Consequently, installation of the gtk+3 package will also install the adwaita-icon-theme package. If you wish to use a different theme, install the relevant package, then specify the theme in /etc/gtk-3.0/settings.ini or ~/.config/gtk-3.0/settings.ini. adwaita-icon-theme can be removed after ignoring the package.

For information about how to specify a different GTK icon theme in settings.ini, refer to the GtkSettings documentation, in particular the "gtk-icon-theme-name" property.

XDG Desktop Portals

Some applications, including Flatpak, use XDG Desktop Portals to provide access to various system interfaces, including file open and save dialogs, the clipboard, screencasting, opening URLs, and more.

Installation

XDG Desktop Portals require a user D-Bus session bus. Install xdg-desktop-portal and one or more backends:

BackendNotes
xdg-desktop-portal-gnomeProvides most common and GNOME-specific interfaces (GTK+ UI)
xdg-desktop-portal-gtkProvides most common interfaces (GTK+ UI)
xdg-desktop-portal-kdeProvides most common and KDE-specific interfaces (Qt/KF5 UI)
xdg-desktop-portal-lxqtOnly provides a file chooser (based on libfm-qt)
io.elementary.filesOnly provides a file chooser
xdg-desktop-portal-wlrOnly provides a screenshot and screencasting interface for wlroots compositors

If unsure what to choose, xdg-desktop-portal-gtk is a good default choice.

Configuration

In most cases, the default configuration, located at /usr/share/xdg-desktop-portal/portals.conf, should suffice. If necessary, this configuration can be overridden for specific desktop environments and portal interfaces by creating $XDG_CURRENT_DESKTOP-portals.conf or portals.conf at the system or user level as described in portals.conf(5).

GNOME

Pre-installation

GNOME supports both X and Wayland sessions. Follow the "Wayland" or "Xorg" sections to setup your preferred environment.

Install the dbus package, ensure the dbus service is enabled, and reboot for the changes to take effect.

Installation

Install the gnome package for a GNOME environment which includes the base GNOME desktop and a subset of GNOME applications. Additional applications are available via the gnome-apps package.

A minimal GNOME environment can be created by installing the gnome-core package. Note, however, that not all GNOME features may be present or functional.

If you require ZeroConf support, install the avahi package and enable the avahi-daemon service.

Starting GNOME

The gdm package provides the gdm service for the GNOME Display Manager; test the service before enabling it. GDM defaults to providing a Wayland session via the mutter window manager, but an X session can be chosen instead.

KDE

Installation

Install the kde-plasma package, and optionally, the kde-baseapps package.

To use the "Networks" widget, install NetworkManager and enable the dbus and NetworkManager services. To use the "Volume" widget, set up PipeWire or PulseAudio.

Installing the kde-plasma package also installs the sddm package, which provides the sddm service for the Simple Desktop Display Manager. This service depends on the dbus service being enabled; test the service before enabling it. If you are not intending to run SDDM via a remote X server, you will need to install either the xorg-minimal package or the xorg package. By default, SDDM will start an X-based Plasma session, but you can request a Wayland-based Plasma session instead.

If you wish to start an X-based session from the console, use startx to run startplasma-x11. For a Wayland-based session, run startplasma-wayland directly.

Dolphin

Dolphin is the default file manager of the KDE desktop environment. It can be installed on its own by installing the dolphin package, or it can be installed as part of the kde-baseapps meta-package.

Thumbnail Previews

To enable thumbnail file previews, install the kdegraphics-thumbnailers package. If you want video thumbnails, the ffmpegthumbs package is also necessary. Enable previews in "Control" -> "Configure Dolphin" -> "General" -> "Previews" by checking the corresponding boxes. File previews will be shown for the selected file types after clicking "Preview" in Dolphin's toolbar.

Multimedia

Audio setup

To setup audio on your Void Linux system you have to decide if you want to use PulseAudio, PipeWire or just ALSA. Sndio is also available, but is neither supported nor recommended.

Some applications require PulseAudio, especially closed source programs, but PipeWire provides a drop-in replacement for PulseAudio.

If elogind is not enabled, it is necessary to be in the audio group in order to have access to audio devices.

ALSA

To use ALSA, install the alsa-utils package and make sure your user is a member of the audio group.

The alsa-utils package provides the alsa service. When enabled, this service saves and restores the state of ALSA (e.g. volume) at shutdown and boot, respectively.

To allow use of software requiring PulseAudio, install the apulse package. apulse provides part of the PulseAudio interface expected by applications, translating calls to that interface into calls to ALSA. For details about using apulse, consult the project README.

Configuration

The default sound card can be specified via ALSA configuration files or via kernel module options.

To obtain information about the order of loaded sound card modules:

$ cat /proc/asound/modules
 0 snd_hda_intel
 1 snd_hda_intel
 2 snd_usb_audio

To set a different card as the default, edit /etc/asound.conf or the per-user configuration file ~/.asoundrc:

defaults.ctl.card 2;
defaults.pcm.card 2;

or specify sound card module order in /etc/modprobe.d/alsa.conf:

options snd_usb_audio index=0

Dmix

The dmix ALSA plugin allows playing sound from multiple sources. dmix is enabled by default for soundcards which do not support hardware mixing. To enable it for digital output, edit /etc/asound.conf:

pcm.dsp {
    type plug
    slave.pcm "dmix"
}

PipeWire

PipeWire is a modern server for handling audio (and video) streams. It is highly flexible and can interface with applications designed for ALSA, PulseAudio, and JACK audio systems. It is also designed to work well with Flatpak applications and provides a method for screenshotting and screensharing on Wayland via xdg-desktop-portal.

Prerequisites

PipeWire requires an active D-Bus user session bus. If your desktop environment, window manager, or Wayland compositor is configured to provide this, no further configuration should be required. If not, the desktop environment, window manager, or Wayland compositor may need to be launched with dbus-run-session(1).

PipeWire also requires the XDG_RUNTIME_DIR environment variable to be defined in your environment to work properly.

If not using elogind, it is necessary to be in the audio group to access audio devices and the video group to access video devices.

Basic Setup

To use PipeWire, install the pipewire package. This will also install a PipeWire session manager, wireplumber.

Session Management

In PipeWire, a session manager assumes responsibility for interconnecting media sources and sinks as well as enforcing routing policy. Without a session manager, PipeWire will not function.

If you have installed an earlier version of the Void pipewire package, make sure to update your system to eliminate any stale system configuration that may attempt to launch pipewire-media-session, the original PipeWire session manager. Users who previously overrode the system configuration to use wireplumber, e.g. by placing a custom pipewire.conf file in /etc/pipewire or ${XDG_CONFIG_HOME}/pipewire, may wish to reconcile these overrides with /usr/share/pipewire/pipewire.conf installed by the most recent pipewire package. If the sole purpose of a prior override was to disable pipewire-media-session, deleting the custom configuration may be sufficient.

Currently, there is only one session manager available: WirePlumber. To configure PipeWire to use this session manager and ensure proper startup ordering, PipeWire should be configured to launch the session manager directly. This can be accomplished by running

# mkdir -p /etc/pipewire/pipewire.conf.d
# ln -s /usr/share/examples/wireplumber/10-wireplumber.conf /etc/pipewire/pipewire.conf.d/

for system-wide configuration, or

$ : "${XDG_CONFIG_HOME:=${HOME}/.config}"
$ mkdir -p "${XDG_CONFIG_HOME}/pipewire/pipewire.conf.d"
$ ln -s /usr/share/examples/wireplumber/10-wireplumber.conf "${XDG_CONFIG_HOME}/pipewire/pipewire.conf.d/"

for per-user configuration.

PulseAudio interface

The PulseAudio interface is optional but highly recommended. Most applications cannot speak directly to PipeWire, but instead speak to PipeWire's PulseAudio interface.

If pulseaudio is installed, uninstall it and ensure pulseaudio is not running.

Modify the PipeWire configuration to launch pipewire-pulse:

# mkdir -p /etc/pipewire/pipewire.conf.d
# ln -s /usr/share/examples/pipewire/20-pipewire-pulse.conf /etc/pipewire/pipewire.conf.d/

for system configurations, or

$ : "${XDG_CONFIG_HOME:=${HOME}/.config}"
$ mkdir -p "${XDG_CONFIG_HOME}/pipewire/pipewire.conf.d"
$ ln -s /usr/share/examples/pipewire/20-pipewire-pulse.conf "${XDG_CONFIG_HOME}/pipewire/pipewire.conf.d/"

for per-user configurations.

Testing

pipewire(1) should be started as your user. To test that PipeWire works, run the pipewire command in a terminal emulator in your session:

$ pipewire

Launching pipewire should be sufficient to establish a working PipeWire session that uses wireplumber for session management.

The status of WirePlumber can be checked with:

$ wpctl status
PipeWire 'pipewire-0' [0.3.82, ...]
[...]

If the PulseAudio interface was configured, use pactl(1) (provided by the pulseaudio-utils package) to ensure it is working properly:

$ pactl info
[...]
Server Name: PulseAudio (on PipeWire 0.3.82)
[...]

Launching Automatically

Once pipewire works as expected, it can be configured to launch when starting a graphical session. There are several ways this can be achieved:

  • Use the autostarting mechanism of your desktop environment: many desktop environments have a way to configure applications and programs to start automatically.
  • Use XDG Desktop Application Autostart: many desktop environments also support the Desktop Application Autostart Specification. The pipewire package ships a Desktop Entry file for pipewire in /usr/share/applications. If your environment supports the Desktop Application Autostart, you can start pipewire by symlinking the desktop file to the system (/etc/xdg/autostart) or user (${XDG_CONFIG_HOME}/autostart or ~/.config/autostart) autostart directory. If you are using a desktop environment, window manager, or Wayland compositor that does not support this, a tool like dex(1) can be used to add support for Desktop Application Autostart, for example: dex --environment <window manager> --autostart --search-paths ~/.config/autostart.
  • Use your window manager's startup scripts: pipewire can be launched directly from your window manager or Wayland compositor's startup script.

Optional Setup

Command-line and Terminal interfaces

A variety of tools for interacting with PipeWire are included in the pipewire package, including pw-cli(1), pw-top(1), and pw-cat(1).

wpctl can be used to control the WirePlumber session manager.

If using the PulseAudio interface, PulseAudio configuration tools like pactl (from pulseaudio-utils) and ncpamixer can also be used.

Graphical interfaces

qpwgraph and helvum provide a node-and-graph-style interface for connecting applications and devices.

If using the PulseAudio interface, PulseAudio configuration tools like pavucontrol, pavucontrol-qt, and the widgets and applets integrated into many desktop environments can also be used to configure PipeWire.

Bluetooth audio

Install the libspa-bluetooth package.

ALSA integration

Install alsa-pipewire, then enable the PipeWire ALSA device and make it the default:

# mkdir -p /etc/alsa/conf.d
# ln -s /usr/share/alsa/alsa.conf.d/50-pipewire.conf /etc/alsa/conf.d
# ln -s /usr/share/alsa/alsa.conf.d/99-pipewire-default.conf /etc/alsa/conf.d

JACK interface

Install libjack-pipewire.

Use pw-jack(1) to launch JACK clients manually:

$ pw-jack <application>

Alternatively, override the library provided by libjack (see ld.so(8)). The following approach will work on glibc-based systems:

# echo "/usr/lib/pipewire-0.3/jack" > /etc/ld.so.conf.d/pipewire-jack.conf
# ldconfig

then reboot.

Troubleshooting

Common errors

[E][...] mod.rt       | [     module-rt.c:  262 pw_rtkit_bus_get()] Failed to connect to system bus: Failed to connect to socket /run/dbus/system_bus_socket: No such file or directory

This indicates the system D-Bus is not running. Enable the dbus service.

[E][...] mod.rt       | [     module-rt.c:  262 pw_rtkit_bus_get()] Failed to connect to session bus: D-Bus library appears to be incorrectly set up: see the manual page for dbus-uuidgen to correct this issue. (Failed to open "/var/lib/dbus/machine-id": No such file or directory; Failed to open "/etc/machine-id": No such file or directory)

This indicates the user session D-Bus is not running.

[E][...] mod.protocol-native | [module-protocol-:  710 init_socket_name()] server 0x55e09658e9d0: name pipewire-0 is not an absolute path and no runtime dir found. Set one of PIPEWIRE_RUNTIME_DIR, XDG_RUNTIME_DIR or USERPROFILE in the environment

This indicates XDG_RUNTIME_DIR is not set up properly.

Only a "dummy" output is found

If a session manager (like wireplumber) is not running, configure it and restart PipeWire.

If a session manager is running, check if your user is in the audio and video groups. If not using elogind, this is necessary for PipeWire to access devices.

PulseAudio

Depending on which applications you use, you might need to provide PulseAudio with a D-BUS session bus (e.g. via dbus-run-session) or a D-BUS system bus (via the dbus service).

For applications which use ALSA directly and don't support PulseAudio, the alsa-plugins-pulseaudio package can make them use PulseAudio through ALSA.

PulseAudio will automatically start when needed. If it is not starting automatically, it can be started manually by invoking pulseaudio(1) from the terminal as follows:

$ pulseaudio --daemonize=no --exit-idle-time=-1

On the other hand, PulseAudio can also end up being auto activated when it isn't desired. To inhibit this behavior, the autospawn directive from pulse-client.conf(5) can be set to no.

There are several methods of allowing PulseAudio to access to audio devices. The simplest one is to add your user to the audio group. Alternatively, you can use a session manager, like elogind.

Bluetooth

Ensure the Bluetooth controller is not blocked. Use rfkill to check whether there are any blocks and to remove soft blocks. If there is a hard block, there is likely either a physical hardware switch or an option in the BIOS to enable the Bluetooth controller.

$ rfkill
ID TYPE     DEVICE      SOFT      HARD
0 wlan      phy0   unblocked unblocked
1 bluetooth hci0     blocked unblocked

# rfkill unblock bluetooth

Installation

Install the bluez package and enable the bluetoothd and dbus services. Then, add your user to the bluetooth group and restart the dbus service, or simply reboot the system. Note that restarting the dbus service may kill processes making use of it.

To use an audio device such as a wireless speaker or headset, ALSA users need to install the bluez-alsa package. PulseAudio users do not need any additional software. PipeWire users need libspa-bluetooth.

Usage

Manage Bluetooth connections and controllers using bluetoothctl, which provides a command line interface and also accepts commands on standard input.

Consult the Arch Wiki for an example of how to pair a device.

Configuration

The main configuration file is /etc/bluetooth/main.conf.

TeX Live

In Void, the texlive-bin package provides a basic TeX installation, including the tlmgr program. Use tlmgr to install TeX packages and package collections from CTAN mirrors. Install the gnupg package to allow tlmgr to verify TeX packages.

The texlive-bin package contains the latest TeX Live version; however, earlier versions, such as texlive2018-bin, are also available.

The texlive package and texlive-* packages are also available, and provide TeX packages directly via xbps. TeX packages installed via those packages cannot interact with TeX packages installed directly from CTAN (via tlmgr). For example: pdflatex from texlive-pdflatex cannot be used to compile a TeX document that uses a package installed via tlmgr; tlmgr install pdflatex would be required for that.

Configuring TeX Live

After installing TeX Live, update the value of PATH:

$ source /etc/profile

Check that /opt/texlive/<year>/bin/x86_64-linux (or /opt/texlive/<year>/bin/i386-linux) is in your PATH:

$ echo $PATH

If required, change the global default paper size:

# tlmgr paper <letter|a4>

Installing/Updating TeX packages

To install all available packages:

# tlmgr install scheme-full

To install specific packages, you can install the collection(s) including them. To list the available collections:

$ tlmgr info collections

To see the list of files owned by a collection:

$ tlmgr info --list collection-<name>

To install the collection:

# tlmgr install collection-<name>

To install a standalone package, first check if the package exists:

$ tlmgr search --global <package>

and then install it:

# tlmgr install <package>

To find the package providing a particular file (for example, a font):

$ tlmgr search --file <filename> --global

To remove a package or a collection:

# tlmgr remove <package>

To update installed packages:

# tlmgr update --all

For a full description, check:

https://www.tug.org/texlive/doc/tlmgr.html

External Applications

Programming Languages

The Void repositories have a number of Python and Lua packages. If possible, install packages from the Void repositories or consider packaging the library or application you need. Packaging your application allows for easier system maintenance and can benefit other Void Linux users, so consider making a pull request for it. The contribution instructions can be found here.

To keep packages smaller, Void has separate devel packages for header files and development tools. If you install a library or application via a language's package manager (e.g. pip, gem), or compile one from source, you may need to install the programming language's -devel package. This is specially relevant for musl libc users, due to pre-built binaries usually targeting glibc instead.

LanguagePackage ManagerVoid Package
Python3pip, anaconda, virtualenv, etcpython3-devel
Python2pip, anaconda, virtualenv, etcpython2-devel
Rubygemruby-devel
lualuarockslua-devel

Java

Void provides LTS versions of the OpenJDK development kits and runtimes. Currently, versions 8, 11, 17, and 21 are available. To run Java-based applications, install the Java Runtime Environment of the desired version. To build Java-based programs, install the Java Development Kit of the desired version (and optionally other components listed below).

Void PackageDescription
openjdkXJava Development Kit
openjdkX-jreJava Runtime Environment
openjdkX-docDeveloper documentation
openjdkX-srcJava source code
openjdkX-jmodsJava modules
openjdkX-static-libsJava static libraries

To facilitate installing multiple Java versions in parallel, Void's OpenJDK packages use xbps-alternatives(1) to select the default JDK and JRE. Each openjdkX package provides the jdk alternative group, and each openjdkX-jre package provides the java alternative group.

These alternative groups manage the symlinks at /usr/lib/jvm/default-jdk and /usr/lib/jvm/default-jre. These are used to set the JAVA_HOME environment variables and add Java utilities to your PATH via a profile script (/etc/profile.d/jdk.sh).

When installing a Java package for the first time, you may need to re-login or run source /etc/profile.d/jdk.sh in your terminal session to update these variables.

Restricted Packages

Some packages have legal restrictions on their distribution (e.g. Discord), may be too large, or have another condition that makes it difficult for Void to distribute. These packages have build templates, but the packages themselves are not built or distributed. As such, they must be built locally. For more information see the page on restricted packages.

Non-x86_64 Architectures

The Void build system runs on x86_64 servers, both for compiling and cross-compiling packages. However, some packages (e.g. pandoc) do not support cross-compilation. These packages have to be built locally on a computer running the same architecture and libc as the system on which the package is to be used. To learn how to build packages, refer to the README for the void-packages repository.

Flatpak

Flatpak is another method for installing external proprietary applications on Linux. For information on using Flatpak with Void Linux, see the official Flatpak documentation.

If sound is not working for programs installed using Flatpak, PulseAudio auto-activation might not be working correctly. Make sure PulseAudio is running before launching the program.

Note that Flatpak's sandboxing will not necessarily protect you from any security and/or privacy-violating features of proprietary software.

Troubleshooting

Some apps may not function properly (e.g. not being able to access the host system's files). Some of these issues can be fixed by installing one or more of the xdg-user-dirs, xdg-user-dirs-gtk or xdg-utils packages, and setting up XDG Desktop Portals.

Some Flatpaks require D-Bus and/or Pulseaudio.

AppImages

An AppImage is a file that bundles an application with everything needed to run it. An AppImage can be used by making it executable and running it; installation is not required. AppImages can be run in a sandbox, such as firejail.

Some of the applications for which an AppImage is available can be found on AppImageHub.

AppImages do not yet work on musl installations.

Octave Packages

Some Octave packages require external dependencies to compile and run. For example, to build the control package, you must install the openblas-devel, libgomp-devel, libgfortran-devel, gcc-fortran, and gcc packages.

MATLAB

To use MATLAB's help browser, live scripts, add-on installer, and simulink, install the libselinux package.

Steam

Steam can be installed either via a native package, which requires enabling the "nonfree" repository, or via Flatpak. The list of dependencies for different platforms and troubleshooting information for the native package can be found in its Void-specific documentation, while this section deals with potential issues faced by Flatpak users.

If you are using a different drive to store your game library, the --filesystem option from flatpak-override(1) can prove useful.

Printing

CUPS (Common Unix Printing System) is the supported mechanism for connecting to printers on Void Linux.

As prerequisites, install the cups package and enable the cupsd service. Wait until the service is marked available.

Installing Printing Drivers

If the printer is being accessed over the network and supports PostScript or PCL, CUPS alone should be sufficient. However, additional driver packages are necessary for local printer support. The cups-filters package provides driver support for CUPS.

Depending on the hardware in question, additional drivers may be necessary.

Some CUPS drivers contain proprietary or binary-only extensions. These are available only in the nonfree repository, and sometimes only for specific architectures.

Driverless printing

Most modern network printers support printing driverlessly using the IPP Everywhere standard. See https://www.pwg.org/printers/ for a list of self-certified printers supporting this standard. Even if a printer is not on this list, there is still a high chance it is supported.

Do note that cups-filters is still required for driverless printing.

Gutenprint drivers

Gutenprint provides support for many printers. These drivers are contained in the gutenprint package.

HP drivers

Printers from Hewlett-Packard require the hplip package.

Running the following command will guide you through the driver installation process. The default configuration selections it suggests are typically sufficient.

# hp-setup -i

Brother drivers

For Brother printer support, install the foomatic drivers, which are contained in the foomatic-db and foomatic-db-nonfree packages. Support for various laser models is provided by the brother-brlaser package.

Drivers for Epson Inkjet printers

Install the epson-inkjet-printer-escpr package for Epson Inkjet printers.

Canon PIXMA/MAXIFY drivers

The cnijfilter2 package contains drivers for various Canon PIXMA and MAXIFY models. Please note that installing the driver package requires enabling the "nonfree" repository.

Configuring a New Printer

CUPS provides a web interface and command line tools that can be used to configure printers. Additionally, various native GUI options are available and may be better suited, depending on the use-case.

Automatically

Printers with support for IPP Everywhere can be discovered and configured automatically using ZeroConf. To enable this, install the avahi and nss-mdns package and enable the avahi-daemon service.

Web interface

To configure the printer using the CUPS web interface, navigate to http://localhost:631 in a browser. Under the "Administration" tab, select "Printers > Add Printer". When asked to log in, use an account that is in the lpadmin group.

Command line

The lpadmin(8) tool may be used to configure a printer using the command line.

Graphical interface

The system-config-printer package offers simple and robust configuration of new printers. Install and invoke it:

# system-config-printer

Normally this tool requires root privileges. However, if you are using PolicyKit, you can install the cups-pk-helper package to allow unprivileged users to use system-config-printer.

While system-config-printer is shown here, your desktop environment may have a native printer dialog, which may be found by consulting the documentation for your DE.

Troubleshooting

USB printer not shown

The device URI can be found manually by running:

# /usr/lib/cups/backend/usb

Containers and Virtual Machines

This section describes how to set up some of the container and virtual machine software available on Void.

Section Contents

Creating and using chroots and containers

chroots and containers can be set up and used for many purposes, including:

  • running glibc software on musl (and vice versa)
  • running software in a more controlled or sandboxed environment
  • creating a rootfs for bootstrapping a system

Chroot Creation

xvoidstrap

xvoidstrap(1) (from xtools) can be used to create the chroot:

# mkdir <chroot_dir>
# XBPS_ARCH=<chroot_arch> xvoidstrap <chroot_dir> base-container <other_pkgs>

<other_pkgs> is only needed if you want to pre-install other packages in the chroot.

Manual Creation

Alternatively, this process can be done manually.

Create a directory that will contain the chroot, then install a base system in it via the base-container package:

# mkdir -p "<chroot_dir>/var/db/xbps/keys"
# cp -a /var/db/xbps/keys/* "<chroot_dir>/var/db/xbps/keys"
# XBPS_ARCH=<chroot_arch> xbps-install -S -r <chroot_dir> -R <repository> base-container <other_pkgs>

The <repository> may vary depending on architecture.

<other_pkgs> is only needed if you want to pre-install other packages in the chroot.

Chroot Usage

xchroot

xchroot(1) (from xtools) can be used to automatically set up and enter the chroot.

Manual Method

Alternatively, this process can be done manually.

If network access is required, copy /etc/resolv.conf into the chroot; /etc/hosts may need to be copied as well.

Several directories then need to be mounted as follows:

# mount -t proc none <chroot_dir>/proc
# mount -t sysfs none <chroot_dir>/sys
# mount --rbind /dev <chroot_dir>/dev
# mount --rbind /run <chroot_dir>/run

Use chroot(1) to change to the new root, then run programs and do tasks as usual. Once finished with the chroot, unmount the chroot using umount(8). If any destructive actions are taken on the chroot directory without unmounting first, you may need to reboot to repopulate the affected directories.

Alternatives

Bubblewrap

bwrap(1) (from the bubblewrap package) has additional features like the ability for sandboxing and does not require root access.

bwrap is very flexible and can be used in many ways, for example:

$ bwrap --bind <chroot_dir> / \
	--dev /dev \
	--proc /proc \
	--bind /sys /sys \
	--bind /run /run \
	--ro-bind /etc/resolv.conf /etc/resolv.conf \
	--ro-bind /etc/passwd /etc/passwd \
	--ro-bind /etc/group /etc/group \
	<command>

In this example, you will not be able to add or edit users or groups. When running graphical applications with Xorg, you may need to also bind-mount ~/.Xauthority or other files or directories.

The bwrap(1) manpage and the Arch Wiki article contain more examples of bwrap usage.

Flatpak

Flatpak is a convenient option for running many applications, including graphical or proprietary ones, on both glibc and musl systems.

Application Containers

If a more integrated and polished solution is desired, Void also provides OCI containers that work with tools like docker and podman. These containers do not require the creation of a chroot directory before usage.

libvirt

libvirt is an API and daemon for managing platform virtualization, supporting virtualization technologies such as LXC, KVM, QEMU, Bhyve, Xen, VMWare, and Hyper-V.

To use libvirt, install the libvirt package, ensure the dbus package is installed, and enable the dbus, libvirtd, virtlockd and virtlogd services. The libvirtd daemon can be reconfigured at runtime via virt-admin(1).

The libvirt package provides the virsh(1) interface to libvirtd. virsh is an interactive shell and batch-scriptable tool for performing management tasks, including creating, configuring and running virtual machines, and managing networks and storage. Note that virsh usually needs to be run as root, as described in the virsh man page:

Most virsh commands require root privileges to run due to the communications channels used to talk to the hypervisor. Running as non root will return an error.

However, if you have the polkit and dbus packages installed and you enable the dbus service, libvirtd will grant necessary privileges to any user added to the libvirt group.

An alternative to virsh is provided by the virt-manager and virt-manager-tools packages. The default QEMU/KVM system connection requires the qemu package.

For general information on libvirt, refer to the libvirt wiki and the wiki's FAQ. For an introduction to libvirt usage, refer to the "VM lifecycle" page.

LXC

The Linux Containers project includes three subprojects: LXC, LXD and LXCFS. The project also included the CGManager project, which has been deprecated in favor of the CGroup namespace in recent kernels.

Configuring LXC

Install the lxc package.

Creating and running privileged containers as root does not require any configuration; simply use the various lxc-* commands, such as lxc-create(1), lxc-start(1), lxc-attach(1), etc.

Creating unprivileged containers

User IDs (UIDs) and group IDs (GIDs) normally range from 0 to 65535. Unprivileged containers enhance security by mapping UID and GID ranges inside each container to ranges not in use by the host system. The unused host ranges must be subordinated to the user who will be running the unprivileged containers.

Subordinate UIDs and GIDs are assigned in the subuid(5) and subgid(5) files, respectively.

To create unprivileged containers, first edit /etc/subuid and /etc/subgid to delegate ranges. For example:

root:1000000:65536
user:2000000:65536

In each colon-delimited entry:

  • the first field is the user to which a subordinate range will be assigned;
  • the second field is the smallest numeric ID defining a subordinate range; and
  • the third field is the number of consecutive IDs in the range.

The usermod(8) program may also be used to manipulate suborinated IDs.

Generally, the number of consecutive IDs should be an integer multiple of 65536; the starting value is not important, except to ensure that the various ranges defined in the file do not overlap. In this example, root controls UIDs (or, from subgid, GIDs) ranging from 1000000 to 1065535, inclusive; user controls IDs ranging from 2000000 to 2065535.

Before creating a container, the user owning the container will need an lxc.conf(5) file specifying the subuid and subgid range to use. For root-owned containers, this file resides at /etc/lxc/default.conf; for unprivileged users, the file resides at ~/.config/lxc/default.conf. Mappings are described in lines of the form

lxc.idmap = u 0 1000000 65536
lxc.idmap = g 0 1000000 65536

The isolated u character indicates a UID mapping, while the isolated g indicates a GID mapping. The first numeric value should generally always be 0; this indicates the start of the UID or GID range as seen from within the container. The second numeric value is the start of the corresponding range as seen from outside the container, and may be an arbitrary value within the range delegated in /etc/subuid or /etc/subgid. The final value is the number of consecutive IDs to map.

Note that, although the external range start is arbitrary, care must be taken to ensure that the end of the range implied by the start and number does not extend beyond the range of IDs delegated to the user.

If configuring a non-root user, edit /etc/lxc/lxc-usernet as root to specify a network device quota. For example, to allow the user named user to create up to 10 veth devices connected to the lxcbr0 bridge:

user veth lxcbr0 10

The user can now create and use unprivileged containers with the lxc-* utilities. To create a simple Void container named mycontainer, use a command similar to:

lxc-create -n mycontainer -t download -- \
	--dist voidlinux --release current --arch amd64

You may substitute another architecture for amd64, and you may specify a musl image by adding --variant musl to the end of the command. See the LXC Image Server for a list of available containers.

By default, configurations and mountpoints for system containers are stored in /var/lib/lxc, while configurations for user containers and mountpoints are stored in ~/.local/share/lxc. Both of these values can be modified by setting lxc.lxcpath in the lxc.system.conf(5) file. The superuser may launch unprivileged containers in the system lxc.lxcpath defined in /etc/lxc/lxc.conf; regular users may launch unprivileged containers in the personal lxc.lxcpath defined in ~/.config/lxc/lxc.conf.

All containers will share the same subordinate UID and GID maps by default. This is permissible, but it means that an attacker who gains elevated access within one container, and can somehow break out of that container, will have similar access to other containers. To isolate containers from each other, alter the lxc.idmap ranges in default.conf to point to a unique range before you create each container. Trying to fix permissions on a container created with the wrong map is possible, but inconvenient.

LXD

LXD provides an alternative interface to LXC's lxc-* utilities. However, it does not require the configuration described in the previous section.

Install the lxd package, and enable the lxd service.

LXD users must belong to the lxd group.

Use the lxc command to manage instances, as described here.

GnuPG

Void ships both GnuPG legacy (as gnupg1) and GnuPG stable (as gnupg).

Smartcards

For using smartcards such as Yubikeys with GnuPG, there are two backends for communicating with them through GnuPG: The internal CCID driver of GnuPG's scdaemon, or the PC/SC driver.

scdaemon with internal CCID driver

By default, scdaemon, which is required for using smartcards with GnuPG, uses its internal CCID driver. For this to work, your smartcard needs to be one of the smartcards in the udev rules here and you need to either be using elogind or be a member of the plugdev group. If these two condition are fulfilled and you don't have pcscd running, gpg --card-status should successfully print your current card status.

scdaemon with pcscd backend

If you need to use pcscd for other reasons, run echo disable-ccid >> ~/.gnupg/scdaemon.conf. Now, assuming your pcscd setup works correctly, gpg --card-status should print your card status.

OpenPGP Card Tools

As an alternative to GnuPG with smartcards, Void also ships openpgp-card-tools, a Rust based utility not reliant on GnuPG. It requires using pcscd for interacting with smart cards, so if you want to use it in parallel with GnuPG, ou need to configure scdaemon to use the pcscd backend, as described above in "scdaemon with pcscd backend".

PHP

There are two ways to install PHP packages with XBPS:

  1. Using the versioned packages (recommended).
  2. Using the meta-packages.

Versioned PHP Packages

It is generally recommended to use versioned PHP packages (e.g. php8.1, php8.1-apcu, etc.) for most use cases as this ensures a consistent environment on updates with minimal or no intervention required.

PHP Meta-packages

In Void, the php package is a meta-package that points to the latest upstream PHP version. This convention is followed by all packages prefixed with php-, such as php-fpm, as well as xdebug and composer. See the php template for a complete list. It is recommended to only use these meta-packages for development purposes.

When using a PHP meta-package, be warned that updating may require manual intervention if a new major PHP version has been added to the repository. As a part of the version change, the configuration location will change to reflect the new version. For example, upgrading from 8.0 to 8.1 would result in the configuration path changing from /etc/php8.0 to /etc/php8.1. Any customizations that have been made need to be manually applied to the new configuration directory.

php-fpm updates require special care since they include a runit service. In this case, ensure that the new runit service is started and that applications using the previous version of php-fpm can access the new php-fpm instance. In particular, make sure any applications accessing the FPM instance have the correct TCP/unix socket address.

XBPS Package Manager

The X Binary Package System (XBPS) is a fast package manager that has been designed and implemented from scratch. XBPS is managed by the Void Linux team and developed at https://github.com/void-linux/xbps.

Most general package management is done with the following commands:

  • xbps-query(1) searches for and displays information about packages installed locally, or, if used with the -R flag, packages contained in repositories.
  • xbps-install(1) installs and updates packages, and syncs repository indexes.
  • xbps-remove(1) removes installed packages, and can also remove orphaned packages and cached package files.
  • xbps-reconfigure(1) runs the configuration steps for installed packages, and can be used to reconfigure certain packages after changes in their configuration files. The latter usually requires the --force flag.
  • xbps-alternatives(1) lists or sets the alternatives provided by installed packages. Alternatives is a system which allows multiple packages to provide common functionality through otherwise conflicting files, by creating symlinks from the common paths to package-specific versions that are selected by the user.
  • xbps-pkgdb(1) can report and fix issues in the package database, as well as modify it.
  • xbps-rindex(1) manages local binary package repositories.

Most questions can be answered by consulting the man pages for these tools, together with the xbps.d(5) man page.

To learn how to build packages from source, refer to the README for the void-packages repository.

Updating

Like any other system, it is important to keep Void up-to-date. Use xbps-install(1) to update:

# xbps-install -Su

XBPS must use a separate transaction to update itself. If your update includes the xbps package, you will need to run the above command a second time to apply the rest of the updates.

Restarting Services

XBPS does not restart services when they are updated. This task is left to the administrator, so they can orchestrate maintenance windows, ensure reasonable backup capacity, and generally be present for service upgrades.

To find processes running different versions than are present on disk, use the xcheckrestart tool provided by the xtools package:

$ xcheckrestart
11339 /opt/google/chrome/chrome (deleted) (google-chrome)

xcheckrestart will print out the PID, path to the executable, status of the path that was launched (almost always deleted) and the process name.

xcheckrestart can and should be run as an unprivileged user.

Kernel Panic After Update

If you get a kernel panic after an update, it is likely your system ran out of space in /boot. Refer to "Removing old kernels" for further information.

Finding Files and Packages

To search available repositories for packages, use xbps-query(1):

$ xbps-query -Rs <search_pattern>

The -R flag specifies that repositories should be searched. Without it, -s searches for locally-installed packages.

If you can't find a file or program you expected to find after installing a package, you can use xbps-query(1) to list the files provided by that package:

$ xbps-query -f <package_name>

The xtools package contains the xlocate(1) utility. xlocate works like locate(1), but for files in the Void package repositories:

$ xlocate -S
Fetching objects: 11688, done.
From https://repo-default.voidlinux.org/xlocate/xlocate
 + e122c3634...a2659176f master     -> master  (forced update)
$ xlocate xlocate
xtools-0.59_1   /usr/bin/xlocate
xtools-0.59_1   /usr/share/man/man1/xlocate.1 -> /usr/share/man/man1/xtools.1

It is also possible to use xbps-query(1) to find files, though this is strongly discouraged:

$ xbps-query -Ro /usr/bin/xlocate
xtools-0.46_1: /usr/bin/xlocate (regular file)

This requires xbps-query to download parts of every package to find the file. xlocate, however, queries a locally cached index of all files, so no network access is required.

To get a list of all installed packages, without their version:

$ xbps-query -l | awk '{ print $2 }' | xargs -n1 xbps-uhelper getpkgname

Advanced Usage

Downgrading

XBPS allows you to downgrade a package to a specific package version.

Via xdowngrade

The easiest way to downgrade is to use xdowngrade from the xtools package, specifying the package version to which you wish to downgrade:

# xdowngrade /var/cache/xbps/pkg-1.0_1.xbps

Via XBPS

XBPS can be used to downgrade to a package version that is no longer available in the repository index.

If the package version had been installed previously, it will be available in /var/cache/xbps/. If not, it will need to be obtained from elsewhere; for the purposes of this example, it will be assumed that the package version has been added to /var/cache/xbps/.

First add the package version to your local repository:

# xbps-rindex -a /var/cache/xbps/pkg-1.0_1.xbps

Then downgrade with xbps-install:

# xbps-install -R /var/cache/xbps/ -f pkg-1.0_1

The -f flag is necessary to force downgrade/re-installation of an already installed package.

Holding packages

To prevent a package from being updated during a system update, use xbps-pkgdb(1):

# xbps-pkgdb -m hold <package>

The hold can be removed with:

# xbps-pkgdb -m unhold <package>

Repository-locking packages

If you've used xbps-src to build and install a package from a customized template, or with custom build options, you may wish to prevent system updates from replacing that package with a non-customized version. To ensure that a package is only updated from the same repository used to install it, you can repolock it via xbps-pkgdb(1):

# xbps-pkgdb -m repolock <package>

To remove the repolock:

# xbps-pkgdb -m repounlock <package>

Ignoring Packages

Sometimes you may wish to remove a package whose functionality is being provided by another package, but will be unable to do so due to dependency issues. For example, you may wish to use doas(1) instead of sudo(8), but will be unable to remove the sudo package due to it being a dependency of the base-system package. To remove it, you will need to ignore the sudo package.

To ignore a package, add an appropriate ignorepkg entry in an xbps.d(5) configuration file. For example:

ignorepkg=sudo

You will then be able to remove the sudo package using xbps-remove(1).

Virtual Packages

Virtual packages can be created with xbps.d(5) virtualpkg entries. Any request to the virtual package will be resolved to the real package. For example, to create a linux virtual package which will resolve to the linux5.6 package, create an xbps.d configuration file with the contents:

virtualpkg=linux:linux5.6

Repositories

Repositories are the heart of the XBPS package system. Repositories can be local or remote. A repository contains binary package files, which may have signatures, and a data file named $ARCH-repodata (e.g. x86_64-repodata), which may also be signed.

Note that, while local repositories do not require signatures, remote repositories must be signed.

The main repository

The locations of the main repository in relation to a base mirror URL are:

  • glibc: /current
  • musl: /current/musl
  • aarch64 and aarch64-musl: /current/aarch64

Subrepositories

In addition to the main repository, which is enabled upon installation, Void provides other official repositories maintained by the Void project, but not enabled by default:

  • nonfree: contains software packages with non-free licenses
  • multilib: contains 32-bit libraries for 64-bit systems (glibc only)
  • multilib/nonfree: contains non-free multilib packages
  • debug: contains debugging symbols for packages

These repositories can be enabled via the installation of the relevant package. These packages only install a repository configuration file in /usr/share/xbps.d.

nonfree

Void has a nonfree repository for packages that don't have free licenses. It can be enabled by installing the void-repo-nonfree package.

Packages can end up in the nonfree repository for a number of reasons:

  • Non-free licensed software with released source-code.
  • Software released only as redistributable binary packages.
  • Patented technology, which may or may not have an (otherwise) open implementation.

multilib

The multilib repository provides 32-bit packages as a compatibility layer inside a 64-bit system. It can be enabled by installing the void-repo-multilib package.

These repositories are only available for x86_64 systems running the glibc C library.

multilib/nonfree

The multilib/nonfree repository provides additional 32-bit packages which have non-free licenses. It can be enabled by installing the void-repo-multilib-nonfree package.

debug

Void Linux packages come without debugging symbols. If you want to debug software or look at a core dump you will need the debugging symbols. These packages are contained in the debug repository. It can be enabled by installing the void-repo-debug package.

Once enabled, symbols may be obtained for <package> by installing <package>-dbg.

Finding debug dependencies

The xtools package contains the xdbg(1) utility to retrieve a list of debug packages, including dependencies, for a package:

$ xdbg bash
bash-dbg
glibc-dbg
# xbps-install -S $(xdbg bash)

Mirrors

Void Linux maintains mirrors in several geographic regions for users. A fresh install will default to repo-default.voidlinux.org, which may map to any Tier 1 mirror, but you may have a better experience by selecting a different mirror.

See xmirror.voidlinux.org for more information and a list of available mirrors.

Tor Mirrors

Void Linux is also mirrored on the Tor network. See Using Tor Mirrors for more information.

Changing Mirrors

Each repository has a file defining the URL for the mirror used. For official repositories, these files are installed by the package manager in /usr/share/xbps.d, but if duplicate files are found in /etc/xbps.d, those values are used instead.

xmirror

To easily modify the currently selected mirror, xmirror(1) (from the xmirror package) can be used. This utility takes care of all steps for updating the selected mirror.

Manual Method

Alternatively, this can be done manually:

To modify mirror URLs cleanly, copy all the repository configuration files to /etc/xbps.d and change the URLs in each copied repository file.

# mkdir -p /etc/xbps.d
# cp /usr/share/xbps.d/*-repository-*.conf /etc/xbps.d/
# sed -i 's|https://repo-default.voidlinux.org|<repository>|g' /etc/xbps.d/*-repository-*.conf

After changing the URLs, you must synchronize xbps with the new mirrors:

# xbps-install -S

You should see the new repository URLs while synchronizing. You can also use xbps-query to verify the repository URLs, but only after they have been synchronized:

$ xbps-query -L
 9970 https://repo-default.voidlinux.org/current (RSA signed)
   27 https://repo-default.voidlinux.org/current/multilib/nonfree (RSA signed)
 4230 https://repo-default.voidlinux.org/current/multilib (RSA signed)
   47 https://repo-default.voidlinux.org/current/nonfree (RSA signed)
 5368 https://repo-default.voidlinux.org/current/debug (RSA signed)

Remember that repositories added afterwards will also need to be changed, or they will use the default mirror.

Using Tor Mirrors

Tor is an anonymizing software that bounces traffic via computers all around the world. It can provide access to regular sites on the internet or to hidden sites only available on the network.

The following Void Linux Mirrors are available on the Tor Network:

Using XBPS with Tor

XBPS can be made to connect to mirrors using Tor. These mirrors can be normal mirrors, via exit relays, or, for potentially greater anonymity, hidden service mirrors on the network.

XBPS respects the SOCKS_PROXY environment variable, which makes it easy to use via Tor.

Installing Tor

Tor is contained in the tor package.

After having installed Tor, you can start it as your own user:

$ tor

or enable its system service.

By default, Tor will act as a client and open a SOCKS5 proxy on TCP port 9050 on localhost.

Making XBPS connect via the SOCKS proxy

XBPS reads the SOCKS_PROXY environment variable and will use any proxy specified in it. By simply setting the variable to the address and port of the proxy opened by the Tor client, all XBPS's connections will go over the Tor network.

An example upgrading your system over Tor:

# export SOCKS_PROXY="socks5://127.0.0.1:9050"
# xbps-install -Su

Using a hidden service mirror

To use a hidden service mirror, the default mirrors need to be overwritten with configuration files pointing to .onion-addresses that are used internally on the Tor network. XBPS allows overriding repository addresses under /etc/xbps.d.

Copy your repository files from /usr/share/xbps.d to /etc/xbps.d and replace the addresses with that of an onion service (Lysator's onion used as an example):

# mkdir -p /etc/xbps.d
# cp /usr/share/xbps.d/*-repository-*.conf /etc/xbps.d/
# sed -i 's|https://repo-default.voidlinux.org|http://lysator7eknrfl47rlyxvgeamrv7ucefgrrlhk7rouv3sna25asetwid.onion/pub/voidlinux|g' /etc/xbps.d/*-repository-*.conf

Tor provides layered end-to-end encryption so HTTPS is not necessary.

When installing packages, with SOCKS_PROXY set like the earlier example, XBPS should indicate that it is synchronizing the repositories from the onion address specified in the override:

# xbps-install -S
[*] Updating `http://lysator7eknrfl47rlyxvgeamrv7ucefgrrlhk7rouv3sna25asetwid.onion/pub/voidlinux/current/aarch64/nonfree/aarch64-repodata' ...
aarch64-repodata: 4030B [avg rate: 54KB/s]
[*] Updating `http://lysator7eknrfl47rlyxvgeamrv7ucefgrrlhk7rouv3sna25asetwid.onion/pub/voidlinux/current/aarch64/aarch64-repodata' ...
aarch64-repodata: 1441KB [avg rate: 773KB/s]

Security consideration

It is advisable to set SOCKS_PROXY automatically in your environment if you are using an onion. If the setting is missing, a DNS query for the name of the hidden service will leak to the configured DNS server.

To automatically set the environment variable, add it to a file in /etc/profile.d:

# cat - <<EOF > /etc/profile.d/socksproxy.sh
#!/bin/sh
export SOCKS_PROXY="socks5://127.0.0.1:9050"
EOF

Restricted Packages

Void offers some packages that are officially maintained, but not distributed. These packages are marked as restricted and must be built from their void-packages template locally.

Packages can be restricted from distribution by either the upstream author or Void. Void reserves the right to restrict distribution of any package for effectively any reason, massive size being the most common. Another common reason is restrictive licensing that does not allow third-party redistribution of source or binary packages.

Building manually

You can use xbps-src in the void-packages repository to build the restricted packages from templates. For instructions on building packages from templates, refer to the void-packages documentation, and the "Quick start" section in particular .

Remember that the building of restricted packages must be enabled explicitly by setting XBPS_ALLOW_RESTRICTED=yes in your xbps-src configuration (in the etc/conf file in the repository.)

Automated building

There is also a tool, xbps-mini-builder which automates the process of building a list of packages. The script can be called periodically and will only rebuild packages if their templates have changed.

Custom Repositories

Void supports user-created repositories, both local and remote. This is only recommended for serving custom packages created personally, or packages from another trusted source. The Void project does not support any third-party package repositories - the use of third-party software packages poses very serious security concerns, and risks serious damage your system.

Adding custom repositories

To add a custom repository, create a file in /etc/xbps.d, with the contents:

repository=<URL>

where <URL> is either a local directory or a URL to a remote repository.

For example, to define a remote repository:

# echo 'repository=http://my.domain.com/repo' > /etc/xbps.d/my-remote-repo.conf

Remote repositories need to be signed. xbps-install(1) refuses to install packages from remote repositories if they are not signed.

To define a local repository:

# echo 'repository=/path/to/repo' > /etc/xbps.d/my-local-repo.conf

Signing Repositories

Remote repositories must be signed. Local repositories do not need to be signed.

The xbps-rindex(1) tool is used to sign repositories.

The private key for signing packages needs to be a PEM-encoded RSA key. The key can be generated with either ssh-keygen(1) or openssl(1):

$ ssh-keygen -t rsa -m PEM -f private.pem
$ openssl genrsa -out private.pem

Once the key is generated, the public part of the private key has to be added to the repository metadata. This step is required only once.

$ xbps-rindex --privkey private.pem --sign --signedby "I'm Groot" /path/to/repository/dir

Then sign one or more packages with the following command:

$ xbps-rindex --privkey private.pem --sign-pkg /path/to/repository/dir/*.xbps

Note that future packages will not be automatically signed.

Troubleshooting XBPS

Sometimes the package manager gets in a weird spot and can't fix itself without help. This section documents important fixes and things that can go wrong when working with XBPS.

Section Contents

Common Issues

Verifying RSA keys

If the Void RSA key has changed, xbps-install(1) will report the new key fingerprint and ask you to confirm it:

<repository> repository has been RSA signed by "Void Linux"
Fingerprint: <rsa_fingerprint>
Do you want to import this public key? [Y/n]

To verify the key, ensure the <rsa_fingerprint> matches one of the fingerprints in both void-packages and void-mklive.

Errors while updating or installing packages

If there are any errors while updating or installing a new package, make sure that you are using the latest version of the remote repository index. Running xbps-install(1) with the -S option will guarantee that.

"Operation not permitted"

An "Operation not permitted" error, such as:

ERROR: [reposync] failed to fetch file https://repo-default.voidlinux.org/current/nonfree/x86_64-repodata': Operation not permitted

can be caused by your system's date and/or time being incorrect. Ensure your date and time are correct.

"Not Found"

A "Not Found" error, such as:

ERROR: [reposync] failed to fetch file `https://repo-default.voidlinux.org/current/musl/x86_64-repodata': Not Found

usually means your XBPS configuration is pointing to the wrong repositories for your system. Confirm that your xbps.d(5) files refer to the correct repositories.

shlib errors

An "unresolvable shlib" error, such as:

libllvm8-8.0.1_2: broken, unresolvable shlib `libffi.so.6'

is probably due to outdated or orphan packages. To check for outdated packages, simply try to update your system. Orphan packages, on the other hand, have been removed from the Void repos, but are still installed on your system; they can be removed by running xbps-remove(1) with the -o option.

If you get an error message saying:

Transaction aborted due to unresolved shlibs

the repositories are in the staging state, which can happen due to large builds. The solution is to wait for the builds to finish. You can view the builds' progress in the Buildbot's web interface.

repodata errors

In March 2020, the compression format used for the repository data (repodata) was changed from gzip to zstd. If XBPS wasn't updated to version 0.54 (released June 2019) or newer, it is not possible to update the system with it. Unfortunately, there isn't an error message for this case, but it can be detected by running xbps-install with the -Sd flags. The debug message for this error is shown below.

[DEBUG] [repo] `//var/db/xbps/https___repo-default_voidlinux_org_current/x86_64-repodata' failed to open repodata archive Invalid or incomplete multibyte or wide character

In this situation, it is necessary to follow the steps in xbps-static.

Broken systems

If your system is for some reason broken and can't perform updates or package installations, using a statically linked version of xbps to update and install packages can help you avoid reinstalling the whole system.

Static XBPS

In rare cases, it is possible to break the system sufficiently that XBPS can no longer function. This usually happens while trying to do unsupported things with libc, but can also happen when an update contains a corrupt glibc archive or otherwise fails to unpack and configure fully.

Another issue that can present itself is in systems with a XBPS version before 0.54 (released June 2019). These systems will be impossible to update from the official repositories using the regular update procedure, due a change in the compression format used for repository data, which was made in March 2020.

In these cases it is possible to recover your system with a separate, statically compiled copy of XBPS.

Obtaining static XBPS

Statically compiled versions of XBPS are available on all mirrors in the static/ directory. The link below points to the static copies on the primary mirror in the EU:

https://repo-default.voidlinux.org/static

Download and unpack the latest version, or the version that matches the broken copy on your system (with a preference for the latest copy).

Using static XBPS

The tools in the static set are identical to the normal ones found on most systems. The only difference is that these tools are statically linked to the musl C library, and should work on systems where nothing else does. On systems that can no longer boot, it is recommended to chroot in using a Void installation medium and use the static tools from there, as it is unlikely that even a shell will work correctly on those systems. When using static XBPS with a glibc installation, the environment variable XBPS_ARCH needs to be set.

Contributing

There's more to running a distribution than just writing code.

To contribute to the Void packages repository, start by reading the CONTRIBUTING document in the void-packages GitHub repository.

To contribute to this Handbook, read CONTRIBUTING in the void-docs repository.

If you have any questions, feel free to ask them via IRC in #voidlinux on irc.libera.chat, or in the voidlinux subreddit.

Usage Statistics

If you would like to contribute usage reports, the PopCorn program reports installation statistics back to the Void project. These statistics are purely opt-in - PopCorn is not installed or enabled by default on any Void systems.

PopCorn only reports which packages are installed, their version, and the host CPU architecture (the output of xuname). This does not report which services are enabled, or any other personal information. Individual systems are tracked persistently by a random (client-generated) UUID, to ensure that each system is only counted once in each 24-hour sampling period.

The data collected by PopCorn is available to view at http://popcorn.voidlinux.org

Setting up PopCorn

First, install the PopCorn package. Then, enable the popcorn service, which will attempt to report statistics once per day.

Contributing To void-docs

The sources for this handbook are hosted in the void-docs repository on GitHub. If you would like to make a contribution, please read about the purpose of the Handbook, follow our style guide and submit a pull request.