diff options
Diffstat (limited to 'opt/limine/limine-sync.conf.example')
| -rw-r--r-- | opt/limine/limine-sync.conf.example | 108 |
1 files changed, 108 insertions, 0 deletions
diff --git a/opt/limine/limine-sync.conf.example b/opt/limine/limine-sync.conf.example new file mode 100644 index 0000000..4e66397 --- /dev/null +++ b/opt/limine/limine-sync.conf.example @@ -0,0 +1,108 @@ +# /etc/limine-sync.conf.example — copy to /etc/limine-sync.conf and edit. +# +# limine-sync generates the kernel entries of the boot menu from what is +# in /boot, and splices them into /etc/limine.conf.in at @KERNELS@. This +# file supplies the few things a directory listing cannot tell it. +# +# Everything below except CMDLINE is optional and shown at its built-in +# default; an absent file means "all defaults", which is why CMDLINE +# being required is what makes the file required. +# +# Syntax: KEY=value, one per line. Keys are uppercase. Surrounding +# quotes are stripped. '#' starts a comment anywhere on the line, so a +# value cannot contain one. An unrecognised key is warned about and +# ignored — check the output after editing. +# +# Nothing here is read at boot. Changes take effect when limine-sync +# next runs: automatically on the boot/ trigger, or by hand. Always +# look before you leap: +# +# limine-sync -n # print what would change, touch nothing +# limine-sync +# + +# ── CMDLINE (required) ─────────────────────────────────────────────── +# +# The kernel command line, used verbatim for every generated entry. +# There is no default: the root device is the one setting that differs +# on every machine, and a guess would produce a bootable-looking menu +# pointing at the wrong disk. limine-sync refuses to run without it. +# +# On an install target `setup-boot /mnt` writes this line for you, and +# reruns are how you change it afterwards. +# +# Unencrypted root: +# +#CMDLINE=root=/dev/nvme0n1p2 ro quiet +# +# LUKS root. cryptroot= is the UUID of the *LUKS partition* (from +# `cryptsetup luksUUID /dev/nvme0n1p2`), not of the filesystem inside +# it. tinyrd creates the mapping under exactly the name given in +# root=, so the name you passed to `cryptsetup open` belongs here and +# is recorded nowhere else: +# +#CMDLINE=cryptroot=UUID=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx root=/dev/mapper/croot ro quiet +# +# Serial console as well as the screen (VPS, or `qemu -nographic`): +# +#CMDLINE=root=/dev/vda2 ro quiet console=tty0 console=ttyS0,115200n8 + +# ── TITLE ──────────────────────────────────────────────────────────── +# +# Menu label. Entries read "<TITLE> <version>", e.g. "rawnix 6.18.50", +# and the Xen chainload entry reads "<TITLE> Xen". +# +#TITLE=rawnix + +# ── INITRD ─────────────────────────────────────────────────────────── +# +# Filename in /boot of the initramfs, added as a module_path after the +# microcode. tinyrd carries no modules and takes the LUKS UUID from +# the cmdline, so one image serves every kernel and this is a plain +# name with no version in it. +# +# Set it empty to boot with no initramfs. That is fine for an +# unencrypted root and impossible for a LUKS one — limine-sync warns if +# CMDLINE has cryptroot= and the image is not on the ESP. +# +#INITRD=tinyrd.img + +# ── MICROCODE ──────────────────────────────────────────────────────── +# +# Filename in /boot of the early microcode image, added as the *first* +# module_path. The kernel only applies an early update from the head +# of the initrd chain, so the order is not adjustable. +# +# Set empty on a machine that needs none, or on Intel once there is an +# intel-ucode port. An entry is only emitted if the file is actually +# on the ESP. +# +#MICROCODE=amd-ucode.img + +# ── Xen ────────────────────────────────────────────────────────────── +# +# Only relevant if a xen-*.efi is on the ESP. The newest one always +# gets a chainload entry in limine.conf; these keys additionally +# generate /boot/efi/xen.cfg, which is what picks dom0's kernel. +# +# XEN_KERNEL pins a *series*, not a version: "6.12" selects the newest +# 6.12.x that has a module tree. Empty means leave xen.cfg alone, +# which is the default and what you want if you maintain it by hand. +# +#XEN_KERNEL= +# +# Appended to dom0's kernel line. Note dom0 usually wants fewer CPUs +# and less memory than the cmdline above gives a bare-metal boot. +# +#XEN_CMDLINE= +# +# The hypervisor's own options line. +# +#XEN_OPTIONS=dom0_mem=4G,max:4G dom0_max_vcpus=4 +# +# Xen takes a single ramdisk, so microcode and an initramfs are +# mutually exclusive here. yes = microcode; anything else = the +# INITRD image, if there is one. An encrypted dom0 root therefore +# cannot get early microcode this way and needs Xen's own ucode=. +# +#XEN_UCODE=no |