Files
ProxMenux/scripts/global/vm_storage_helpers.sh
2026-04-06 19:13:31 +02:00

403 lines
12 KiB
Bash

#!/usr/bin/env bash
if [[ -n "${__PROXMENUX_VM_STORAGE_HELPERS__}" ]]; then
return 0
fi
__PROXMENUX_VM_STORAGE_HELPERS__=1
function _array_contains() {
local needle="$1"
shift
local item
for item in "$@"; do
[[ "$item" == "$needle" ]] && return 0
done
return 1
}
function _refresh_host_storage_cache() {
MOUNTED_DISKS=$(lsblk -ln -o NAME,MOUNTPOINT | awk '$2!="" {print "/dev/" $1}')
SWAP_DISKS=$(swapon --noheadings --raw --show=NAME 2>/dev/null)
LVM_DEVICES=$(pvs --noheadings -o pv_name 2> >(grep -v 'File descriptor .* leaked') | xargs -r -n1 readlink -f | sort -u)
CONFIG_DATA=$(grep -vE '^\s*#' /etc/pve/qemu-server/*.conf /etc/pve/lxc/*.conf 2>/dev/null)
ZFS_DISKS=""
local zfs_raw entry path base_disk
zfs_raw=$(zpool list -v -H 2>/dev/null | awk '{print $1}' | grep -v '^NAME$' | grep -v '^-' | grep -v '^mirror')
for entry in $zfs_raw; do
path=""
if [[ "$entry" == wwn-* || "$entry" == ata-* ]]; then
[[ -e "/dev/disk/by-id/$entry" ]] && path=$(readlink -f "/dev/disk/by-id/$entry")
elif [[ "$entry" == /dev/* ]]; then
path="$entry"
fi
if [[ -n "$path" ]]; then
base_disk=$(lsblk -no PKNAME "$path" 2>/dev/null)
[[ -n "$base_disk" ]] && ZFS_DISKS+="/dev/$base_disk"$'\n'
fi
done
ZFS_DISKS=$(echo "$ZFS_DISKS" | sort -u)
}
function _disk_is_host_system_used() {
local disk="$1"
local disk_real part fstype part_path
DISK_USAGE_REASON=""
while read -r part fstype; do
[[ -z "$part" ]] && continue
part_path="/dev/$part"
if grep -qFx "$part_path" <<< "$MOUNTED_DISKS"; then
DISK_USAGE_REASON="$(translate "Mounted filesystem detected") ($part_path)"
return 0
fi
if grep -qFx "$part_path" <<< "$SWAP_DISKS"; then
DISK_USAGE_REASON="$(translate "Swap partition detected") ($part_path)"
return 0
fi
case "$fstype" in
zfs_member)
DISK_USAGE_REASON="$(translate "ZFS member detected") ($part_path)"
return 0
;;
linux_raid_member)
DISK_USAGE_REASON="$(translate "RAID member detected") ($part_path)"
return 0
;;
LVM2_member)
DISK_USAGE_REASON="$(translate "LVM physical volume detected") ($part_path)"
return 0
;;
esac
done < <(lsblk -ln -o NAME,FSTYPE "$disk" 2>/dev/null)
disk_real=$(readlink -f "$disk" 2>/dev/null)
if [[ -n "$disk_real" && -n "$LVM_DEVICES" ]] && grep -qFx "$disk_real" <<< "$LVM_DEVICES"; then
DISK_USAGE_REASON="$(translate "Disk is part of host LVM")"
return 0
fi
if [[ -n "$ZFS_DISKS" && "$ZFS_DISKS" == *"$disk"* ]]; then
DISK_USAGE_REASON="$(translate "Disk is part of a host ZFS pool")"
return 0
fi
return 1
}
function _disk_used_in_guest_configs() {
local disk="$1"
local real_path
real_path=$(readlink -f "$disk" 2>/dev/null)
if [[ -n "$real_path" ]] && grep -Fq "$real_path" <<< "$CONFIG_DATA"; then
return 0
fi
local symlink
for symlink in /dev/disk/by-id/*; do
[[ -e "$symlink" ]] || continue
if [[ "$(readlink -f "$symlink")" == "$real_path" ]] && grep -Fq "$symlink" <<< "$CONFIG_DATA"; then
return 0
fi
done
return 1
}
function _controller_block_devices() {
local pci_full="$1"
local pci_root="/sys/bus/pci/devices/$pci_full"
[[ -d "$pci_root" ]] || return 0
local sys_block dev_name cur base
# Walk /sys/block and resolve each block device back to its ancestor PCI device.
# This avoids unbounded recursive scans while still handling NVMe/SATA paths.
for sys_block in /sys/block/*; do
[[ -e "$sys_block/device" ]] || continue
dev_name=$(basename "$sys_block")
[[ -b "/dev/$dev_name" ]] || continue
cur=$(readlink -f "$sys_block/device" 2>/dev/null)
[[ -n "$cur" ]] || continue
while [[ "$cur" != "/" ]]; do
base=$(basename "$cur")
if [[ "$base" == "$pci_full" ]]; then
echo "/dev/$dev_name"
break
fi
cur=$(dirname "$cur")
done
done
}
function _vm_is_q35() {
local vmid="$1"
local machine_line
machine_line=$(qm config "$vmid" 2>/dev/null | awk -F': ' '/^machine:/ {print $2}')
[[ "$machine_line" == *q35* ]]
}
function _vm_storage_enable_iommu_cmdline() {
local cpu_vendor iommu_param
cpu_vendor=$(grep -m1 "vendor_id" /proc/cpuinfo 2>/dev/null | awk '{print $3}')
if [[ "$cpu_vendor" == "GenuineIntel" ]]; then
iommu_param="intel_iommu=on"
elif [[ "$cpu_vendor" == "AuthenticAMD" ]]; then
iommu_param="amd_iommu=on"
else
return 1
fi
local cmdline_file="/etc/kernel/cmdline"
local grub_file="/etc/default/grub"
if [[ -f "$cmdline_file" ]] && grep -qE 'root=ZFS=|root=ZFS/' "$cmdline_file" 2>/dev/null; then
if ! grep -q "$iommu_param" "$cmdline_file"; then
cp "$cmdline_file" "${cmdline_file}.bak.$(date +%Y%m%d_%H%M%S)"
sed -i "s|\\s*$| ${iommu_param} iommu=pt|" "$cmdline_file"
proxmox-boot-tool refresh >/dev/null 2>&1 || true
fi
elif [[ -f "$grub_file" ]]; then
if ! grep -q "$iommu_param" "$grub_file"; then
cp "$grub_file" "${grub_file}.bak.$(date +%Y%m%d_%H%M%S)"
sed -i "/GRUB_CMDLINE_LINUX_DEFAULT=/ s|\"$| ${iommu_param} iommu=pt\"|" "$grub_file"
update-grub >/dev/null 2>&1 || true
fi
else
return 1
fi
return 0
}
function _vm_storage_ensure_iommu_or_offer() {
local reboot_policy="${VM_STORAGE_IOMMU_REBOOT_POLICY:-ask_now}"
if declare -F _pci_is_iommu_active >/dev/null 2>&1 && _pci_is_iommu_active; then
return 0
fi
if grep -qE 'intel_iommu=on|amd_iommu=on' /proc/cmdline 2>/dev/null && \
[[ -d /sys/kernel/iommu_groups ]] && \
[[ -n "$(ls /sys/kernel/iommu_groups/ 2>/dev/null)" ]]; then
return 0
fi
# Wizard flow: if IOMMU was already configured in this run and reboot is pending,
# allow the user to continue planning storage selections without re-prompting.
if [[ "$reboot_policy" == "defer" && "${VM_STORAGE_IOMMU_PENDING_REBOOT:-0}" == "1" ]]; then
return 0
fi
local prompt
prompt="$(translate "IOMMU is not active on this system.")\n\n"
prompt+="$(translate "Controller/NVMe passthrough to VMs requires IOMMU enabled in BIOS/UEFI and kernel.")\n\n"
prompt+="$(translate "Do you want to enable IOMMU now?")\n\n"
prompt+="$(translate "A host reboot is required after this change.")"
whiptail --title "IOMMU Required" --yesno "$prompt" 14 78
[[ $? -ne 0 ]] && return 1
if ! _vm_storage_enable_iommu_cmdline; then
whiptail --title "IOMMU" --msgbox \
"$(translate "Failed to configure IOMMU automatically.")\n\n$(translate "Please configure it manually and reboot.")" \
10 72
return 1
fi
if [[ "$reboot_policy" == "defer" ]]; then
VM_STORAGE_IOMMU_PENDING_REBOOT=1
export VM_STORAGE_IOMMU_PENDING_REBOOT
whiptail --title "Reboot Required" --msgbox \
"$(translate "IOMMU configured successfully.")\n\n$(translate "Continue the VM wizard and reboot the host at the end.")\n\n$(translate "You can now select Controller/NVMe devices in Storage Plan.")\n$(translate "Device assignments will be written now and become active after reboot.")" \
12 78
return 0
fi
if whiptail --title "Reboot Required" --yesno \
"$(translate "IOMMU configured successfully.")\n\n$(translate "Do you want to reboot now?")" 10 68; then
reboot
else
whiptail --title "Reboot Required" --msgbox \
"$(translate "Please reboot manually and run the passthrough step again.")" 9 68
fi
return 1
}
function _vm_storage_confirm_controller_passthrough_risk() {
local vmid="${1:-}"
local vm_name="${2:-}"
local title="${3:-Controller + NVMe}"
local vm_label=""
if [[ -n "$vmid" ]]; then
vm_label="$vmid"
[[ -n "$vm_name" ]] && vm_label="${vm_label} (${vm_name})"
fi
local msg
msg="$(translate "Important compatibility notice")\n\n"
msg+="$(translate "Not all motherboards support physical Controller/NVMe passthrough to VMs reliably, especially systems with old platforms or limited BIOS/UEFI firmware.")\n\n"
msg+="$(translate "On some systems, the VM may fail to start or the host may freeze when the VM boots.")\n\n"
local reinforce_limited_firmware="no"
local bios_date bios_year current_year cpu_model
bios_date=$(cat /sys/class/dmi/id/bios_date 2>/dev/null)
bios_year=$(echo "$bios_date" | grep -oE '[0-9]{4}' | tail -n1)
current_year=$(date +%Y 2>/dev/null)
if [[ -n "$bios_year" && -n "$current_year" ]]; then
if (( current_year - bios_year >= 7 )); then
reinforce_limited_firmware="yes"
fi
fi
cpu_model=$(grep -m1 'model name' /proc/cpuinfo 2>/dev/null | cut -d: -f2- | xargs)
if echo "$cpu_model" | grep -qiE 'J4[0-9]{3}|J3[0-9]{3}|N4[0-9]{3}|N3[0-9]{3}|Apollo Lake'; then
reinforce_limited_firmware="yes"
fi
if [[ "$reinforce_limited_firmware" == "yes" ]]; then
msg+="$(translate "Detected risk factor: this host may use an older or limited firmware platform, which increases passthrough instability risk.")\n\n"
fi
if [[ -n "$vm_label" ]]; then
msg+="$(translate "Target VM"): ${vm_label}\n\n"
fi
msg+="$(translate "If this happens after assignment"):\n"
msg+=" - $(translate "Power cycle the host if it is frozen.")\n"
msg+=" - $(translate "Remove the hostpci controller/NVMe entries from the VM config file.")\n"
msg+=" /etc/pve/qemu-server/${vmid:-<VMID>}.conf\n"
msg+=" - $(translate "Start the VM again without that passthrough device.")\n\n"
msg+="$(translate "Do you want to continue with this assignment?")"
whiptail --title "$title" --yesno "$msg" 21 96
}
function _shorten_text() {
local text="$1"
local max_len="${2:-42}"
[[ -z "$text" ]] && { echo ""; return; }
if (( ${#text} > max_len )); then
echo "${text:0:$((max_len-3))}..."
else
echo "$text"
fi
}
function _pci_slot_base() {
local pci_full="$1"
local slot
slot="${pci_full#0000:}"
slot="${slot%.*}"
echo "$slot"
}
function _vm_status_is_running() {
local vmid="$1"
qm status "$vmid" 2>/dev/null | grep -q "status: running"
}
function _vm_onboot_is_enabled() {
local vmid="$1"
qm config "$vmid" 2>/dev/null | grep -qE '^onboot:\s*1'
}
function _vm_name_by_id() {
local vmid="$1"
local conf="/etc/pve/qemu-server/${vmid}.conf"
local vm_name
vm_name=$(awk '/^name:/ {print $2}' "$conf" 2>/dev/null)
[[ -z "$vm_name" ]] && vm_name="VM-${vmid}"
echo "$vm_name"
}
function _vm_has_pci_slot() {
local vmid="$1"
local slot_base="$2"
local conf="/etc/pve/qemu-server/${vmid}.conf"
[[ -f "$conf" ]] || return 1
grep -qE "^hostpci[0-9]+:.*(0000:)?${slot_base}(\\.[0-7])?([,[:space:]]|$)" "$conf"
}
function _pci_assigned_vm_ids() {
local pci_full="$1"
local exclude_vmid="${2:-}"
local slot_base conf vmid
slot_base=$(_pci_slot_base "$pci_full")
for conf in /etc/pve/qemu-server/*.conf; do
[[ -f "$conf" ]] || continue
vmid=$(basename "$conf" .conf)
[[ -n "$exclude_vmid" && "$vmid" == "$exclude_vmid" ]] && continue
if grep -qE "^hostpci[0-9]+:.*(0000:)?${slot_base}(\\.[0-7])?([,[:space:]]|$)" "$conf"; then
echo "$vmid"
fi
done
}
function _remove_pci_slot_from_vm_config() {
local vmid="$1"
local slot_base="$2"
local conf="/etc/pve/qemu-server/${vmid}.conf"
[[ -f "$conf" ]] || return 1
local tmpf
tmpf=$(mktemp)
awk -v slot="$slot_base" '
$0 ~ "^hostpci[0-9]+:.*(0000:)?" slot "(\\.[0-7])?([,[:space:]]|$)" {next}
{print}
' "$conf" > "$tmpf" && cat "$tmpf" > "$conf"
rm -f "$tmpf"
}
function _pci_assigned_vm_summary() {
local pci_full="$1"
local slot_base conf vmid vm_name running onboot
local -a refs=()
local running_count=0 onboot_count=0
slot_base="${pci_full#0000:}"
slot_base="${slot_base%.*}"
for conf in /etc/pve/qemu-server/*.conf; do
[[ -f "$conf" ]] || continue
if ! grep -qE "^hostpci[0-9]+:.*(0000:)?${slot_base}(\\.[0-7])?([,[:space:]]|$)" "$conf"; then
continue
fi
vmid=$(basename "$conf" .conf)
vm_name=$(awk '/^name:/ {print $2}' "$conf" 2>/dev/null)
[[ -z "$vm_name" ]] && vm_name="VM-${vmid}"
if qm status "$vmid" 2>/dev/null | grep -q "status: running"; then
running="running"
running_count=$((running_count + 1))
else
running="stopped"
fi
if grep -qE "^onboot:\s*1" "$conf" 2>/dev/null; then
onboot="1"
onboot_count=$((onboot_count + 1))
else
onboot="0"
fi
refs+=("${vmid}[${running},onboot=${onboot}]")
done
[[ ${#refs[@]} -eq 0 ]] && return 1
local joined summary
joined=$(IFS=', '; echo "${refs[*]}")
summary="$(translate "Assigned to VM(s)"): ${joined}"
if [[ "$running_count" -gt 0 ]]; then
summary+=" ($(translate "running"): ${running_count})"
fi
if [[ "$onboot_count" -gt 0 ]]; then
summary+=", onboot=1: ${onboot_count}"
fi
echo "$summary"
return 0
}