#!/bin/sh
# Transparent huge pages as the mesh sets them on this machine (module kernel, novox/hq issue 746).
# Written by the mesh from this machine's setting transparent-huge-pages and run by
# mesh-kernel-huge-pages.service: `apply` at boot and whenever the setting changes, `restore` when the
# unit stops (the module unassigned, or the machine shutting down). Replaced on every push; change the
# setting, never this file. A change of the setting reloads the unit, which runs `apply` again: the mode
# moves straight to the new value and never passes back through the one the machine was found with.
#
#   huge-pages apply|restore [<the kernel's transparent_hugepage directory> [<this module's state>]]
#
# The two directories are arguments only so the module's tests can run this against a stand-in tree.
set -eu

want='${config:transparent-huge-pages}'
sys=/sys/kernel/mm/transparent_hugepage
state='${dir:state}'
[ $# -ge 2 ] && sys=$2
[ $# -ge 3 ] && state=$3

case "$want" in
always | madvise | never) ;;
*)
	echo "the setting transparent-huge-pages is '$want'; the kernel takes always, madvise or never" >&2
	exit 2
	;;
esac
[ -f "$sys/enabled" ] || {
	echo "this kernel has no $sys/enabled: it was built without transparent huge pages, so there is nothing to set" >&2
	exit 1
}

# The mode in force is the word the kernel brackets: "always [madvise] never". A file holding one bare
# word (the stand-in tree of the tests) is read as that word.
mode() { sed -n -e 's/.*\[\([a-z]*\)\].*/\1/p' -e t -e 's/^ *\([a-z]*\) *$/\1/p' "$sys/enabled"; }

[ $# -ge 1 ] || set -- usage
case "$1" in
apply)
	now=$(mode)
	# What the machine had before the mesh first set it, kept once, so `restore` gives it back.
	[ -s "$state/found" ] || printf '%s\n' "$now" >"$state/found"
	if [ "$now" = "$want" ]; then
		echo "transparent huge pages: $want, as set"
		exit 0
	fi
	printf '%s\n' "$want" >"$sys/enabled"
	got=$(mode)
	[ "$got" = "$want" ] || {
		echo "wrote $want to $sys/enabled and the kernel says $got" >&2
		exit 1
	}
	echo "transparent huge pages: $now -> $want"
	# A file folio already in the page cache keeps its size: the active journal files, mapped by journald
	# and by every reader, stay in 2 MiB folios until they are archived (issue 746). Rotating starts new
	# files that are mapped under the new mode. A failed rotation is said and does not fail the setting.
	# At boot nothing is in huge folios yet, and the persistent journal opens only after this unit: no rotation
	# while the service manager is still bringing the machine up.
	booting=$(systemctl is-system-running 2>/dev/null || true)
	case "$booting" in
	initializing | starting) echo "the machine is still starting: nothing to rotate yet" ;;
	*)
		if [ "$want" != always ] && command -v journalctl >/dev/null 2>&1; then
			journalctl --rotate || echo "the journal could not be rotated: its active files keep their huge pages until they are archived" >&2
		fi
		;;
	esac
	;;
restore)
	found=$(cat "$state/found" 2>/dev/null || true)
	case "$found" in
	always | madvise | never)
		printf '%s\n' "$found" >"$sys/enabled"
		echo "transparent huge pages: back to $found, as the machine had them"
		;;
	*) echo "transparent huge pages: nothing recorded as found, left at $(mode)" ;;
	esac
	# Unassigned or shut down: nothing of the module is kept; the next start records what it finds again.
	rm -f "$state/found"
	;;
*)
	echo "usage: huge-pages apply|restore [<transparent_hugepage directory> [<state directory>]]" >&2
	exit 2
	;;
esac