Amend Dracut module to export ZFS root on shutdown

Make use of Dracut's ability to restore the initramfs on shutdown and
pivot to it, allowing for a clean unmount and export of the ZFS root.
No need to force-import on every reboot anymore.

Signed-off-by: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #2195
Issue #2476
Issue #2498
Issue #2556
Issue #2563
Issue #2575
Issue #2600
Issue #2755
Issue #2766

dracut/90zfs/.gitignore

@@ -1,3 +1,4 @@
+export-zfs.sh
 module-setup.sh
 mount-zfs.sh
 parse-zfs.sh

dracut/90zfs/Makefile.am

@@ -1,10 +1,12 @@
 pkgdracutdir = $(dracutdir)/modules.d/90zfs
 pkgdracut_SCRIPTS = \
+	$(top_srcdir)/dracut/90zfs/export-zfs.sh \
 	$(top_srcdir)/dracut/90zfs/module-setup.sh \
 	$(top_srcdir)/dracut/90zfs/mount-zfs.sh \
 	$(top_srcdir)/dracut/90zfs/parse-zfs.sh
 
 EXTRA_DIST = \
+	$(top_srcdir)/dracut/90zfs/export-zfs.sh.in \
 	$(top_srcdir)/dracut/90zfs/module-setup.sh.in \
 	$(top_srcdir)/dracut/90zfs/mount-zfs.sh.in \
 	$(top_srcdir)/dracut/90zfs/parse-zfs.sh.in

dracut/90zfs/export-zfs.sh.in

@@ -0,0 +1,29 @@
+#!/bin/sh
+
+_do_zpool_export() {
+	local ret=0
+	local final=$1
+	local force
+
+	if [ "x$final" != "x" ]; then
+		force="-f"
+	fi
+
+	info "Exporting ZFS storage pools"
+	zpool list -H | while read fs rest ; do
+		zpool export $force "$fs" || ret=$?
+	done
+
+	if [ "x$final" != "x" ]; then
+		info "zpool list"
+		zpool list 2>&1 | vinfo
+	fi
+
+	return $ret
+}
+
+if command -v zpool >/dev/null; then
+	_do_zpool_export $1
+else
+	:
+fi

dracut/90zfs/module-setup.sh.in

@@ -39,6 +39,7 @@ install() {
 	dracut_install hostid
 	inst_hook cmdline 95 "$moddir/parse-zfs.sh"
 	inst_hook mount 98 "$moddir/mount-zfs.sh"
+	inst_hook shutdown 30 "$moddir/export-zfs.sh"
 
 	if [ -e @sysconfdir@/zfs/zpool.cache ]; then
 		inst @sysconfdir@/zfs/zpool.cache

dracut/90zfs/mount-zfs.sh.in

@@ -67,5 +67,7 @@ case "$root" in
 		else
 			mount -o zfsutil -t zfs "$zfsbootfs" "$NEWROOT" && ROOTFS_MOUNTED=yes
 		fi
+
+		need_shutdown
 		;;
 esac

dracut/README.dracut.markdown

@@ -77,6 +77,9 @@ command line and determine if ZFS is the active root filesystem.
 * `mount-zfs.sh`: Run later in initramfs boot process after udev has settled
 to mount the root dataset.
 
+* `export-zfs.sh`: Run on shutdown after dracut has restored the initramfs
+and pivoted to it, allowing for a clean unmount and export of the ZFS root.
+
 `module-setup.sh`
 ---------------
 
@@ -164,3 +167,25 @@ import can lead to serious data corruption and loss of pools, so this option
 should be used with extreme caution.  Note that even with this flag set, if
 the required zpool was auto-imported by the kernel module, no additional
 `zpool import` commands are run, so nothing is forced.
+
+`export-zfs.sh`
+-------------
+
+Normally the zpool containing the root dataset cannot be exported on
+shutdown as it is still in use by the init process. To work around this,
+Dracut is able to restore the initramfs on shutdown and pivot to it.
+All remaining process are then running from a ramdisk, allowing for a
+clean unmount and export of the ZFS root. The theory of operation is
+described in detail in the [Dracut manual](https://www.kernel.org/pub/linux/utils/boot/dracut/dracut.html#_dracut_on_shutdown).
+
+This script will try to export all remaining zpools after Dracut has
+pivoted to the initramfs. If an initial regular export is not successful,
+Dracut will call this script once more with the `final` option,
+in which case a forceful export is attempted.
+
+Other Dracut modules include similar shutdown scripts and Dracut
+invokes these scripts round-robin until they succeed. In particular,
+the `90dm` module installs a script which tries to close and remove
+all device mapper targets. Thus, if there are ZVOLs containing
+dm-crypt volumes or if the zpool itself is backed by a dm-crypt
+volume, the shutdown scripts will try to untangle this.