Use a different technique to detect whether to mount-zfs

The behavior of the Dracut module was very wrong before.

The correct behavior: initramfs should not run `zfs-mount` to completion
if the two generator files exist.  If, however, one of them is missing,
it indicates one of three cases:

* The kernel command line did not specify a root ZFS file system, and
  another Dracut module is already handling root mount (via systemd).
  `mount-zfs` can run, but it will do nothing.
* There is no systemd to run `sysroot.mount` to begin with.
  `mount-zfs` must run.
* The root parameter is zfs:AUTO, which cannot be run in sysroot.mount.
  `mount-zfs` must run.

In any of these three cases, it is safe to run `zfs-mount` to completion.

`zfs-mount` must also delete itself if it determines it should not run,
or else Dracut will do the insane thing of running it over and over again.
Literally, the definition of insanity, doing the same thing that did not
work before, expecting different results.  Doing that may have had a great
result before, when we had a race between devices appearing and pools
being mounted, and `mount-zfs` was tasked with the full responsibility
of importing the needed pool, but nowadays it is wrong behavior and
should be suppressed.

I deduced that self-deletion was the correct thing to do by looking at
other Dracut code, because (as we all are very fully aware of) Dracut
is entirely, ahem, "implementation-defined".

Tested-by: @wphilips 
Reviewed-by: Brian Behlendorf <[email protected]>
Signed-off-by: Manuel Amador (Rudd-O) <[email protected]>
Closes #5157 
Closes #5204 

2 files changed, 52 insertions, 19 deletions

diff --git a/contrib/dracut/90zfs/mount-zfs.sh.in b/contrib/dracut/90zfs/mount-zfs.sh.in
index 51e107bb3..3b6be314c 100755
--- a/contrib/dracut/90zfs/mount-zfs.sh.in
+++ b/contrib/dracut/90zfs/mount-zfs.sh.in
@@ -10,17 +10,29 @@ case "${root}" in
 	*) return ;;
 esac
 
-if command -v systemctl >/dev/null; then
-	# If sysroot.mount exists, the initial RAM disk configured
-	# it to mount ZFS on root.  In that case, we bail early.
-	loadstate="$(systemctl --system --show -p LoadState sysroot.mount || true)"
-	if [ "${loadstate}" = "LoadState=not-found" -o "${loadstate}" = "" ] ; then
-		info "ZFS: sysroot.mount absent, mounting root with mount-zfs.sh"
-	else
-		info "ZFS: sysroot.mount present, delegating root mount to it"
-		return
-	fi
+GENERATOR_FILE=/run/systemd/generator/sysroot.mount
+GENERATOR_EXTENSION=/run/systemd/generator/sysroot.mount.d/zfs-enhancement.conf
+
+if [ -e "$GENERATOR_FILE" -a -e "$GENERATOR_EXTENSION" ] ; then
+	# If the ZFS sysroot.mount flag exists, the initial RAM disk configured
+	# it to mount ZFS on root.  In that case, we bail early.  This flag
+	# file gets created by the zfs-generator program upon successful run.
+	info "ZFS: There is a sysroot.mount and zfs-generator has extended it."
+	info "ZFS: Delegating root mount to sysroot.mount."
+	# Let us tell the initrd to run on shutdown.
+	# We have a shutdown hook to run
+	# because we imported the pool.
+	need_shutdown
+	# We now prevent Dracut from running this thing again.
+	for zfsmounthook in "$hookdir"/mount/*zfs* ; do
+		if [ -f "$zfsmounthook" ] ; then
+			rm -f "$zfsmounthook"
+		fi
+	done
+	return
 fi
+info "ZFS: No sysroot.mount exists or zfs-generator did not extend it."
+info "ZFS: Mounting root with the traditional mount-zfs.sh instead."
 
 # Delay until all required block devices are present.
 udevadm settle
@@ -45,6 +57,10 @@ ZFS_DATASET="${ZFS_DATASET:-${root#zfs:}}"
 ZFS_POOL="${ZFS_DATASET%%/*}"
 
 if import_pool "${ZFS_POOL}" ; then
+	# Let us tell the initrd to run on shutdown.
+	# We have a shutdown hook to run
+	# because we imported the pool.
+	need_shutdown
 	info "ZFS: Mounting dataset ${ZFS_DATASET}..."
 	if mount_dataset "${ZFS_DATASET}" ; then
 		ROOTFS_MOUNTED=yes
@@ -53,4 +69,3 @@ if import_pool "${ZFS_POOL}" ; then
 fi
 
 rootok=0
-need_shutdown
diff --git a/contrib/dracut/90zfs/zfs-generator.sh.in b/contrib/dracut/90zfs/zfs-generator.sh.in
index aa12fb856..c6384f583 100755
--- a/contrib/dracut/90zfs/zfs-generator.sh.in
+++ b/contrib/dracut/90zfs/zfs-generator.sh.in
@@ -1,21 +1,32 @@
 #!/bin/bash
 
+echo "zfs-generator: starting" >> /dev/kmsg
+
 GENERATOR_DIR="$1"
-[ -z "$GENERATOR_DIR" ] && exit 1
+[ -n "$GENERATOR_DIR" ] || {
+    echo "zfs-generator: no generator directory specified, exiting" >> /dev/kmsg
+    exit 1
+}
 
 [ -f /lib/dracut-lib.sh ] && dracutlib=/lib/dracut-lib.sh
 [ -f /usr/lib/dracut/modules.d/99base/dracut-lib.sh ] && dracutlib=/usr/lib/dracut/modules.d/99base/dracut-lib.sh
 
-type getarg >/dev/null 2>&1 || . "$dracutlib"
+type getarg >/dev/null 2>&1 || {
+    echo "zfs-generator: loading Dracut library from $dracutlib" >> /dev/kmsg
+    . "$dracutlib"
+}
 
 [ -z "$root" ]       && root=$(getarg root=)
 [ -z "$rootfstype" ] && rootfstype=$(getarg rootfstype=)
 [ -z "$rootflags" ]  && rootflags=$(getarg rootflags=)
 
+# If root is not ZFS= or zfs: or rootfstype is not zfs
+# then we are not supposed to handle it.
 [ "${root##zfs:}" = "${root}" -a "${root##ZFS=}" = "${root}" -a "$rootfstype" != "zfs" ] && exit 0
-# If root is set to zfs:AUTO, then we know sysroot.mount will not be generated
-# so we have no need to enhance it.
-# See https://github.com/zfsonlinux/zfs/pull/4558#discussion_r61118952 for details.
+# If root is set to zfs:AUTO, then we are also not
+# supposed to handle it, and it should be handled
+# by the traditional Dracut mount hook.
+# See https://github.com/zfsonlinux/zfs/pull/4558#discussion_r61118952
 if [ "${root}" = "zfs:AUTO" ] ; then
   exit 0
 fi
@@ -32,16 +43,23 @@ fi
 root="${root##zfs:}"
 root="${root##ZFS=}"
 
+echo "zfs-generator: writing extension for sysroot.mount to $GENERATOR_DIR"/sysroot.mount.d/zfs-enhancement.conf >> /dev/kmsg
+
 [ -d "$GENERATOR_DIR" ] || mkdir "$GENERATOR_DIR"
-[ -d "$GENERATOR_DIR/sysroot.mount.d" ] || mkdir "$GENERATOR_DIR/sysroot.mount.d"
+[ -d "$GENERATOR_DIR"/sysroot.mount.d ] || mkdir "$GENERATOR_DIR"/sysroot.mount.d
 
 {
     echo "[Unit]"
+    echo "Before=initrd-root-fs.target"
     echo "After=zfs-import-scan.service"
     echo "After=zfs-import-cache.service"
-    echo ""
     echo "[Mount]"
     echo "What=${root}"
     echo "Type=${rootfstype}"
     echo "Options=${rootflags}"
-} > "$GENERATOR_DIR/sysroot.mount.d/zfs-enhancement.conf"
+} > "$GENERATOR_DIR"/sysroot.mount.d/zfs-enhancement.conf
+
+[ -d "$GENERATOR_DIR"/initrd-root-fs.target.requires ] || mkdir -p "$GENERATOR_DIR"/initrd-root-fs.target.requires
+ln -s ../sysroot.mount "$GENERATOR_DIR"/initrd-root-fs.target.requires/sysroot.mount
+
+echo "zfs-generator: finished" >> /dev/kmsg