cidata: fix virtiofs mounts with a space in the path

.ls-lint.yml

@@ -32,6 +32,8 @@ ignore:
 - hack/common.inc.sh
 - pkg/cidata/cidata.TEMPLATE.d
 - pkg/cidata/cidata.TEMPLATE.d/util/compare_version.sh
+- pkg/cidata/cidata.TEMPLATE.d/util/escape_fstab.sh
+- pkg/cidata/cidata.TEMPLATE.d/util/unescape_fstab.sh
 - pkg/cidata/wincidata.TEMPLATE.d
 - website
 # "ubuntu-24.04" does not follow the kebab-case

pkg/cidata/cidata.TEMPLATE.d/boot.Linux/04-persistent-data-volume.sh

@@ -25,7 +25,7 @@ for DIR in ${DATADIRS}; do
 		MNTDEV="$(echo "${LINE}" | awk '{print $1}')"
 		# unmangle " \t\n\\#"
 		# https://github.com/torvalds/linux/blob/v6.6/fs/proc_namespace.c#L89
-		MNTPNT="$(echo "${LINE}" | awk '{print $2}' | sed -e 's/\\040/ /g; s/\\011/\t/g; s/\\012/\n/g; s/\\134/\\/g; s/\\043/#/g')"
+		MNTPNT="$(echo "${LINE}" | awk '{print $2}' | unescape_fstab.sh)"
 		# Ignore if MNTPNT is neither DIR nor a parent directory of DIR.
 		# It is not a parent if MNTPNT doesn't start with DIR, or the first
 		# character after DIR isn't a slash.

pkg/cidata/cidata.TEMPLATE.d/boot.Linux/05-lima-mounts.sh

@@ -10,6 +10,21 @@ if ! [[ ${LIMA_CIDATA_VMTYPE} == "vz" && ${LIMA_CIDATA_MOUNTTYPE} == "virtiofs"
 	exit 0
 fi
 
+# cloud-init's cc_mounts writes the mount point into /etc/fstab without octal-escaping it, so a
+# space/tab in the path makes an unparsable line that mount(8) silently skips via nofail.
+# cc_mounts already created the directory from the unescaped value, so just repair the fstab
+# syntax (see util/escape_fstab.sh) and (re)mount.
+# https://github.com/lima-vm/lima/issues/5136
+# https://github.com/abiosoft/colima/issues/1471
+if grep -q virtiofs /etc/fstab; then
+	escape_fstab.sh </etc/fstab >/etc/fstab.lima.tmp &&
+		cat /etc/fstab.lima.tmp >/etc/fstab && rm -f /etc/fstab.lima.tmp
+	# Mount entries cc_mounts skipped due to the previously broken line (already-mounted ones
+	# are a no-op). On Oracle Linux the virtiofs module is installed later in
+	# 30-install-packages.sh (REMOUNT_VIRTIOFS=1 remounts then), so tolerate failure here.
+	mount -t virtiofs -a || true
+fi
+
 # Update fstab entries and unmount/remount the volumes with secontext options
 # when selinux is enabled in kernel
 if [ -d /sys/fs/selinux ]; then
@@ -42,7 +57,9 @@ if [ -d /sys/fs/selinux ]; then
 			fi
 			sed -i -e "$line""s/comment=cloudconfig/comment=cloudconfig,context=\"$label\"/g" /etc/fstab
 			if [[ ${MOUNT_OPTIONS} != *"$label"* ]]; then
-				MOUNT_POINT=$(awk -v line="$line" 'NR==line {print $2}' /etc/fstab)
+				# fstab stores the mount point octal-escaped (e.g. space = "\040"); decode
+				# it before passing the path to mount(8).
+				MOUNT_POINT=$(awk -v line="$line" 'NR==line {print $2}' /etc/fstab | unescape_fstab.sh)
 				OPTIONS=$(awk -v line="$line" 'NR==line {print $4}' /etc/fstab)
 
 				#########################################################

pkg/cidata/cidata.TEMPLATE.d/util/escape_fstab.sh

@@ -0,0 +1,68 @@
+#!/bin/bash
+
+# SPDX-FileCopyrightText: Copyright The Lima Authors
+# SPDX-License-Identifier: Apache-2.0
+
+# Read an /etc/fstab on stdin and write it to stdout with the mount-point field
+# octal-escaped (per fstab(5)) for cloud-config virtiofs entries whose path
+# contains a space or tab. cloud-init's cc_mounts writes the mount point verbatim,
+# so a space/tab produces an unparsable line that mount(8) silently skips via the
+# nofail option. Fields are tab-separated, so -F'\t' isolates the mount point;
+# already-escaped paths have no literal space/tab, so the transformation is
+# idempotent (and stays correct once cloud-init escapes the field itself).
+#
+# See:
+# https://github.com/lima-vm/lima/issues/5136
+# https://github.com/abiosoft/colima/issues/1471
+# https://github.com/canonical/cloud-init/issues/3603 (cc_mounts does not escape)
+# https://github.com/canonical/cloud-init/issues/6911 (the upstream cloud-init fix)
+
+set -eu
+
+: "${SELFTEST:=}"
+if [ -n "${SELFTEST}" ]; then
+	unset SELFTEST
+	tab=$(printf '\t')
+	check() {
+		local desc=$1 input=$2 want=$3 got
+		got=$(printf '%s\n' "${input}" | "$0")
+		if [ "${got}" = "${want}" ]; then
+			echo "ok: ${desc}"
+		else
+			echo "FAIL: ${desc}" >&2
+			printf '  want: %q\n  got:  %q\n' "${want}" "${got}" >&2
+			return 1
+		fi
+	}
+	echo >&2 "=== Running tests ==="
+	check "space in the mount point is escaped" \
+		"tag${tab}/tmp/dir with spaces${tab}virtiofs${tab}rw,nofail,comment=cloudconfig${tab}0${tab}0" \
+		"tag${tab}/tmp/dir\\040with\\040spaces${tab}virtiofs${tab}rw,nofail,comment=cloudconfig${tab}0${tab}0"
+	check "already-escaped path is unchanged (idempotent)" \
+		"tag${tab}/tmp/dir\\040with\\040spaces${tab}virtiofs${tab}rw,nofail,comment=cloudconfig${tab}0${tab}0" \
+		"tag${tab}/tmp/dir\\040with\\040spaces${tab}virtiofs${tab}rw,nofail,comment=cloudconfig${tab}0${tab}0"
+	check "path without whitespace is unchanged" \
+		"tag${tab}/mnt/nospace${tab}virtiofs${tab}rw,nofail,comment=cloudconfig${tab}0${tab}0" \
+		"tag${tab}/mnt/nospace${tab}virtiofs${tab}rw,nofail,comment=cloudconfig${tab}0${tab}0"
+	check "backslash is escaped before the space" \
+		"tag${tab}/a b\\c${tab}virtiofs${tab}rw,comment=cloudconfig${tab}0${tab}0" \
+		"tag${tab}/a\\040b\\134c${tab}virtiofs${tab}rw,comment=cloudconfig${tab}0${tab}0"
+	check "entry without comment=cloudconfig is unchanged" \
+		"tag${tab}/tmp/dir with spaces${tab}virtiofs${tab}rw${tab}0${tab}0" \
+		"tag${tab}/tmp/dir with spaces${tab}virtiofs${tab}rw${tab}0${tab}0"
+	check "non-virtiofs entry is unchanged" \
+		"/dev/sda1${tab}/data dir${tab}ext4${tab}defaults,comment=cloudconfig${tab}0${tab}0" \
+		"/dev/sda1${tab}/data dir${tab}ext4${tab}defaults,comment=cloudconfig${tab}0${tab}0"
+	echo >&2 "=== All tests passed ==="
+	exit 0
+fi
+
+awk -F'\t' 'BEGIN { OFS = "\t" }
+	$3 == "virtiofs" && $4 ~ /comment=cloudconfig/ && $2 ~ /[ \t]/ {
+		p = $2
+		gsub(/\\/, "\\134", p) # backslash first so introduced escapes are not re-escaped
+		gsub(/ /, "\\040", p)
+		gsub(/\t/, "\\011", p)
+		$2 = p
+	}
+	{ print }'

pkg/cidata/cidata.TEMPLATE.d/util/unescape_fstab.sh

@@ -0,0 +1,41 @@
+#!/bin/bash
+
+# SPDX-FileCopyrightText: Copyright The Lima Authors
+# SPDX-License-Identifier: Apache-2.0
+
+# Read a string on stdin and write it to stdout with the octal escapes used in
+# /etc/fstab and /proc/mounts decoded to their literal characters:
+# "\040" -> space, "\011" -> tab, "\012" -> newline, "\134" -> backslash,
+# "\043" -> "#". The inverse of util/escape_fstab.sh.
+# https://github.com/torvalds/linux/blob/v6.6/fs/proc_namespace.c#L89
+
+set -eu
+
+: "${SELFTEST:=}"
+if [ -n "${SELFTEST}" ]; then
+	unset SELFTEST
+	tab=$(printf '\t')
+	check() {
+		local desc=$1 input=$2 want=$3 got
+		got=$(printf '%s\n' "${input}" | "$0")
+		if [ "${got}" = "${want}" ]; then
+			echo "ok: ${desc}"
+		else
+			echo "FAIL: ${desc}" >&2
+			printf '  want: %q\n  got:  %q\n' "${want}" "${got}" >&2
+			return 1
+		fi
+	}
+	echo >&2 "=== Running tests ==="
+	check "spaces are decoded" '/tmp/dir\040with\040spaces' "/tmp/dir with spaces"
+	check "tab is decoded" '/a\011b' "/a${tab}b"
+	check "newline is decoded" 'a\012b' "$(printf 'a\nb')"
+	check "backslash is decoded" '/a\134b' '/a\b'
+	check "hash is decoded" '/a\043b' "/a#b"
+	check "a string without escapes is unchanged" "/mnt/nospace" "/mnt/nospace"
+	check "mixed escapes are decoded" '/a\040b\134c' '/a b\c'
+	echo >&2 "=== All tests passed ==="
+	exit 0
+fi
+
+sed -e 's/\\040/ /g; s/\\011/\t/g; s/\\012/\n/g; s/\\134/\\/g; s/\\043/#/g'