ext4: add test for all ext4/ext3/ext2 mount options
authorLukas Czerner <lczerner@redhat.com>
Mon, 8 Nov 2021 15:19:16 +0000 (16:19 +0100)
committerEryu Guan <guaneryu@gmail.com>
Sun, 14 Nov 2021 10:57:39 +0000 (18:57 +0800)
Add test to validate that all the ext4, ext3 and ext2 are properly
recognized, validate and applied to avoid regressions as ext4 moves to
the new mount API.

Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
tests/ext4/053 [new file with mode: 0755]
tests/ext4/053.out [new file with mode: 0644]

diff --git a/tests/ext4/053 b/tests/ext4/053
new file mode 100755 (executable)
index 0000000..eaa0250
--- /dev/null
@@ -0,0 +1,633 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (c) 2021 Red Hat, Inc., Lukas Czerner <lczerner@redhat.com>.
+#
+# FS QA Test 053
+#
+# Sanity check of ext4 mount options
+#
+. ./common/preamble
+_begin_fstest auto mount
+
+seq=`basename $0`
+seqres=$RESULT_DIR/$seq
+
+here=`pwd`
+tmp=/tmp/$$
+status=1       # failure is the default!
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+_cleanup()
+{
+       cd /
+       $UMOUNT_PROG $SCRATCH_MNT > /dev/null 2>&1
+       if [ -n "$LOOP_LOGDEV" ];then
+               _destroy_loop_device $LOOP_LOGDEV 2>/dev/null
+       fi
+       rm -f $tmp.*
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+. ./common/quota
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+
+echo "Silence is golden."
+
+SIZE=$((1024 * 1024))  # 1GB in KB
+LOGSIZE=$((10 *1024))  # 10MB in KB
+
+_supported_fs ext2 ext3 ext4
+_require_scratch_size $SIZE
+_require_quota
+_require_loop
+_require_command "$TUNE2FS_PROG" tune2fs
+MKE2FS_PROG=$(type -P mke2fs)
+_require_command "$MKE2FS_PROG" mke2fs
+_require_kernel_config CONFIG_QFMT_V2
+
+LOG=""
+print_log() {
+       LOG="$LOG $@"
+}
+
+KERNEL_VERSION=`uname -r | cut -d'.' -f1,2`
+KERNEL_MAJ=${KERNEL_VERSION%.*}
+KERNEL_MIN=${KERNEL_VERSION#*.}
+
+kernel_gte() {
+       major=${1%.*}
+       minor=${1#*.}
+
+       if [ $KERNEL_MAJ -gt $major ]; then
+               return 0
+       elif [[ $KERNEL_MAJ -eq $major && $KERNEL_MIN -ge $minor ]]; then
+               return 0
+       fi
+       return 1
+}
+
+
+# The aim here is to keep the mount options behaviour stable going forward
+# so there is not much point in testing older kernels.
+kernel_gte 5.12 || _notrun "This test is only relevant for kernel versions 5.12 and higher"
+
+IGNORED="remount,defaults,ignored,removed"
+CHECK_MINFO="lazytime,noatime,nodiratime,noexec,nosuid,ro"
+ERR=0
+
+test_mnt() {
+       findmnt -n $SCRATCH_DEV > /dev/null 2>&1
+       [ $? -ne 0 ] && return $?
+
+       if [ $# -eq 1 ]; then
+               OPTS=$1
+       elif [ $# -eq 2 ]; then
+               OPTS=$2
+       else
+               return 0
+       fi
+
+       print_log "checking \"$OPTS\" "
+       # test options in /proc/fs/ext4/dev/options
+       (
+       ret=0
+       IFS=','
+       for option in $OPTS; do
+               if echo $IGNORED | grep -w $option; then
+                       continue
+               fi
+
+               [ $option = "noload" ] && option="norecovery"
+
+               if [[ $option = ^* ]]; then
+                       expected=1
+               else
+                       expected=0
+               fi
+               option=${option#^}
+
+               if echo $CHECK_MINFO | grep -w $option; then
+                       findmnt -n -o OPTIONS $SCRATCH_DEV | grep $option
+                       ret=$?
+               else
+                       grep $option /proc/fs/ext4/$(_short_dev $SCRATCH_DEV)/options
+                       ret=$?
+               fi
+
+               if [ $ret -ne $expected ]; then
+                       exit 1
+               fi
+       done
+       ) > /dev/null 2>&1
+       return $?
+}
+
+fail() {
+       print_log "   FAILED"
+       ERR=$((ERR+1))
+       echo $LOG | tee -a $seqres.full
+       LOG=""
+}
+
+ok() {
+       print_log "   OK"
+       echo $LOG >> $seqres.full
+       LOG=""
+}
+
+simple_mount() {
+       _mount $* >> $seqres.full 2>&1
+}
+
+# $1 - can hold -n option, if it does argumetns are shifted
+# $1 - options to test
+# $2 - if provided it's the option string to check for
+do_mnt() {
+       device=$SCRATCH_DEV
+       # If -n argument is provided do not specify $SCRATCH_DEV
+       # usefull for remount
+       if [ "$1" == "-n" ]; then
+               unset device
+               shift
+       fi
+
+       if [ -z "$1" ]; then
+               simple_mount $device $SCRATCH_MNT
+               ret=$?
+       else
+               simple_mount -o $1 $device $SCRATCH_MNT
+               ret=$?
+       fi
+       if [ $ret -eq 0 ]; then
+               test_mnt $1 $2
+               ret=$?
+               [ $ret -ne 0 ] && print_log "(not found)"
+       else
+               print_log "(failed mount)"
+       fi
+
+       return $ret
+}
+
+not_mnt() {
+       # We don't need -t not not_ variant
+       if [ "$1" == "-t" ]; then
+               shift
+       fi
+
+       print_log "SHOULD FAIL mounting $fstype \"$1\" "
+       do_mnt $@
+       if [ $? -eq 0 ]; then
+               fail
+       else
+               ok
+       fi
+       $UMOUNT_PROG $SCRATCH_MNT 2> /dev/null
+}
+
+mnt_only() {
+       print_log "mounting $fstype \"$1\" "
+       do_mnt $@
+       if [ $? -ne 0 ]; then
+               fail
+       else
+               ok
+       fi
+}
+
+mnt() {
+       # Do we need to run the tune2fs mount option test ?
+       t2fs=0
+       if [ "$1" == "-t" ]; then
+               t2fs=1
+               shift
+       fi
+
+       mnt_only $*
+       $UMOUNT_PROG $SCRATCH_MNT 2> /dev/null
+
+       [ "$t2fs" -eq 0 ] && return
+
+       op_set=$1
+       if [ $# -eq 1 ]; then
+               check=$1
+       elif [ $# -eq 2 ]; then
+               check=$2
+       else
+               return 0
+       fi
+
+       # some options need translation for tune2fs
+       op_set=$(echo $op_set | sed -e 's/data=journal/journal_data/' \
+                                   -e 's/data=ordered/journal_data_ordered/' \
+                                   -e 's/data=writeback/journal_data_writeback/')
+       $TUNE2FS_PROG -o $op_set $SCRATCH_DEV > /dev/null 2>&1
+       mnt_only "defaults" $check
+       $UMOUNT_PROG $SCRATCH_MNT 2> /dev/null
+       if [ "$op_set" = ^* ]; then
+               op_set=${op_set#^}
+       else
+               op_set="^${op_set}"
+       fi
+       $TUNE2FS_PROG -o $op_set $SCRATCH_DEV > /dev/null 2>&1
+}
+
+# $1 - options to mount with
+# $2 - options to remount with
+remount() {
+       # First do this specifying both dev and mnt
+       print_log "mounting $fstype \"$1\" "
+       do_mnt $1
+       [ $? -ne 0 ] && fail && return
+       print_log "remounting \"$2\" "
+       do_mnt remount,$2 $3
+       if [ $? -ne 0 ]; then
+               fail
+               $UMOUNT_PROG $SCRATCH_MNT 2> /dev/null
+               return
+       else
+               ok
+       fi
+       $UMOUNT_PROG $SCRATCH_MNT 2> /dev/null
+
+       # Now just specify mnt
+       print_log "mounting $fstype \"$1\" "
+       do_mnt $1
+       [ $? -ne 0 ] && fail && return
+       print_log "remounting (MNT ONLY) \"$2\" "
+       do_mnt -n remount,$2 $3
+       if [ $? -ne 0 ]; then
+               fail
+       else
+               ok
+       fi
+
+       $UMOUNT_PROG $SCRATCH_MNT 2> /dev/null
+}
+
+# $1 - options to mount with, or -r argument
+# $2 - options to remount with
+not_remount_noumount() {
+       remount_only=0
+       # If -r is specified we're going to do remount only
+       if [ "$1" == "-r" ]; then
+               remount_only=1
+               # Dont need shift since first argument would
+               # have been consumed by mount anyway
+       fi
+
+       if [ $remount_only -eq 0 ]; then
+               print_log "mounting $fstype \"$1\" "
+               do_mnt $1
+               [ $? -ne 0 ] && fail && return
+       fi
+       print_log "SHOULD FAIL remounting $fstype \"$2\" "
+       do_mnt remount,$2 $3
+       if [ $? -eq 0 ]; then
+               fail
+       else
+               ok
+       fi
+
+       # Now just specify mnt
+       print_log "SHOULD FAIL remounting $fstype (MNT ONLY) \"$2\" "
+       do_mnt -n remount,$2 $3
+       if [ $? -eq 0 ]; then
+               fail
+       else
+               ok
+       fi
+}
+
+not_remount() {
+       not_remount_noumount $*
+       $UMOUNT_PROG $SCRATCH_MNT 2> /dev/null
+}
+
+
+do_mkfs() {
+       $MKE2FS_PROG -T $fstype -Fq $* >> $seqres.full 2>&1 ||
+       _fail "mkfs failed - $MKFS_EXT4_PROG -Fq $* $SCRATCH_DEV"
+}
+
+not_ext2() {
+       if [[ $fstype == "ext2" ]]; then
+               not_$*
+       else
+               $*
+       fi
+}
+
+only_ext4() {
+       if [[ $fstype == "ext4" ]]; then
+               $*
+       else
+               not_$*
+       fi
+}
+
+# Create logdev for external journal
+LOOP_IMG=$tmp.logdev
+truncate -s ${LOGSIZE}k $LOOP_IMG
+LOOP_LOGDEV=`_create_loop_device $LOOP_IMG`
+majmin=`stat -c "%t:%T" $LOOP_LOGDEV`
+LOGDEV_DEVNUM=`echo "${majmin%:*}*2^8 + ${majmin#*:}" | bc`
+
+# Test all the extN file system supported by ext4 driver
+fstype=
+for fstype in ext2 ext3 ext4; do
+
+       $UMOUNT_PROG $SCRATCH_MNT 2> /dev/null
+       $UMOUNT_PROG $SCRATCH_DEV 2> /dev/null
+
+       do_mkfs $SCRATCH_DEV ${SIZE}k
+
+       # do we have fstype support ?
+       do_mnt
+       if [ $? -ne 0 ]; then
+               print_log "$fstype not supported. Skipping..."
+               ok
+               continue
+       fi
+       if [ ! -f /proc/fs/ext4/$(_short_dev $SCRATCH_DEV)/options ]; then
+               print_log "$fstype not supported. Skipping..."
+               ok
+               continue
+       fi
+
+       $UMOUNT_PROG $SCRATCH_MNT 2> /dev/null
+
+       not_mnt failme
+       mnt
+       mnt bsddf
+       mnt minixdf
+       mnt grpid
+       mnt -t bsdgroups grpid
+       mnt nogrpid
+       mnt sysvgroups nogrpid
+       mnt resgid=1001
+       mnt resuid=1001
+       mnt sb=131072
+       mnt errors=continue
+       mnt errors=panic
+       mnt errors=remount-ro
+       mnt nouid32
+       mnt debug
+       mnt oldalloc removed
+       mnt orlov removed
+       mnt -t user_xattr
+       mnt nouser_xattr
+
+       if _has_kernel_config CONFIG_EXT4_FS_POSIX_ACL; then
+               mnt -t acl
+       else
+               mnt -t acl ^acl
+       fi
+
+       not_ext2 mnt noload norecovery
+       mnt bh removed
+       mnt nobh removed
+       not_ext2 mnt commit=7
+       mnt min_batch_time=200
+       mnt max_batch_time=10000
+       only_ext4 mnt journal_checksum
+       only_ext4 mnt nojournal_checksum
+       only_ext4 mnt journal_async_commit,data=writeback
+       mnt abort ignored
+       not_ext2 mnt -t data=journal
+       not_ext2 mnt -t data=ordered
+       not_ext2 mnt -t data=writeback
+       not_ext2 mnt data_err=abort
+       not_ext2 mnt data_err=ignore ignored
+       mnt usrjquota=aquota.user,jqfmt=vfsv0
+       not_mnt usrjquota=aquota.user
+       mnt usrjquota= ignored
+       mnt grpjquota=aquota.group,jqfmt=vfsv0
+       not_mnt grpjquota=aquota.group
+       mnt grpjquota= ignored
+       mnt jqfmt=vfsold
+       mnt jqfmt=vfsv0
+       mnt jqfmt=vfsv1
+       mnt grpquota
+       mnt quota
+       mnt noquota
+       mnt usrquota
+       mnt grpquota
+       mnt barrier
+       mnt barrier=0 nobarrier
+       mnt barrier=1 barrier
+       mnt barrier=99 barrier
+       mnt -t nobarrier
+       mnt i_version
+       mnt stripe=512
+       only_ext4 mnt delalloc
+       only_ext4 mnt -t nodelalloc
+       mnt warn_on_error
+       mnt nowarn_on_error
+       not_mnt debug_want_extra_isize=512
+       mnt debug_want_extra_isize=32 ignored
+       mnt mblk_io_submit removed
+       mnt nomblk_io_submit removed
+       mnt -t block_validity
+       mnt noblock_validity
+       mnt inode_readahead_blks=16
+       not_ext2 mnt journal_ioprio=6 ignored
+       mnt auto_da_alloc=0 noauto_da_alloc
+       mnt auto_da_alloc=1 auto_da_alloc
+       mnt auto_da_alloc=95 auto_da_alloc
+       mnt auto_da_alloc
+       mnt noauto_da_alloc
+       only_ext4 mnt dioread_nolock
+       only_ext4 mnt nodioread_nolock
+       only_ext4 mnt dioread_lock nodioread_nolock
+       mnt -t discard
+       mnt nodiscard
+       mnt init_itable=20
+       mnt init_itable
+       mnt init_itable=0
+       mnt noinit_itable
+       mnt max_dir_size_kb=4096
+
+       if _has_kernel_config CONFIG_FS_ENCRYPTION; then
+               mnt test_dummy_encryption
+               mnt test_dummy_encryption=v1
+               mnt test_dummy_encryption=v2
+               not_mnt test_dummy_encryption=v3
+               not_mnt test_dummy_encryption=
+       else
+               mnt test_dummy_encryption ^test_dummy_encryption
+               mnt test_dummy_encryption=v1 ^test_dummy_encryption=v1
+               mnt test_dummy_encryption=v2 ^test_dummy_encryption=v2
+               not_mnt test_dummy_encryption=v3
+               not_mnt test_dummy_encryption=
+       fi
+
+       if _has_kernel_config CONFIG_FS_ENCRYPTION_INLINE_CRYPT; then
+               mnt inlinecrypt
+       else
+               mnt inlinecrypt ^inlinecrypt
+       fi
+
+       mnt prefetch_block_bitmaps removed
+       mnt no_prefetch_block_bitmaps
+       # We don't currently have a way to know that the option has been
+       # applied, so comment it out for now. This should be fixed in the
+       # future.
+       #mnt mb_optimize_scan=0
+       #mnt mb_optimize_scan=1
+       #not_mnt mb_optimize_scan=9
+       #not_mnt mb_optimize_scan=
+       mnt nombcache
+       mnt no_mbcache nombcache
+       mnt check=none removed
+       mnt nocheck removed
+       mnt reservation removed
+       mnt noreservation removed
+       mnt journal=20 ignored
+       not_mnt nonsenseoption
+       not_mnt nonsenseoption=value
+
+       # generic mount options
+       mnt lazytime
+       mnt nolazytime ^lazytime
+       mnt noatime
+       mnt nodiratime
+       mnt noexec
+       mnt nosuid
+       mnt ro
+
+       # generic remount check
+       remount barrier nobarrier
+       remount nobarrier barrier
+       remount discard nodiscard
+       remount nodiscard discard
+
+       # dax mount options
+       simple_mount -o dax=always $SCRATCH_DEV $SCRATCH_MNT > /dev/null 2>&1
+       if [ $? -eq 0 ]; then
+               $UMOUNT_PROG $SCRATCH_MNT 2> /dev/null
+               mnt dax
+               mnt dax=always
+               mnt dax=never
+               mnt dax=inode
+
+               not_remount lazytime dax
+               not_remount dax=always dax=never
+
+               if [[ $fstype != "ext2" ]]; then
+                       not_remount data=journal dax
+                       not_remount data=journal dax=always
+                       not_remount data=journal dax=never
+                       not_remount data=journal dax=inode
+               fi
+       fi
+
+       # Quota remount check
+       remount grpquota usrquota
+       remount usrquota quota
+       remount usrquota usrjquota=q.u,jqfmt=vfsv0
+       remount grpquota grpjquota=q.g,jqfmt=vfsv0
+
+       not_remount usrquota grpjquota=q.g,jqfmt=vfsv0
+       not_remount grpquota usrjquota=q.u,jqfmt=vfsv0
+
+       remount quota usrjquota=q.u,jqfmt=vfsv0
+       not_remount quota grpjquota=q.g,jqfmt=vfsv0
+
+       remount usrjquota=q.u,jqfmt=vfsv0 grpjquota=q.g
+       not_remount usrjquota=q.u,jqfmt=vfsv0 usrjquota=q.ua
+       not_remount grpjquota=q.g,jqfmt=vfsv0 grpjquota=q.ga
+
+       remount usrjquota=q.u,jqfmt=vfsv0 usrquota usrjquota=q.u,jqfmt=vfsv0
+       remount grpjquota=q.g,jqfmt=vfsv0 grpquota grpjquota=q.g,jqfmt=vfsv0
+       not_remount usrjquota=q.u,jqfmt=vfsv0 grpquota
+       not_remount grpjquota=q.g,jqfmt=vfsv0 usrquota
+
+       remount grpjquota=q.g,jqfmt=vfsv0 grpjquota= ^grpjquota=
+       remount usrjquota=q.u,jqfmt=vfsv0 usrjquota= ^usrjquota=
+       remount grpjquota=q.g,usrjquota=q.u,jqfmt=vfsv0 grpjquota=,usrjquota= ^grpjquota=,^usrjquota=
+
+       remount jqfmt=vfsv0 grpjquota=q.g
+       remount jqfmt=vfsv0 usrjquota=q.u
+
+       if [[ $fstype != "ext2" ]]; then
+               remount noload data=journal norecovery
+               not_remount data=ordered data=journal
+               not_remount data=journal data=writeback
+               not_remount data=writeback data=ordered
+       fi
+
+       do_mkfs -O journal_dev $LOOP_LOGDEV ${LOGSIZE}k
+       do_mkfs -J device=$LOOP_LOGDEV $SCRATCH_DEV ${SIZE}k
+       mnt defaults
+       mnt journal_path=$LOOP_LOGDEV ignored
+       mnt journal_dev=$LOGDEV_DEVNUM ignored
+       not_mnt journal_path=${LOOP_LOGDEV}_nonexistent ignored
+       not_mnt journal_dev=123456 ignored
+       not_mnt journal_dev=999999999999999 ignored
+
+       do_mkfs -E quotatype=prjquota $SCRATCH_DEV ${SIZE}k
+       mnt prjquota
+
+       # test clearing/changing journalled quota when enabled
+       echo "== Testing active journalled quota" >> $seqres.full
+       mnt_only prjquota,grpjquota=aquota.group,usrjquota=aquota.user,jqfmt=vfsv0
+
+       # Prepare and enable quota
+       quotacheck -vugm $SCRATCH_MNT >> $seqres.full 2>&1
+       quotaon -vug $SCRATCH_MNT >> $seqres.full 2>&1
+
+       not_remount_noumount -r grpjquota=
+       not_remount_noumount -r usrjquota=aaquota.user
+       not_remount_noumount -r grpjquota=aaquota.group
+       not_remount_noumount -r jqfmt=vfsv1
+       not_remount_noumount -r noquota
+       mnt_only remount,usrquota,grpquota ^usrquota,^grpquota
+       $UMOUNT_PROG $SCRATCH_MNT > /dev/null 2>&1
+
+       # test clearing/changing quota when enabled
+       do_mkfs -E quotatype=^prjquota $SCRATCH_DEV ${SIZE}k
+       not_mnt prjquota
+       echo "== Testing active non-journalled quota" >> $seqres.full
+       mnt_only grpquota,usrquota
+
+       # Prepare and enable quota
+       quotacheck -vugm $SCRATCH_MNT >> $seqres.full 2>&1
+       quotaon -vug $SCRATCH_MNT >> $seqres.full 2>&1
+
+       not_remount_noumount -r noquota
+       not_remount_noumount -r usrjquota=aquota.user
+       not_remount_noumount -r grpjquota=aquota.group
+       not_remount_noumount -r jqfmt=vfsv1
+       mnt_only remount,grpjquota= grpquota,^grpjquota
+       mnt_only remount,usrjquota= usrquota,^usrjquota
+       mnt_only remount,usrquota,grpquota usrquota,grpquota
+       quotaoff -f $SCRATCH_MNT >> $seqres.full 2>&1
+       mnt_only remount,noquota ^usrquota,^grpquota,quota
+       $UMOUNT_PROG $SCRATCH_MNT > /dev/null 2>&1
+
+       # Quota feature
+       echo "== Testing quota feature " >> $seqres.full
+       do_mkfs -O quota -E quotatype=prjquota $SCRATCH_DEV ${SIZE}k
+       mnt usrjquota=aquota.user,jqfmt=vfsv0 ^usrjquota=
+       mnt grpjquota=aquota.user,jqfmt=vfsv0 ^grpjquota=
+       mnt jqfmt=vfsv1 ^jqfmt=
+       mnt prjquota
+       mnt usrquota
+       mnt grpquota
+       not_remount defaults usrjquota=aquota.user
+       not_remount defaults grpjquota=aquota.user
+       not_remount defaults jqfmt=vfsv1
+       remount defaults grpjquota=,usrjquota= ignored
+
+done #for fstype in ext2 ext3 ext4; do
+
+$UMOUNT_PROG $SCRATCH_MNT > /dev/null 2>&1
+echo "$ERR errors encountered" >> $seqres.full
+
+status=$ERR
+exit
diff --git a/tests/ext4/053.out b/tests/ext4/053.out
new file mode 100644 (file)
index 0000000..d58db7e
--- /dev/null
@@ -0,0 +1,2 @@
+QA output created by 053
+Silence is golden.