Skip to contentSkip to conversation

common/xfs: wipe the XFS superblock of each AGs

PATCH Superseded Zorro Lang 2 participants 4 messages Sep 20, 2019 – Sep 23, 2019

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Zorro Lang

author
#0 lore raw

PATCH v2 1/1 common/xfs: wipe the XFS superblock of each AGs

xfs/030 always fails after d0e484ac699f ("check: wipe scratch devices between tests") get merged.

Due to xfs/030 does a sized(100m) mkfs. Before we merge above commit, mkfs.xfs detects an old primary superblock, it will write zeroes to all superblocks before formatting the new filesystem. But this won't be done if we wipe the first superblock(by merging above commit).

That means if we make a (smaller) sized xfs after wipefs, those *old* superblocks which created by last time mkfs.xfs will be left on disk. Then when we do xfs_repair, if xfs_repair can't find the first SB, it will go to find those *old* SB at first. When it finds them, everyting goes wrong.

So I try to get XFS AG geometry(by default) and then try to erase all superblocks. Thanks Darrick J. Wong helped to analyze this issue.

---

Hi,

V2 did below changes:
1) Use xfs_db to detect the real xfs geometry
2) Do a $FSTYP specified wipe before trying to wipefs all scratch devices

Thanks,
Zorro

 common/rc  |  8 ++++++++
 common/xfs | 20 ++++++++++++++++++++
 2 files changed, 28 insertions(+)
common/rc +8 -0
index 66c7fd4d..56329747 100644
@@ -4045,6 +4045,14 @@ _try_wipe_scratch_devs()
 {
 	test -x "$WIPEFS_PROG" || return 0
 
+	# Do specified filesystem wipe at first
+	case "$FSTYP" in
+	"xfs")
+		_try_wipe_scratch_xfs
+		;;
+	esac
+
+	# Then do wipefs on all scratch devices
 	for dev in $SCRATCH_DEV_POOL $SCRATCH_DEV $SCRATCH_LOGDEV $SCRATCH_RTDEV; do
 		test -b $dev && $WIPEFS_PROG -a $dev
 	done
common/xfs +20 -0
index 1bce3c18..082a1744 100644
@@ -884,3 +884,23 @@ _xfs_mount_agcount()
 {
 	$XFS_INFO_PROG "$1" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g'
 }
+
+# Wipe the superblock of each XFS AGs
+_try_wipe_scratch_xfs()
+{
+	local num='^[0-9]+$'
+	local agcount
+	local agsize
+	local dbsize
+
+	agcount=`_scratch_xfs_get_sb_field agcount 2>/dev/null`
+	agsize=`_scratch_xfs_get_sb_field agblocks 2>/dev/null`
+	dbsize=`_scratch_xfs_get_sb_field blocksize 2>/dev/null`
+
+	if [[ $agcount =~ $num && $agsize =~ $num && $dbsize =~ $num ]];then
+		for ((i = 0; i < agcount; i++)); do
+			$XFS_IO_PROG -c "pwrite $((i * dbsize * agsize)) $dbsize" \
+				$SCRATCH_DEV >/dev/null;
+		done
+	fi
+}
Signature · 1 line
2.20.1
#1 lore raw
Quoted 70 lines · Zorro Lang

xfs/030 always fails after d0e484ac699f ("check: wipe scratch devices between tests") get merged.

Due to xfs/030 does a sized(100m) mkfs. Before we merge above commit, mkfs.xfs detects an old primary superblock, it will write zeroes to all superblocks before formatting the new filesystem. But this won't be done if we wipe the first superblock(by merging above commit).

That means if we make a (smaller) sized xfs after wipefs, those *old* superblocks which created by last time mkfs.xfs will be left on disk. Then when we do xfs_repair, if xfs_repair can't find the first SB, it will go to find those *old* SB at first. When it finds them, everyting goes wrong.

So I try to get XFS AG geometry(by default) and then try to erase all superblocks. Thanks Darrick J. Wong helped to analyze this issue.

---

Hi,

V2 did below changes:
1) Use xfs_db to detect the real xfs geometry
2) Do a $FSTYP specified wipe before trying to wipefs all scratch devices

Thanks,
Zorro

 common/rc  |  8 ++++++++
 common/xfs | 20 ++++++++++++++++++++
 2 files changed, 28 insertions(+)
common/rc +8 -0
index 66c7fd4d..56329747 100644
@@ -4045,6 +4045,14 @@ _try_wipe_scratch_devs()
 {
 	test -x "$WIPEFS_PROG" || return 0
 
+	# Do specified filesystem wipe at first
+	case "$FSTYP" in
+	"xfs")
+		_try_wipe_scratch_xfs
+		;;
+	esac
+
+	# Then do wipefs on all scratch devices
 	for dev in $SCRATCH_DEV_POOL $SCRATCH_DEV $SCRATCH_LOGDEV $SCRATCH_RTDEV; do
 		test -b $dev && $WIPEFS_PROG -a $dev
 	done
common/xfs +9 -0
index 1bce3c18..082a1744 100644
@@ -884,3 +884,23 @@ _xfs_mount_agcount()
 {
 	$XFS_INFO_PROG "$1" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g'
 }
+
+# Wipe the superblock of each XFS AGs
+_try_wipe_scratch_xfs()
+{
+	local num='^[0-9]+$'
+	local agcount
+	local agsize
+	local dbsize
+

common/xfs _xfs_mount_agcount() · line 905

Quoted 10 lines
common/xfs +10 -0
+	agcount=`_scratch_xfs_get_sb_field agcount 2>/dev/null`
+	agsize=`_scratch_xfs_get_sb_field agblocks 2>/dev/null`
+	dbsize=`_scratch_xfs_get_sb_field blocksize 2>/dev/null`
+
+	if [[ $agcount =~ $num && $agsize =~ $num && $dbsize =~ $num ]];then
+		for ((i = 0; i < agcount; i++)); do
+			$XFS_IO_PROG -c "pwrite $((i * dbsize * agsize)) $dbsize" \
+				$SCRATCH_DEV >/dev/null;
+		done
+	fi

What happened to the loop that simulates a _scratch_mkfs_xfs run (to get the AG geometry) and then zaps that too? You need both zeroing loops to make sure xfs/030 doesn't trip over old superblocks, right?

--D

Quoted 4 lines
common/xfs +1 -0
+}
Signature · 1 line
2.20.1
author
#2 lore raw
Quoted 77 lines · Darrick J. Wong
Quoted 76 lines · Zorro Lang

xfs/030 always fails after d0e484ac699f ("check: wipe scratch devices between tests") get merged.

Due to xfs/030 does a sized(100m) mkfs. Before we merge above commit, mkfs.xfs detects an old primary superblock, it will write zeroes to all superblocks before formatting the new filesystem. But this won't be done if we wipe the first superblock(by merging above commit).

That means if we make a (smaller) sized xfs after wipefs, those *old* superblocks which created by last time mkfs.xfs will be left on disk. Then when we do xfs_repair, if xfs_repair can't find the first SB, it will go to find those *old* SB at first. When it finds them, everyting goes wrong.

So I try to get XFS AG geometry(by default) and then try to erase all superblocks. Thanks Darrick J. Wong helped to analyze this issue.

---

Hi,

V2 did below changes:
1) Use xfs_db to detect the real xfs geometry
2) Do a $FSTYP specified wipe before trying to wipefs all scratch devices

Thanks,
Zorro

 common/rc  |  8 ++++++++
 common/xfs | 20 ++++++++++++++++++++
 2 files changed, 28 insertions(+)
common/rc +8 -0
index 66c7fd4d..56329747 100644
@@ -4045,6 +4045,14 @@ _try_wipe_scratch_devs()
 {
 	test -x "$WIPEFS_PROG" || return 0
 
+	# Do specified filesystem wipe at first
+	case "$FSTYP" in
+	"xfs")
+		_try_wipe_scratch_xfs
+		;;
+	esac
+
+	# Then do wipefs on all scratch devices
 	for dev in $SCRATCH_DEV_POOL $SCRATCH_DEV $SCRATCH_LOGDEV $SCRATCH_RTDEV; do
 		test -b $dev && $WIPEFS_PROG -a $dev
 	done
common/xfs +15 -0
index 1bce3c18..082a1744 100644
@@ -884,3 +884,23 @@ _xfs_mount_agcount()
 {
 	$XFS_INFO_PROG "$1" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g'
 }
+
+# Wipe the superblock of each XFS AGs
+_try_wipe_scratch_xfs()
+{
+	local num='^[0-9]+$'
+	local agcount
+	local agsize
+	local dbsize
+
+	agcount=`_scratch_xfs_get_sb_field agcount 2>/dev/null`
+	agsize=`_scratch_xfs_get_sb_field agblocks 2>/dev/null`
+	dbsize=`_scratch_xfs_get_sb_field blocksize 2>/dev/null`
+
+	if [[ $agcount =~ $num && $agsize =~ $num && $dbsize =~ $num ]];then
+		for ((i = 0; i < agcount; i++)); do
Quoted 8 lines

common/xfs _xfs_mount_agcount() · line 905

Quoted 4 lines
common/xfs +4 -0
+			$XFS_IO_PROG -c "pwrite $((i * dbsize * agsize)) $dbsize" \
+				$SCRATCH_DEV >/dev/null;
+		done
+	fi

What happened to the loop that simulates a _scratch_mkfs_xfs run (to get the AG geometry) and then zaps that too? You need both zeroing loops to make sure xfs/030 doesn't trip over old superblocks, right?

Hi,

Sorry I'm not sure what you mean. Do you mean I'd better to do two zeroing loops, one is as above, the other is for default mkfs.xfs geometry as below[1]?

Thanks,
Zorro

[1]
local tmp=mktemp -u
_scratch_mkfs_xfs -N 2>/dev/null | perl -ne '

    if (/^meta-data=.*\s+agcount=(\d+), agsize=(\d+) blks/) {
      print STDOUT "agcount=$1\nagsize=$2\n";
    }
    if (/^data\s+=\s+bsize=(\d+)\s/) {
      print STDOUT "dbsize=$1\n";
    }' > $tmp.mkfs
  . $tmp.mkfs
  if [ -n "$agcount" -a -n "$agsize" -a -n "$dbsize" ];then
    for ((i = 0; i < agcount; i++)); do
      $XFS_IO_PROG -c "pwrite $((i * dbsize * agsize)) $dbsize" \
         $SCRATCH_DEV >/dev/null;
    done
  fi
  rm -f $tmp.mkfs
Quoted 7 lines

--D

Quoted 4 lines
common/xfs +1 -0
+}
Signature · 1 line
2.20.1
#3 lore raw
Quoted 87 lines · Zorro Lang
Quoted 77 lines · Darrick J. Wong
Quoted 76 lines · Zorro Lang

xfs/030 always fails after d0e484ac699f ("check: wipe scratch devices between tests") get merged.

Due to xfs/030 does a sized(100m) mkfs. Before we merge above commit, mkfs.xfs detects an old primary superblock, it will write zeroes to all superblocks before formatting the new filesystem. But this won't be done if we wipe the first superblock(by merging above commit).

That means if we make a (smaller) sized xfs after wipefs, those *old* superblocks which created by last time mkfs.xfs will be left on disk. Then when we do xfs_repair, if xfs_repair can't find the first SB, it will go to find those *old* SB at first. When it finds them, everyting goes wrong.

So I try to get XFS AG geometry(by default) and then try to erase all superblocks. Thanks Darrick J. Wong helped to analyze this issue.

---

Hi,

V2 did below changes:
1) Use xfs_db to detect the real xfs geometry
2) Do a $FSTYP specified wipe before trying to wipefs all scratch devices

Thanks,
Zorro

 common/rc  |  8 ++++++++
 common/xfs | 20 ++++++++++++++++++++
 2 files changed, 28 insertions(+)
common/rc +8 -0
index 66c7fd4d..56329747 100644
@@ -4045,6 +4045,14 @@ _try_wipe_scratch_devs()
 {
 	test -x "$WIPEFS_PROG" || return 0
 
+	# Do specified filesystem wipe at first
+	case "$FSTYP" in
+	"xfs")
+		_try_wipe_scratch_xfs
+		;;
+	esac
+
+	# Then do wipefs on all scratch devices
 	for dev in $SCRATCH_DEV_POOL $SCRATCH_DEV $SCRATCH_LOGDEV $SCRATCH_RTDEV; do
 		test -b $dev && $WIPEFS_PROG -a $dev
 	done
common/xfs +15 -0
index 1bce3c18..082a1744 100644
@@ -884,3 +884,23 @@ _xfs_mount_agcount()
 {
 	$XFS_INFO_PROG "$1" | grep agcount= | sed -e 's/^.*agcount=\([0-9]*\),.*$/\1/g'
 }
+
+# Wipe the superblock of each XFS AGs
+_try_wipe_scratch_xfs()
+{
+	local num='^[0-9]+$'
+	local agcount
+	local agsize
+	local dbsize
+
+	agcount=`_scratch_xfs_get_sb_field agcount 2>/dev/null`
+	agsize=`_scratch_xfs_get_sb_field agblocks 2>/dev/null`
+	dbsize=`_scratch_xfs_get_sb_field blocksize 2>/dev/null`
+
+	if [[ $agcount =~ $num && $agsize =~ $num && $dbsize =~ $num ]];then
+		for ((i = 0; i < agcount; i++)); do
Quoted 8 lines

common/xfs _xfs_mount_agcount() · line 905

Quoted 4 lines
common/xfs +4 -0
+			$XFS_IO_PROG -c "pwrite $((i * dbsize * agsize)) $dbsize" \
+				$SCRATCH_DEV >/dev/null;
+		done
+	fi

What happened to the loop that simulates a _scratch_mkfs_xfs run (to get the AG geometry) and then zaps that too? You need both zeroing loops to make sure xfs/030 doesn't trip over old superblocks, right?

Quoted 5 lines

Hi,

Sorry I'm not sure what you mean. Do you mean I'd better to do two zeroing loops, one is as above, the other is for default mkfs.xfs geometry as below[1]?

Yes.

--D

Quoted 31 lines

Thanks,
Zorro

[1]
local tmp=mktemp -u
_scratch_mkfs_xfs -N 2>/dev/null | perl -ne '

    if (/^meta-data=.*\s+agcount=(\d+), agsize=(\d+) blks/) {
      print STDOUT "agcount=$1\nagsize=$2\n";
    }
    if (/^data\s+=\s+bsize=(\d+)\s/) {
      print STDOUT "dbsize=$1\n";
    }' > $tmp.mkfs
  . $tmp.mkfs
  if [ -n "$agcount" -a -n "$agsize" -a -n "$dbsize" ];then
    for ((i = 0; i < agcount; i++)); do
      $XFS_IO_PROG -c "pwrite $((i * dbsize * agsize)) $dbsize" \
         $SCRATCH_DEV >/dev/null;
    done
  fi
  rm -f $tmp.mkfs
Quoted 7 lines

--D

Quoted 4 lines
common/xfs +1 -0
+}
Signature · 1 line
2.20.1