On Fri, Sep 28, 2018 at 3:51 PM Anand Jain <anand.jain@xxxxxxxxxx> wrote:
>
> Try to punch hole with unaligned size and offset when the FS
> returns ENOSPC
The FS returns ENOSPC is confusing. It's more clear to say when the
filesystem doesn't have more space available for data allocation.
>
> Signed-off-by: Anand Jain <anand.jain@xxxxxxxxxx>
> ---
> v2->v3:
> add _require_xfs_io_command "fpunch"
> add more logs to $seqfull.full
> mount options and
> group profile info
> add sync after dd upto ENOSPC
> drop fallocate -p and use xfs_io punch to create holes
> v1->v2: Use at least 256MB to test.
> This test case fails on btrfs as of now.
>
> tests/btrfs/172 | 74 +++++++++++++++++++++++++++++++++++++++++++++++++++++
> tests/btrfs/172.out | 2 ++
> tests/btrfs/group | 1 +
> 3 files changed, 77 insertions(+)
> create mode 100755 tests/btrfs/172
> create mode 100644 tests/btrfs/172.out
>
> diff --git a/tests/btrfs/172 b/tests/btrfs/172
> new file mode 100755
> index 000000000000..59413a5de12f
> --- /dev/null
> +++ b/tests/btrfs/172
> @@ -0,0 +1,74 @@
> +#! /bin/bash
> +# SPDX-License-Identifier: GPL-2.0
> +# Copyright (c) 2018 Oracle. All Rights Reserved.
> +#
> +# FS QA Test 172
> +#
> +# Test if the unaligned (by size and offset) punch hole is successful when FS
> +# is at ENOSPC.
> +#
> +seq=`basename $0`
> +seqres=$RESULT_DIR/$seq
> +echo "QA output created by $seq"
> +
> +here=`pwd`
> +tmp=/tmp/$$
> +status=1 # failure is the default!
> +trap "_cleanup; exit \$status" 0 1 2 3 15
> +
> +_cleanup()
> +{
> + cd /
> + rm -f $tmp.*
> +}
> +
> +# get standard environment, filters and checks
> +. ./common/rc
> +. ./common/filter
> +
> +# remove previous $seqres.full before test
> +rm -f $seqres.full
> +
> +# real QA test starts here
> +
> +# Modify as appropriate.
> +_supported_fs btrfs
> +_supported_os Linux
> +_require_scratch
> +_require_xfs_io_command "fpunch"
> +
> +_scratch_mkfs_sized $((256 * 1024 *1024)) >> $seqres.full
> +
> +# max_inline helps to create regular extent
max_inline ensures data is not inlined within metadata extents
> +_scratch_mount "-o max_inline=0,nodatacow"
> +
> +cat /proc/self/mounts | grep $SCRATCH_DEV >> $seqres.full
> +$BTRFS_UTIL_PROG filesystem df $SCRATCH_MNT >> $seqres.full
> +
> +extent_size=$(_scratch_btrfs_sectorsize)
> +unalign_by=512
> +echo extent_size=$extent_size unalign_by=$unalign_by >> $seqres.full
> +
> +$XFS_IO_PROG -f -d -c "pwrite -S 0xab 0 $((extent_size * 10))" \
> + $SCRATCH_MNT/testfile >> $seqres.full
> +
> +echo "Fill fs upto ENOSPC" >> $seqres.full
Fill all space available for data and all unallocated space.
> +dd status=none if=/dev/zero of=$SCRATCH_MNT/filler bs=512 >> $seqres.full 2>&1
Why do you use dd here and not xfs_io?
> +sync
Why is the sync needed?
> +
> +hole_offset=0
> +hole_len=$unalign_by
> +$XFS_IO_PROG -f -d -c "fpunch $hole_offset $hole_len" $SCRATCH_MNT/testfile
No need to pass -f anymore. No need for -d either.
> +
> +hole_offset=$(($extent_size + $unalign_by))
> +hole_len=$(($extent_size - $unalign_by))
> +$XFS_IO_PROG -f -d -c "fpunch $hole_offset $hole_len" $SCRATCH_MNT/testfile
No need to pass -f anymore. No need for -d either.
> +
> +hole_offset=$(($extent_size * 2 + $unalign_by))
> +hole_len=$(($extent_size * 5))
> +$XFS_IO_PROG -f -d -c "fpunch $hole_offset $hole_len" $SCRATCH_MNT/testfile
No need to pass -f anymore. No need for -d either.
> +
> +# success, all done
> +echo "Silence is golden"
> +status=0
> +exit
> diff --git a/tests/btrfs/172.out b/tests/btrfs/172.out
> new file mode 100644
> index 000000000000..ce2de3f0d107
> --- /dev/null
> +++ b/tests/btrfs/172.out
> @@ -0,0 +1,2 @@
> +QA output created by 172
> +Silence is golden
> diff --git a/tests/btrfs/group b/tests/btrfs/group
> index feffc45b6564..7e1a638ab7e1 100644
> --- a/tests/btrfs/group
> +++ b/tests/btrfs/group
> @@ -174,3 +174,4 @@
> 169 auto quick send
> 170 auto quick snapshot
> 171 auto quick qgroup
> +172 auto quick
> --
> 1.8.3.1
>
--
Filipe David Manana,
“Whether you think you can, or you think you can't — you're right.”
