Skip to contentSkip to conversation

btrfs: add test case for newly supported cases of cloning inline extents

PATCH In review [email protected] 2 participants 2 messages Feb 19, 2020 – Feb 20, 2020 Btrfs file system

lore
Share & tools
lore
#0 lore raw

PATCH 2/2 btrfs: add test case for newly supported cases of cloning inline extents

From: Filipe Manana <[email protected]>

Test several scenarios of cloning operations where the source range includes inline extents. They used to not be supported on btrfs because their implementation was not straightforward, and therefore these operations used to fail with errno EOPNOTSUPP on older kernels.

This currently fails on any released kernel. It passes only when the patch with the following subject is applied:

"Btrfs: implement full reflink support for inline extents"

---
 tests/btrfs/205     | 173 +++++++++++++++++++++++++++++++
 tests/btrfs/205.out | 241 ++++++++++++++++++++++++++++++++++++++++++++
 tests/btrfs/group   |   1 +
 3 files changed, 415 insertions(+)
 create mode 100755 tests/btrfs/205
 create mode 100644 tests/btrfs/205.out
tests/btrfs/205 new file +173 -0
new file mode 100755
index 00000000..9bec2bfa
@@ -0,0 +1,173 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (C) 2020 SUSE Linux Products GmbH. All Rights Reserved.
+#
+# FSQA Test No. 205
+#
+# Test several scenarios of cloning operations where the source range includes
+# inline extents. They used to not be supported on btrfs because their
+# implementation was not straightforward, and therefore these operations used
+# to fail with errno EOPNOTSUPP on older kernels.
+#
+# Support for this was added by a patch with the following subject:
+#
+#   "Btrfs: implement full reflink support for inline extents"
+#
+seq=`basename $0`
+seqres=$RESULT_DIR/$seq
+echo "QA output created by $seq"
+tmp=/tmp/$$
+status=1	# failure is the default!
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+_cleanup()
+{
+	cd /
+	rm -f $tmp.*
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+. ./common/reflink
+
+# real QA test starts here
+_supported_fs btrfs
+_supported_os Linux
+_require_scratch_reflink
+_require_xfs_io_command "falloc" "-k"
+_require_command "$CHATTR_PROG" chattr
+_require_btrfs_fs_feature "no_holes"
+_require_btrfs_mkfs_feature "no-holes"
+
+run_tests()
+{
+    rm -f $SCRATCH_MNT/foo* $SCRATCH_MNT/bar*
+
+    # File foo1 has an inline extent encoding 4K of data followed by a regular
+    # extent. It has a file size of 128K.
+    echo "Creating file foo1"
+    touch $SCRATCH_MNT/foo1
+    $CHATTR_PROG +c $SCRATCH_MNT/foo1
+    $XFS_IO_PROG -c "pwrite -S 0xab 0 4K" \
+		 -c "fsync" \
+		 -c "pwrite -S 0xab 4K 124K" \
+		 $SCRATCH_MNT/foo1 | _filter_xfs_io
+
+    # File bar1 has a single 128K extent, and a file size of 128K.
+    echo "Creating file bar1"
+    $XFS_IO_PROG -f -c "pwrite -S 0xcd 0 128K" $SCRATCH_MNT/bar1 | _filter_xfs_io
+
+    echo "Cloning foo1 into the end of bar1"
+    $XFS_IO_PROG -c "reflink $SCRATCH_MNT/foo1 0 128K 128K" $SCRATCH_MNT/bar1 \
+        | _filter_xfs_io
+
+    echo "File bar1 digest = $(_md5_checksum $SCRATCH_MNT/bar1)"
+
+    # File foo2 has an inline extent with 1000 bytes of data and it's followed
+    # by a regular extent of 60K. It has a file size of 64K.
+    echo "Creating file foo2"
+    $XFS_IO_PROG -f -c "pwrite -S 0xab 0 1000" \
+		 -c "fsync" \
+		 -c "falloc 0 4K" \
+		 -c "pwrite -S 0xab 4K 60K" \
+		 $SCRATCH_MNT/foo2 | _filter_xfs_io
+
+    # File bar2 has a regular extent of 64K and a file size of 64K too.
+    echo "Creating file bar2"
+    $XFS_IO_PROG -f -c "pwrite -S 0xcd 0 64K" $SCRATCH_MNT/bar2 | _filter_xfs_io
+
+    echo "Cloning foo2 into the end of bar2"
+    $XFS_IO_PROG -c "reflink $SCRATCH_MNT/foo2 0 64K 64K" $SCRATCH_MNT/bar2 \
+        | _filter_xfs_io
+
+    echo "File bar2 digest = $(_md5_checksum $SCRATCH_MNT/bar2)"
+
+    # File bar3 has a regular extent of 128K and a file size of 128K too.
+    echo "Creating file bar3"
+    $XFS_IO_PROG -f -c "pwrite -S 0xcd 0 128K" $SCRATCH_MNT/bar3 \
+        | _filter_xfs_io
+
+    echo "Cloning foo2 into the middle of bar3"
+    $XFS_IO_PROG -c "reflink $SCRATCH_MNT/foo2 0 64K 64K" $SCRATCH_MNT/bar3 \
+        | _filter_xfs_io
+
+    echo "File bar3 digest = $(_md5_checksum $SCRATCH_MNT/bar3)"
+
+    # File bar4 has a 64K hole at offset 0 followed by a 64K regular extent, and
+    # a file size of 128K.
+    echo "Creating file bar4"
+    $XFS_IO_PROG -f -c "pwrite -S 0xcd 64K 64K" $SCRATCH_MNT/bar4 | _filter_xfs_io
+
+    echo "Cloning foo1 into bar4"
+    $XFS_IO_PROG -c "reflink $SCRATCH_MNT/foo1 0 0 128K" $SCRATCH_MNT/bar4 \
+        | _filter_xfs_io
+
+    echo "File bar4 digest = $(_md5_checksum $SCRATCH_MNT/bar4)"
+
+    # File bar5 has a 1Mb prealloc extent at file offset 0 and a file size of 0.
+    echo "Creating file bar5"
+    $XFS_IO_PROG -f -c "falloc -k 0 1M" $SCRATCH_MNT/bar5
+
+    echo "Cloning foo1 into bar5"
+    $XFS_IO_PROG -c "reflink $SCRATCH_MNT/foo1 0 0 128K" $SCRATCH_MNT/bar5 \
+        | _filter_xfs_io
+
+    echo "File bar5 digest = $(_md5_checksum $SCRATCH_MNT/bar5)"
+
+    # File bar6 has an inline extent encoding 500 bytes of data followed by a
+    # prealloc extent of 1Mb at file offset 4K. The file size is 500 bytes.
+    echo "Creating file bar6"
+    $XFS_IO_PROG -f -c "pwrite -S 0xef 0 500" \
+		 -c "falloc -k 4K 1M" \
+		 $SCRATCH_MNT/bar6 | _filter_xfs_io
+
+    echo "Cloning foo1 into bar6"
+    $XFS_IO_PROG -c "reflink $SCRATCH_MNT/foo1 0 0 128K" $SCRATCH_MNT/bar6 \
+        | _filter_xfs_io
+
+    echo "File bar6 digest = $(_md5_checksum $SCRATCH_MNT/bar6)"
+
+    # Unmount and mount again the filesystem. We want to verify the reflink
+    # operations were durably persisted.
+    _scratch_cycle_mount
+
+    echo "File digests after mounting again the filesystem:"
+    echo "File bar1 digest = $(_md5_checksum $SCRATCH_MNT/bar1)"
+    echo "File bar2 digest = $(_md5_checksum $SCRATCH_MNT/bar2)"
+    echo "File bar3 digest = $(_md5_checksum $SCRATCH_MNT/bar3)"
+    echo "File bar4 digest = $(_md5_checksum $SCRATCH_MNT/bar4)"
+    echo "File bar5 digest = $(_md5_checksum $SCRATCH_MNT/bar5)"
+    echo "File bar6 digest = $(_md5_checksum $SCRATCH_MNT/bar6)"
+}
+
+_scratch_mkfs "-O ^no-holes" >>$seqres.full 2>&1
+_scratch_mount
+
+echo
+echo "Testing with defaults"
+echo
+run_tests
+
+echo
+echo "Testing with -o compress"
+echo
+_scratch_cycle_mount "compress"
+run_tests
+
+echo
+echo "Testing with -o nodatacow"
+echo
+_scratch_cycle_mount "nodatacow"
+run_tests
+
+echo
+echo "Testing with -O no-holes"
+echo
+_scratch_unmount
+_scratch_mkfs "-O no-holes" >>$seqres.full 2>&1
+_scratch_mount
+run_tests
+
+status=0
+exit
tests/btrfs/205.out new file +241 -0

241 lines not on this page · raw message

tests/btrfs/group +1 -0
index 79893747..0e77dbaa 100644
@@ -207,3 +207,4 @@
 202 auto quick subvol snapshot
 203 auto quick send clone
 204 auto quick punch
+205 auto quick clone compress
Signature · 1 line
2.25.0
Reviewed-by on v1 2/2
#1 lore raw
Quoted 7 lines · On 2/19/20 9:06 AM, [email protected] wrote:

From: Filipe Manana <[email protected]>

Test several scenarios of cloning operations where the source range includes inline extents. They used to not be supported on btrfs because their implementation was not straightforward, and therefore these operations used to fail with errno EOPNOTSUPP on older kernels.

Quoted 6 lines

This currently fails on any released kernel. It passes only when the patch with the following subject is applied:

"Btrfs: implement full reflink support for inline extents"

Thanks,

Josef