[10/13] xfs/443: use file allocation unit, not dbsize

Commit Message

Darrick J. Wong Dec. 30, 2022, 10:20 p.m. UTC

From: Darrick J. Wong <[e-mail]>

We can only punch in units of file allocation boundaries, so update this
test to use that instead of the fs blocksize.

Signed-off-by: Darrick J. Wong <[e-mail]>
---
 tests/xfs/443 |    9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

Patch

diff --git a/tests/xfs/443 b/tests/xfs/443
index 56828decae..ab3cda59f3 100755
--- a/tests/xfs/443
+++ b/tests/xfs/443
@@ -40,14 +40,15 @@  _scratch_mount
 
 file1=$SCRATCH_MNT/file1
 file2=$SCRATCH_MNT/file2
+file_blksz=$(_get_file_block_size $SCRATCH_MNT)
 
 # The goal is run an extent swap where one of the associated files has the
 # minimum number of extents to remain in btree format. First, create a couple
 # files with large enough extent counts (200 or so should be plenty) to ensure
 # btree format on the largest possible inode size filesystems.
-$XFS_IO_PROG -fc "falloc 0 $((400 * dbsize))" $file1
+$XFS_IO_PROG -fc "falloc 0 $((400 * file_blksz))" $file1
 $here/src/punch-alternating $file1
-$XFS_IO_PROG -fc "falloc 0 $((400 * dbsize))" $file2
+$XFS_IO_PROG -fc "falloc 0 $((400 * file_blksz))" $file2
 $here/src/punch-alternating $file2
 
 # Now run an extent swap at every possible extent count down to 0. Depending on
@@ -55,12 +56,12 @@  $here/src/punch-alternating $file2
 # btree format.
 for i in $(seq 1 2 399); do
 	# punch one extent from the tmpfile and swap
-	$XFS_IO_PROG -c "fpunch $((i * dbsize)) $dbsize" $file2
+	$XFS_IO_PROG -c "fpunch $((i * file_blksz)) $file_blksz" $file2
 	$XFS_IO_PROG -c "swapext $file2" $file1
 
 	# punch the same extent from the old fork (now in file2) to resync the
 	# extent counts and repeat
-	$XFS_IO_PROG -c "fpunch $((i * dbsize)) $dbsize" $file2
+	$XFS_IO_PROG -c "fpunch $((i * file_blksz)) $file_blksz" $file2
 done
 
 # sanity check that no extents are left over