I recently found and fixed a bug in btrfs where it wasn't updating the citme on the target inode when unlinking [1]. Add a fstest for this. [1]: https://lore.kernel.org/linux-btrfs/20240812-btrfs-unlink-v1-1-ee5c2ef538eb@xxxxxxxxxx/ Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx> --- HCH suggested I roll a fstest for this problem that I found in btrfs the other day. This just creates a file and a hardlink to it, statx's it and then unlinks the hardlink and statx's it again. The ctimes are then compared. --- Changes in v2: - Turn it into a shell script. - Link to v1: https://lore.kernel.org/r/20240813-master-v1-1-862678cc4000@xxxxxxxxxx --- tests/generic/755 | 38 ++++++++++++++++++++++++++++++++++++++ tests/generic/755.out | 2 ++ 2 files changed, 40 insertions(+) diff --git a/tests/generic/755 b/tests/generic/755 new file mode 100755 index 000000000000..68da3b20073f --- /dev/null +++ b/tests/generic/755 @@ -0,0 +1,38 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2024, Jeff Layton <jlayton@xxxxxxxxxx> +# +# FS QA Test No. 755 +# +# Create a file, stat it and then unlink it. Does the ctime of the +# target inode change? +# +. ./common/preamble +_begin_fstest auto quick + +_require_test +_require_test_program unlink-ctime + +testfile="$TEST_DIR/unlink-ctime1.$$" +testlink="$TEST_DIR/unlink-ctime2.$$" + +rm -f $testfile $testlink +touch $testfile +ln $testfile $testlink + +time1=$(stat -c "%Z" $testfile) + +sleep 2 +unlink $testlink + +time2=$(stat -c "%Z" $testfile) + +unlink $testfile + +if [ $time1 -eq $time2 ]; then + echo "Target's ctime did not change after unlink!" +fi + +echo Silence is golden +status=0 +exit diff --git a/tests/generic/755.out b/tests/generic/755.out new file mode 100644 index 000000000000..7c9ea51cd298 --- /dev/null +++ b/tests/generic/755.out @@ -0,0 +1,2 @@ +QA output created by 755 +Silence is golden --- base-commit: f5ada754d5838d29fd270257003d0d123a9d1cd2 change-id: 20240813-master-e3b46de630bd Best regards, -- Jeff Layton <jlayton@xxxxxxxxxx>
