On Tue, Apr 26, 2022 at 08:52:09PM -0400, Theodore Ts'o wrote: > The ext4/054 and ext4/055 tests create a scratch file system with a 1k > block size. This is not compatible with mounting with the DAX option, > which requires a block size equal to the page size (which is 4k on > x86). Following up on a point I made during this morning's ext4 concall -- I avoided these problems by adding a cli option to mkfs.xfs to set the DAX flag on the root directory (since it's advisory and propagates to all new children) and then updated my fstests config to use that. No more mount failures due to blocksize != pagesize! :) --D > Also, the ext4/054 test doesn't use the test device, so remove the > _require_test declaration. > > Signed-off-by: Theodore Ts'o <tytso@xxxxxxx> > --- > tests/ext4/054 | 2 +- > tests/ext4/055 | 3 ++- > 2 files changed, 3 insertions(+), 2 deletions(-) > > diff --git a/tests/ext4/054 b/tests/ext4/054 > index 9a11719f..6c722f32 100755 > --- a/tests/ext4/054 > +++ b/tests/ext4/054 > @@ -19,8 +19,8 @@ _begin_fstest auto quick dangerous_fuzzers > > # real QA test starts here > _supported_fs ext4 > -_require_test > _require_scratch_nocheck > +_exclude_scratch_mount_option dax > _require_xfs_io_command "falloc" > _require_xfs_io_command "pwrite" > _require_xfs_io_command "fsync" > diff --git a/tests/ext4/055 b/tests/ext4/055 > index 8f466f1b..1ae42b89 100755 > --- a/tests/ext4/055 > +++ b/tests/ext4/055 > @@ -17,8 +17,9 @@ > _begin_fstest auto quota > > # real QA test starts here > -_require_scratch_nocheck > _supported_fs ext4 > +_require_scratch_nocheck > +_exclude_scratch_mount_option dax > _require_user fsgqa > _require_user fsgqa2 > _require_command "$DEBUGFS_PROG" debugfs > -- > 2.31.0 >
