Assorted cleanups in various filesystems. Currently
that pile sits in vfs.git #work.misc; if anyone wants to pick
these into relevant filesystem tree, I'll be glad to drop those
from the queue...
Diffstat:
fs/affs/namei.c | 3 ---
fs/befs/linuxvfs.c | 3 ---
fs/bfs/dir.c | 5 -----
fs/ceph/dir.c | 21 +++++++++++++--------
fs/ceph/export.c | 2 --
fs/ext4/namei.c | 2 --
fs/gfs2/export.c | 2 --
fs/gfs2/super.c | 12 +-----------
fs/hostfs/hostfs_kern.c | 8 ++------
fs/kernfs/mount.c | 3 ---
fs/nfsd/nfsctl.c | 4 ----
fs/nilfs2/namei.c | 7 +------
fs/ocfs2/dcache.c | 7 -------
fs/ocfs2/dir.c | 9 ---------
fs/orangefs/dir.c | 32 ++++++++++++--------------------
fs/proc/proc_sysctl.c | 14 ++------------
fs/reiserfs/namei.c | 7 -------
fs/udf/namei.c | 11 +----------
fs/zonefs/super.c | 2 --
security/apparmor/apparmorfs.c | 7 +------
20 files changed, 33 insertions(+), 128 deletions(-)
Shortlog:
Al Viro (22):
hostfs: use d_splice_alias() calling conventions to simplify failure exits
/proc/sys: use d_splice_alias() calling conventions to simplify failure exits
zonefs: d_splice_alias() will do the right thing on ERR_PTR() inode
udf: d_splice_alias() will do the right thing on ERR_PTR() inode
affs: d_obtain_alias(ERR_PTR(...)) will do the right thing
befs: d_obtain_alias(ERR_PTR(...)) will do the right thing
ceph: d_obtain_{alias,root}(ERR_PTR(...)) will do the right thing
gfs2: d_obtain_alias(ERR_PTR(...)) will do the right thing
kernfs: d_obtain_alias(NULL) will do the right thing...
nilfs2: d_obtain_alias(ERR_PTR(...)) will do the right thing...
udf: d_obtain_alias(ERR_PTR(...)) will do the right thing...
ext4_add_entry(): ->d_name.len is never 0
bfs_add_entry(): get rid of pointless ->d_name.len checks
__ocfs2_add_entry(), ocfs2_prepare_dir_for_insert()
reiserfs_add_entry(): get rid of pointless namelen checks
udf_fiiter_add_entry(): check for zero ->d_name.len is bogus...
get rid of passing callbacks to ceph __dentry_leases_walk()
nfsd: kill stale comment about simple_fill_super() requirements
ocfs2_find_match(): there's no such thing as NULL or negative ->d_parent
gfs2: use is_subdir()
orangefs: saner arguments passing in readdir guts
apparmorfs: don't duplicate kfree_link()
First group (1--11) is about making use of d_splice_alias() and d_obtain_alias()
calling conventions; a bunch of failure exits are pointless, since the primitive
will do the right thing. Then (12--16) killing impossible checks for dentry
name length (never 0, never exceeds what ->lookup() would reject with ENAMETOOLONG).
The rest is really without any common theme...
As far as I can tell, nothing in there has conflicts in linux-next.
Individual patches in followups. Please, review; again, if anything of
that gets picked by relevant filesystem tree, I'll gladly drop it from
this branch.
