new ->follow_link() and ->put_link() calling conventions
authorAl Viro <viro@zeniv.linux.org.uk>
Sat, 2 May 2015 17:32:22 +0000 (13:32 -0400)
committerAl Viro <viro@zeniv.linux.org.uk>
Mon, 11 May 2015 02:19:45 +0000 (22:19 -0400)
commit680baacbca69d18a6d7315374ad83d05ac9c0977
treea69822ef5234d4a308b780ff51c5d9b77bd3a89b
parent46afd6f61cc33ae4b3a2aed4bb454d11d4114c27
new ->follow_link() and ->put_link() calling conventions

a) instead of storing the symlink body (via nd_set_link()) and returning
an opaque pointer later passed to ->put_link(), ->follow_link() _stores_
that opaque pointer (into void * passed by address by caller) and returns
the symlink body.  Returning ERR_PTR() on error, NULL on jump (procfs magic
symlinks) and pointer to symlink body for normal symlinks.  Stored pointer
is ignored in all cases except the last one.

Storing NULL for opaque pointer (or not storing it at all) means no call
of ->put_link().

b) the body used to be passed to ->put_link() implicitly (via nameidata).
Now only the opaque pointer is.  In the cases when we used the symlink body
to free stuff, ->follow_link() now should store it as opaque pointer in addition
to returning it.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
31 files changed:
Documentation/filesystems/Locking
Documentation/filesystems/vfs.txt
drivers/staging/lustre/lustre/llite/symlink.c
fs/9p/vfs_inode.c
fs/9p/vfs_inode_dotl.c
fs/autofs4/symlink.c
fs/befs/linuxvfs.c
fs/cifs/cifsfs.h
fs/cifs/link.c
fs/configfs/symlink.c
fs/ecryptfs/inode.c
fs/ext4/symlink.c
fs/f2fs/namei.c
fs/fuse/dir.c
fs/gfs2/inode.c
fs/hostfs/hostfs_kern.c
fs/hppfs/hppfs.c
fs/kernfs/symlink.c
fs/libfs.c
fs/namei.c
fs/nfs/symlink.c
fs/overlayfs/inode.c
fs/proc/base.c
fs/proc/inode.c
fs/proc/namespaces.c
fs/proc/self.c
fs/proc/thread_self.c
fs/xfs/xfs_iops.c
include/linux/fs.h
include/linux/namei.h
mm/shmem.c