vfs: add fileattr ops

There's a substantial amount of boilerplate in filesystems handling
FS_IOC_[GS]ETFLAGS/ FS_IOC_FS[GS]ETXATTR ioctls.

Also due to userspace buffers being involved in the ioctl API this is
difficult to stack, as shown by overlayfs issues related to these ioctls.

Introduce a new internal API named "fileattr" (fsxattr can be confused with
xattr, xflags is inappropriate, since this is more than just flags).

There's significant overlap between flags and xflags and this API handles
the conversions automatically, so filesystems may choose which one to use.

In ->fileattr_get() a hint is provided to the filesystem whether flags or
xattr are being requested by userspace, but in this series this hint is
ignored by all filesystems, since generating all the attributes is cheap.

If a filesystem doesn't implemement the fileattr API, just fall back to
f_op->ioctl().  When all filesystems are converted, the fallback can be
removed.

32bit compat ioctls are now handled by the generic code as well.

Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>

2 files changed, 23 insertions, 3 deletions

diff --git a/Documentation/filesystems/locking.rst b/Documentation/filesystems/locking.rst
index b7dcc86c92a4..729ce492b662 100644
--- a/Documentation/filesystems/locking.rst
+++ b/Documentation/filesystems/locking.rst
@@ -80,13 +80,16 @@ prototypes::
 				struct file *, unsigned open_flag,
 				umode_t create_mode);
 	int (*tmpfile) (struct inode *, struct dentry *, umode_t);
+	int (*fileattr_set)(struct user_namespace *mnt_userns,
+			    struct dentry *dentry, struct fileattr *fa);
+	int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
 
 locking rules:
 	all may block
 
-============	=============================================
+=============	=============================================
 ops		i_rwsem(inode)
-============	=============================================
+=============	=============================================
 lookup:		shared
 create:		exclusive
 link:		exclusive (both)
@@ -107,7 +110,9 @@ fiemap:		no
 update_time:	no
 atomic_open:	shared (exclusive if O_CREAT is set in open flags)
 tmpfile:	no
-============	=============================================
+fileattr_get:	no or exclusive
+fileattr_set:	exclusive
+=============	=============================================
 
 
 	Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_rwsem
diff --git a/Documentation/filesystems/vfs.rst b/Documentation/filesystems/vfs.rst
index 2049bbf5e388..14c31eced416 100644
--- a/Documentation/filesystems/vfs.rst
+++ b/Documentation/filesystems/vfs.rst
@@ -441,6 +441,9 @@ As of kernel 2.6.22, the following members are defined:
 				   unsigned open_flag, umode_t create_mode);
 		int (*tmpfile) (struct user_namespace *, struct inode *, struct dentry *, umode_t);
 	        int (*set_acl)(struct user_namespace *, struct inode *, struct posix_acl *, int);
+		int (*fileattr_set)(struct user_namespace *mnt_userns,
+				    struct dentry *dentry, struct fileattr *fa);
+		int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
 	};
 
 Again, all methods are called without any locks being held, unless
@@ -588,6 +591,18 @@ otherwise noted.
 	atomically creating, opening and unlinking a file in given
 	directory.
 
+``fileattr_get``
+	called on ioctl(FS_IOC_GETFLAGS) and ioctl(FS_IOC_FSGETXATTR) to
+	retrieve miscellaneous file flags and attributes.  Also called
+	before the relevant SET operation to check what is being changed
+	(in this case with i_rwsem locked exclusive).  If unset, then
+	fall back to f_op->ioctl().
+
+``fileattr_set``
+	called on ioctl(FS_IOC_SETFLAGS) and ioctl(FS_IOC_FSSETXATTR) to
+	change miscellaneous file flags and attributes.  Callers hold
+	i_rwsem exclusive.  If unset, then fall back to f_op->ioctl().
+
 
 The Address Space Object
 ========================