[PATCH] VFS: Permit filesystem to override root dentry on mount

Extend the get_sb() filesystem operation to take an extra argument that
permits the VFS to pass in the target vfsmount that defines the mountpoint.

The filesystem is then required to manually set the superblock and root dentry
pointers.  For most filesystems, this should be done with simple_set_mnt()
which will set the superblock pointer and then set the root dentry to the
superblock's s_root (as per the old default behaviour).

The get_sb() op now returns an integer as there's now no need to return the
superblock pointer.

This patch permits a superblock to be implicitly shared amongst several mount
points, such as can be done with NFS to avoid potential inode aliasing.  In
such a case, simple_set_mnt() would not be called, and instead the mnt_root
and mnt_sb would be set directly.

The patch also makes the following changes:

 (*) the get_sb_*() convenience functions in the core kernel now take a vfsmount
     pointer argument and return an integer, so most filesystems have to change
     very little.

 (*) If one of the convenience function is not used, then get_sb() should
     normally call simple_set_mnt() to instantiate the vfsmount. This will
     always return 0, and so can be tail-called from get_sb().

 (*) generic_shutdown_super() now calls shrink_dcache_sb() to clean up the
     dcache upon superblock destruction rather than shrink_dcache_anon().

     This is required because the superblock may now have multiple trees that
     aren't actually bound to s_root, but that still need to be cleaned up. The
     currently called functions assume that the whole tree is rooted at s_root,
     and that anonymous dentries are not the roots of trees which results in
     dentries being left unculled.

     However, with the way NFS superblock sharing are currently set to be
     implemented, these assumptions are violated: the root of the filesystem is
     simply a dummy dentry and inode (the real inode for '/' may well be
     inaccessible), and all the vfsmounts are rooted on anonymous[*] dentries
     with child trees.

     [*] Anonymous until discovered from another tree.

 (*) The documentation has been adjusted, including the additional bit of
     changing ext2_* into foo_* in the documentation.

[[email protected]: convert ipath_fs, do other stuff]
Signed-off-by: David Howells <[email protected]>
Acked-by: Al Viro <[email protected]>
Cc: Nathan Scott <[email protected]>
Cc: Roland Dreier <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

Documentation/filesystems/Locking

@@ -142,15 +142,16 @@ see also dquot_operations section.
 
 --------------------------- file_system_type ---------------------------
 prototypes:
-	struct super_block *(*get_sb) (struct file_system_type *, int,
-			const char *, void *);
+	struct int (*get_sb) (struct file_system_type *, int,
+			const char *, void *, struct vfsmount *);
 	void (*kill_sb) (struct super_block *);
 locking rules:
 		may block	BKL
 get_sb		yes		yes
 kill_sb		yes		yes
 
-->get_sb() returns error or a locked superblock (exclusive on ->s_umount).
+->get_sb() returns error or 0 with locked superblock attached to the vfsmount
+(exclusive on ->s_umount).
 ->kill_sb() takes a write-locked superblock, does all shutdown work on it,
 unlocks and drops the reference.
 

Documentation/filesystems/porting

@@ -50,10 +50,11 @@ Turn your foo_read_super() into a function that would return 0 in case of
 success and negative number in case of error (-EINVAL unless you have more
 informative error value to report).  Call it foo_fill_super().  Now declare
 
-struct super_block foo_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+int foo_get_sb(struct file_system_type *fs_type,
+	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
 {
-	return get_sb_bdev(fs_type, flags, dev_name, data, ext2_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, foo_fill_super,
+			   mnt);
 }
 
 (or similar with s/bdev/nodev/ or s/bdev/single/, depending on the kind of

Documentation/filesystems/vfs.txt

@@ -113,8 +113,8 @@ members are defined:
 struct file_system_type {
 	const char *name;
 	int fs_flags;
-        struct super_block *(*get_sb) (struct file_system_type *, int,
-                                       const char *, void *);
+        struct int (*get_sb) (struct file_system_type *, int,
+                              const char *, void *, struct vfsmount *);
         void (*kill_sb) (struct super_block *);
         struct module *owner;
         struct file_system_type * next;

arch/ia64/kernel/perfmon.c

@@ -595,10 +595,11 @@ pfm_get_unmapped_area(struct file *file, unsigned long addr, unsigned long len,
 }
 
 
-static struct super_block *
-pfmfs_get_sb(struct file_system_type *fs_type, int flags, const char *dev_name, void *data)
+static int
+pfmfs_get_sb(struct file_system_type *fs_type, int flags, const char *dev_name, void *data,
+	     struct vfsmount *mnt)
 {
-	return get_sb_pseudo(fs_type, "pfm:", NULL, PFMFS_MAGIC);
+	return get_sb_pseudo(fs_type, "pfm:", NULL, PFMFS_MAGIC, mnt);
 }
 
 static struct file_system_type pfm_fs_type = {

arch/powerpc/platforms/cell/spufs/inode.c

@@ -436,11 +436,11 @@ spufs_fill_super(struct super_block *sb, void *data, int silent)
 	return spufs_create_root(sb, data);
 }
 
-static struct super_block *
+static int
 spufs_get_sb(struct file_system_type *fstype, int flags,
-		const char *name, void *data)
+		const char *name, void *data, struct vfsmount *mnt)
 {
-	return get_sb_single(fstype, flags, data, spufs_fill_super);
+	return get_sb_single(fstype, flags, data, spufs_fill_super, mnt);
 }
 
 static struct file_system_type spufs_type = {

drivers/infiniband/core/uverbs_main.c

@@ -821,11 +821,12 @@ static void ib_uverbs_remove_one(struct ib_device *device)
 	kref_put(&uverbs_dev->ref, ib_uverbs_release_dev);
 }
 
-static struct super_block *uverbs_event_get_sb(struct file_system_type *fs_type, int flags,
-					       const char *dev_name, void *data)
+static int uverbs_event_get_sb(struct file_system_type *fs_type, int flags,
+			       const char *dev_name, void *data,
+			       struct vfsmount *mnt)
 {
 	return get_sb_pseudo(fs_type, "infinibandevent:", NULL,
-			     INFINIBANDEVENTFS_MAGIC);
+			     INFINIBANDEVENTFS_MAGIC, mnt);
 }
 
 static struct file_system_type uverbs_event_fs = {

drivers/infiniband/hw/ipath/ipath_fs.c

@@ -542,13 +542,14 @@ static int ipathfs_fill_super(struct super_block *sb, void *data,
 	return ret;
 }
 
-static struct super_block *ipathfs_get_sb(struct file_system_type *fs_type,
-				        int flags, const char *dev_name,
-					void *data)
+static int ipathfs_get_sb(struct file_system_type *fs_type, int flags,
+			const char *dev_name, void *data, struct vfsmount *mnt)
 {
-	ipath_super = get_sb_single(fs_type, flags, data,
-				    ipathfs_fill_super);
-	return ipath_super;
+	int ret = get_sb_single(fs_type, flags, data,
+				    ipathfs_fill_super, mnt);
+	if (ret >= 0)
+		ipath_super = mnt->mnt_sb;
+	return ret;
 }
 
 static void ipathfs_kill_super(struct super_block *s)

drivers/isdn/capi/capifs.c

@@ -121,10 +121,10 @@ capifs_fill_super(struct super_block *s, void *data, int silent)
 	return -ENOMEM;
 }
 
-static struct super_block *capifs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int capifs_get_sb(struct file_system_type *fs_type,
+	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
 {
-	return get_sb_single(fs_type, flags, data, capifs_fill_super);
+	return get_sb_single(fs_type, flags, data, capifs_fill_super, mnt);
 }
 
 static struct file_system_type capifs_fs_type = {

drivers/misc/ibmasm/ibmasmfs.c

@@ -90,10 +90,11 @@ static void ibmasmfs_create_files (struct super_block *sb, struct dentry *root);
 static int ibmasmfs_fill_super (struct super_block *sb, void *data, int silent);
 
 
-static struct super_block *ibmasmfs_get_super(struct file_system_type *fst,
-			int flags, const char *name, void *data)
+static int ibmasmfs_get_super(struct file_system_type *fst,
+			int flags, const char *name, void *data,
+			struct vfsmount *mnt)
 {
-	return get_sb_single(fst, flags, data, ibmasmfs_fill_super);
+	return get_sb_single(fst, flags, data, ibmasmfs_fill_super, mnt);
 }
 
 static struct super_operations ibmasmfs_s_ops = {

drivers/oprofile/oprofilefs.c

@@ -272,10 +272,10 @@ static int oprofilefs_fill_super(struct super_block * sb, void * data, int silen
 }
 
 
-static struct super_block *oprofilefs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int oprofilefs_get_sb(struct file_system_type *fs_type,
+	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
 {
-	return get_sb_single(fs_type, flags, data, oprofilefs_fill_super);
+	return get_sb_single(fs_type, flags, data, oprofilefs_fill_super, mnt);
 }
 
 

drivers/usb/core/inode.c

@@ -543,10 +543,10 @@ static void fs_remove_file (struct dentry *dentry)
 
 /* --------------------------------------------------------------------- */
 
-static struct super_block *usb_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int usb_get_sb(struct file_system_type *fs_type,
+	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
 {
-	return get_sb_single(fs_type, flags, data, usbfs_fill_super);
+	return get_sb_single(fs_type, flags, data, usbfs_fill_super, mnt);
 }
 
 static struct file_system_type usb_fs_type = {

drivers/usb/gadget/inode.c

@@ -2070,11 +2070,11 @@ gadgetfs_fill_super (struct super_block *sb, void *opts, int silent)
 }
 
 /* "mount -t gadgetfs path /dev/gadget" ends up here */
-static struct super_block *
+static int
 gadgetfs_get_sb (struct file_system_type *t, int flags,
-		const char *path, void *opts)
+		const char *path, void *opts, struct vfsmount *mnt)
 {
-	return get_sb_single (t, flags, opts, gadgetfs_fill_super);
+	return get_sb_single (t, flags, opts, gadgetfs_fill_super, mnt);
 }
 
 static void

fs/9p/vfs_super.c

@@ -99,12 +99,13 @@ v9fs_fill_super(struct super_block *sb, struct v9fs_session_info *v9ses,
  * @flags: mount flags
  * @dev_name: device name that was mounted
  * @data: mount options
+ * @mnt: mountpoint record to be instantiated
  *
  */
 
-static struct super_block *v9fs_get_sb(struct file_system_type
-				       *fs_type, int flags,
-				       const char *dev_name, void *data)
+static int v9fs_get_sb(struct file_system_type *fs_type, int flags,
+		       const char *dev_name, void *data,
+		       struct vfsmount *mnt)
 {
 	struct super_block *sb = NULL;
 	struct v9fs_fcall *fcall = NULL;
@@ -123,17 +124,19 @@ static struct super_block *v9fs_get_sb(struct file_system_type
 
 	v9ses = kzalloc(sizeof(struct v9fs_session_info), GFP_KERNEL);
 	if (!v9ses)
-		return ERR_PTR(-ENOMEM);
+		return -ENOMEM;
 
 	if ((newfid = v9fs_session_init(v9ses, dev_name, data)) < 0) {
 		dprintk(DEBUG_ERROR, "problem initiating session\n");
-		sb = ERR_PTR(newfid);
+		retval = newfid;
 		goto out_free_session;
 	}
 
 	sb = sget(fs_type, NULL, v9fs_set_super, v9ses);
-	if (IS_ERR(sb))
+	if (IS_ERR(sb)) {
+		retval = PTR_ERR(sb);
 		goto out_close_session;
+	}
 	v9fs_fill_super(sb, v9ses, flags);
 
 	inode = v9fs_get_inode(sb, S_IFDIR | mode);
@@ -184,19 +187,19 @@ static struct super_block *v9fs_get_sb(struct file_system_type
 		goto put_back_sb;
 	}
 
-	return sb;
+	return simple_set_mnt(mnt, sb);
 
 out_close_session:
 	v9fs_session_close(v9ses);
 out_free_session:
 	kfree(v9ses);
-	return sb;
+	return retval;
 
 put_back_sb:
 	/* deactivate_super calls v9fs_kill_super which will frees the rest */
 	up_write(&sb->s_umount);
 	deactivate_super(sb);
-	return ERR_PTR(retval);
+	return retval;
 }
 
 /**

fs/adfs/super.c

@@ -470,10 +470,11 @@ static int adfs_fill_super(struct super_block *sb, void *data, int silent)
 	return -EINVAL;
 }
 
-static struct super_block *adfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int adfs_get_sb(struct file_system_type *fs_type,
+	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
 {
-	return get_sb_bdev(fs_type, flags, dev_name, data, adfs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, adfs_fill_super,
+			   mnt);
 }
 
 static struct file_system_type adfs_fs_type = {

fs/affs/super.c

@@ -524,10 +524,11 @@ affs_statfs(struct super_block *sb, struct kstatfs *buf)
 	return 0;
 }
 
-static struct super_block *affs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int affs_get_sb(struct file_system_type *fs_type,
+	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
 {
-	return get_sb_bdev(fs_type, flags, dev_name, data, affs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, affs_fill_super,
+			   mnt);
 }
 
 static struct file_system_type affs_fs_type = {

fs/afs/super.c

@@ -38,9 +38,9 @@ struct afs_mount_params {
 static void afs_i_init_once(void *foo, kmem_cache_t *cachep,
 			    unsigned long flags);
 
-static struct super_block *afs_get_sb(struct file_system_type *fs_type,
-				      int flags, const char *dev_name,
-				      void *data);
+static int afs_get_sb(struct file_system_type *fs_type,
+		      int flags, const char *dev_name,
+		      void *data, struct vfsmount *mnt);
 
 static struct inode *afs_alloc_inode(struct super_block *sb);
 
@@ -294,10 +294,11 @@ static int afs_fill_super(struct super_block *sb, void *data, int silent)
  * get an AFS superblock
  * - TODO: don't use get_sb_nodev(), but rather call sget() directly
  */
-static struct super_block *afs_get_sb(struct file_system_type *fs_type,
-				      int flags,
-				      const char *dev_name,
-				      void *options)
+static int afs_get_sb(struct file_system_type *fs_type,
+		      int flags,
+		      const char *dev_name,
+		      void *options,
+		      struct vfsmount *mnt)
 {
 	struct afs_mount_params params;
 	struct super_block *sb;
@@ -311,7 +312,7 @@ static struct super_block *afs_get_sb(struct file_system_type *fs_type,
 	ret = afscm_start();
 	if (ret < 0) {
 		_leave(" = %d", ret);
-		return ERR_PTR(ret);
+		return ret;
 	}
 
 	/* parse the options */
@@ -348,18 +349,19 @@ static struct super_block *afs_get_sb(struct file_system_type *fs_type,
 		goto error;
 	}
 	sb->s_flags |= MS_ACTIVE;
+	simple_set_mnt(mnt, sb);
 
 	afs_put_volume(params.volume);
 	afs_put_cell(params.default_cell);
-	_leave(" = %p", sb);
-	return sb;
+	_leave(" = 0 [%p]", 0, sb);
+	return 0;
 
  error:
 	afs_put_volume(params.volume);
 	afs_put_cell(params.default_cell);
 	afscm_stop();
 	_leave(" = %d", ret);
-	return ERR_PTR(ret);
+	return ret;
 } /* end afs_get_sb() */
 
 /*****************************************************************************/

fs/autofs/init.c

@@ -14,10 +14,10 @@
 #include <linux/init.h>
 #include "autofs_i.h"
 
-static struct super_block *autofs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int autofs_get_sb(struct file_system_type *fs_type,
+	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
 {
-	return get_sb_nodev(fs_type, flags, data, autofs_fill_super);
+	return get_sb_nodev(fs_type, flags, data, autofs_fill_super, mnt);
 }
 
 static struct file_system_type autofs_fs_type = {

fs/autofs4/init.c

@@ -14,10 +14,10 @@
 #include <linux/init.h>
 #include "autofs_i.h"
 
-static struct super_block *autofs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int autofs_get_sb(struct file_system_type *fs_type,
+	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
 {
-	return get_sb_nodev(fs_type, flags, data, autofs4_fill_super);
+	return get_sb_nodev(fs_type, flags, data, autofs4_fill_super, mnt);
 }
 
 static struct file_system_type autofs_fs_type = {

fs/befs/linuxvfs.c

@@ -899,11 +899,12 @@ befs_statfs(struct super_block *sb, struct kstatfs *buf)
 	return 0;
 }
 
-static struct super_block *
+static int
 befs_get_sb(struct file_system_type *fs_type, int flags, const char *dev_name,
-	    void *data)
+	    void *data, struct vfsmount *mnt)
 {
-	return get_sb_bdev(fs_type, flags, dev_name, data, befs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, befs_fill_super,
+			   mnt);
 }
 
 static struct file_system_type befs_fs_type = {

fs/bfs/inode.c

@@ -410,10 +410,10 @@ static int bfs_fill_super(struct super_block *s, void *data, int silent)
 	return -EINVAL;
 }
 
-static struct super_block *bfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int bfs_get_sb(struct file_system_type *fs_type,
+	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
 {
-	return get_sb_bdev(fs_type, flags, dev_name, data, bfs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, bfs_fill_super, mnt);
 }
 
 static struct file_system_type bfs_fs_type = {