Skip to contentSkip to conversation

NFS: Permit filesystem to override root dentry on mount [try #7]

PATCH In review David Howells 3 participants 14 messages Mar 8, 2006 – Mar 9, 2006

Subsystems (42)
lore
Share & tools
lore

David Howells

author maintainer
#0 lore raw

PATCH 1/6 NFS: Permit filesystem to override root dentry on mount [try #7]

The attached patch extends 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).

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 (see patch #5). In such a case, simple_set_mnt() would not be called, and instead the mnt_root and mnt_sb would be set directly.

This patch also changes the superblock cleanup routine to make it use shrink_dcache_sb() instead of shrink_dcache_anon(). This required is 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.

Following discussion with Al Viro, the following changes [try #2] were made to the previous attempt at this set of patches:

(*) The vfsmount is now passed into the get_sb() method for a filesystem
instead of passing a pointer to a pointer to a dentry into which get_sb() could stick a root dentry if it wanted. get_sb() now instantiates the superblock and root dentry pointers in the vfsmount itself.

(*) The get_sb() method now returns an integer (0 or -ve error number) rather
than the superblock pointer or cast error number.

(*) 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_parent() and shrink_dcache_anon(). This is because, as far as I can tell, the current code assumes that all the dentries will be linked into a tree depending from sb->s_root, and that anonymous dentries won't have children.

However, with the way these patches implement NFS superblock sharing, 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.

Further changes [try #3] that have been made:

(*) The patches are now against Trond's NFS git tree, so won't apply to
Linus's tree.

(*) Core documentation changes have been moved to the first patch.

Further changes [try #4]:

(*) shrink_dcache_parent() has been reintroduced to handle final cleanup of
filesystem types such as nfsd and binfmt_misc that have pinned dentries.

(*) get_sb_bdev() has been fixed to handle the superblock sharing case
correctly (it should not go down the error path).

Further changes [try #5]:

(*) Make futex_get_sb() static once again.

(*) Fold together the patch that changes the core and the patch that modifies
all the filesystems to cope with the core change as some people are unable to cope with two interdependent patches.

---

 Documentation/filesystems/Locking         |    7 +-
 Documentation/filesystems/porting         |    7 +-
 Documentation/filesystems/vfs.txt         |    4 +
 arch/ia64/kernel/perfmon.c                |    7 +-
 arch/powerpc/platforms/cell/spufs/inode.c |    6 +-
 drivers/infiniband/core/uverbs_main.c     |    7 +-
 drivers/isdn/capi/capifs.c                |    6 +-
 drivers/misc/ibmasm/ibmasmfs.c            |    7 +-
 drivers/oprofile/oprofilefs.c             |    6 +-
 drivers/usb/core/inode.c                  |    6 +-
 drivers/usb/gadget/inode.c                |    6 +-
 fs/9p/vfs_super.c                         |   17 +++--
 fs/adfs/super.c                           |    7 +-
 fs/affs/super.c                           |    7 +-
 fs/afs/super.c                            |   24 ++++----
 fs/autofs/init.c                          |    6 +-
 fs/autofs4/init.c                         |    6 +-
 fs/befs/linuxvfs.c                        |    7 +-
 fs/bfs/inode.c                            |    6 +-
 fs/binfmt_misc.c                          |    6 +-
 fs/block_dev.c                            |    6 +-
 fs/cifs/cifsfs.c                          |   10 ++-
 fs/coda/inode.c                           |    6 +-
 fs/configfs/mount.c                       |    6 +-
 fs/cramfs/inode.c                         |    7 +-
 fs/debugfs/inode.c                        |    8 +--
 fs/devfs/base.c                           |    8 +--
 fs/devpts/inode.c                         |    6 +-
 fs/efs/super.c                            |    6 +-
 fs/eventpoll.c                            |   13 ++--
 fs/ext2/super.c                           |    6 +-
 fs/ext3/super.c                           |    6 +-
 fs/freevxfs/vxfs_super.c                  |    7 +-
 fs/fuse/inode.c                           |    8 +--
 fs/hfs/super.c                            |    7 +-
 fs/hfsplus/super.c                        |    8 ++-
 fs/hostfs/hostfs_kern.c                   |    8 +--
 fs/hpfs/super.c                           |    7 +-
 fs/hppfs/hppfs_kern.c                     |    8 +--
 fs/hugetlbfs/inode.c                      |    6 +-
 fs/inotify.c                              |    6 +-
 fs/isofs/inode.c                          |    7 +-
 fs/jffs/inode-v23.c                       |    7 +-
 fs/jffs2/super.c                          |   49 +++++++++------
 fs/jfs/super.c                            |    7 +-
 fs/libfs.c                                |   12 ++--
 fs/minix/inode.c                          |    7 +-
 fs/msdos/namei.c                          |    9 ++-
 fs/namespace.c                            |    9 +++
 fs/ncpfs/inode.c                          |    6 +-
 fs/nfs/inode.c                            |   82 +++++++++++++++-----------
 fs/nfsd/nfsctl.c                          |    6 +-
 fs/ntfs/super.c                           |    7 +-
 fs/ocfs2/dlm/dlmfs.c                      |    6 +-
 fs/ocfs2/super.c                          |   12 ++--
 fs/openpromfs/inode.c                     |    6 +-
 fs/pipe.c                                 |    6 +-
 fs/proc/root.c                            |    6 +-
 fs/qnx4/inode.c                           |    7 +-
 fs/ramfs/inode.c                          |   13 ++--
 fs/reiserfs/super.c                       |    9 ++-
 fs/relayfs/inode.c                        |    8 +--
 fs/romfs/inode.c                          |    7 +-
 fs/smbfs/inode.c                          |    6 +-
 fs/super.c                                |   92 ++++++++++++++++-------------
 fs/sysfs/mount.c                          |    6 +-
 fs/sysv/super.c                           |   13 ++--
 fs/udf/super.c                            |    6 +-
 fs/ufs/super.c                            |    6 +-
 fs/vfat/namei.c                           |    9 ++-
 fs/xfs/linux-2.6/xfs_super.c              |    8 ++-
 include/linux/fs.h                        |   25 +++++---
 include/linux/ramfs.h                     |    4 +
 ipc/mqueue.c                              |    8 +--
 kernel/cpuset.c                           |    8 +--
 kernel/futex.c                            |    8 +--
 mm/shmem.c                                |    6 +-
 net/socket.c                              |    7 +-
 net/sunrpc/rpc_pipe.c                     |    6 +-
 security/inode.c                          |    8 +--
 security/selinux/selinuxfs.c              |    7 +-
 81 files changed, 455 insertions(+), 374 deletions(-)
Documentation/filesystems/Locking +4 -3
index 1045da5..3abf08f 100644
@@ -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 +4 -3
index 2f38846..ae8db97 100644
@@ -50,10 +50,11 @@ Turn your foo_read_super() into a functi
 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, ext2_fill_super,
+			   mnt);
 }
 
 (or similar with s/bdev/nodev/ or s/bdev/single/, depending on the kind of
Documentation/filesystems/vfs.txt +2 -2
index e56e842..9b1748e 100644
@@ -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 +4 -3
index 9c5194b..d9d173b 100644
@@ -595,10 +595,11 @@ pfm_get_unmapped_area(struct file *file,
 }
 
 
-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 +3 -3
index b3962c3..c2c4e0f 100644
@@ -428,11 +428,11 @@ spufs_fill_super(struct super_block *sb,
 	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 +4 -3
index 903f85a..ed962e1 100644
@@ -818,11 +818,12 @@ static void ib_uverbs_remove_one(struct 
 	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/isdn/capi/capifs.c +3 -3
index 0a37ade..9ea6bd0 100644
@@ -121,10 +121,10 @@ fail:
 	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 +4 -3
index 5c550fc..3fe4a95 100644
@@ -90,10 +90,11 @@ static void ibmasmfs_create_files (struc
 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 +3 -3
index d6bae69..fdfb7bd 100644
@@ -272,10 +272,10 @@ static int oprofilefs_fill_super(struct 
 }
 
 
-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 +3 -3
index 3cf945c..95f5ad9 100644
@@ -543,10 +543,10 @@ static void fs_remove_file (struct dentr
 
 /* --------------------------------------------------------------------- */
 
-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 +3 -3
index 0aab7d2..5cc09a3 100644
@@ -2065,11 +2065,11 @@ gadgetfs_fill_super (struct super_block 
 }
 
 /* "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 +10 -7
index 2c4fa75..bca1efb 100644
@@ -100,12 +100,13 @@ v9fs_fill_super(struct super_block *sb, 
  * @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;
@@ -124,15 +125,17 @@ static struct super_block *v9fs_get_sb(s
 
 	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");
 		kfree(v9ses);
-		return ERR_PTR(newfid);
+		return newfid;
 	}
 
 	sb = sget(fs_type, NULL, v9fs_set_super, v9ses);
+	if (IS_ERR(sb))
+		return PTR_ERR(sb);
 
 	v9fs_fill_super(sb, v9ses, flags);
 
@@ -180,13 +183,13 @@ static struct super_block *v9fs_get_sb(s
 		goto put_back_sb;
 	}
 
-	return sb;
+	return simple_set_mnt(mnt, sb);
 
 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 +4 -3
index 2439632..ff07c9b 100644
@@ -469,10 +469,11 @@ error:
 	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 +4 -3
index aaec015..337893a 100644
@@ -523,10 +523,11 @@ affs_statfs(struct super_block *sb, stru
 	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 +13 -11
index d6fa8e5..b809ac2 100644
@@ -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_b
  * 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(st
 	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(st
 		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 +3 -3
index b977ece..aca1237 100644
@@ -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 +3 -3
index acecec8..5d91933 100644
@@ -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 +4 -3
index 2d365cb..617742f 100644
@@ -898,11 +898,12 @@ befs_statfs(struct super_block *sb, stru
 	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 +3 -3
index 3af6c73..d6b6592 100644
@@ -409,10 +409,10 @@ out:
 	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 = {
fs/binfmt_misc.c +3 -3
index 6a7b730..480ea07 100644
@@ -740,10 +740,10 @@ static int bm_fill_super(struct super_bl
 	return err;
 }
 
-static struct super_block *bm_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int bm_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, bm_fill_super);
+	return get_sb_single(fs_type, flags, data, bm_fill_super, mnt);
 }
 
 static struct linux_binfmt misc_format = {
fs/block_dev.c +3 -3
index 6e50346..101fb61 100644
@@ -300,10 +300,10 @@ static struct super_operations bdev_sops
 	.clear_inode = bdev_clear_inode,
 };
 
-static struct super_block *bd_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int bd_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, "bdev:", &bdev_sops, 0x62646576);
+	return get_sb_pseudo(fs_type, "bdev:", &bdev_sops, 0x62646576, mnt);
 }
 
 static struct file_system_type bd_type = {
fs/cifs/cifsfs.c +5 -5
index 79eeccd..25064b6 100644
@@ -465,9 +465,9 @@ struct super_operations cifs_super_ops =
 	.remount_fs = cifs_remount,
 };
 
-static struct super_block *
+static int
 cifs_get_sb(struct file_system_type *fs_type,
-	    int flags, const char *dev_name, void *data)
+	    int flags, const char *dev_name, void *data, struct vfsmount *mnt)
 {
 	int rc;
 	struct super_block *sb = sget(fs_type, NULL, set_anon_super, NULL);
@@ -475,7 +475,7 @@ cifs_get_sb(struct file_system_type *fs_
 	cFYI(1, ("Devname: %s flags: %d ", dev_name, flags));
 
 	if (IS_ERR(sb))
-		return sb;
+		return PTR_ERR(sb);
 
 	sb->s_flags = flags;
 
@@ -483,10 +483,10 @@ cifs_get_sb(struct file_system_type *fs_
 	if (rc) {
 		up_write(&sb->s_umount);
 		deactivate_super(sb);
-		return ERR_PTR(rc);
+		return rc;
 	}
 	sb->s_flags |= MS_ACTIVE;
-	return sb;
+	return simple_set_mnt(mnt, sb);
 }
 
 static ssize_t cifs_file_writev(struct file *file, const struct iovec *iov,
fs/coda/inode.c +3 -3
index 04a73fb..f3fb09d 100644
@@ -305,10 +305,10 @@ static int coda_statfs(struct super_bloc
 
 /* init_coda: used by filesystems.c to register coda */
 
-static struct super_block *coda_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int coda_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, coda_fill_super);
+	return get_sb_nodev(fs_type, flags, data, coda_fill_super, mnt);
 }
 
 struct file_system_type coda_fs_type = {
fs/configfs/mount.c +3 -3
index f920d30..94dab7b 100644
@@ -103,10 +103,10 @@ static int configfs_fill_super(struct su
 	return 0;
 }
 
-static struct super_block *configfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int configfs_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, configfs_fill_super);
+	return get_sb_single(fs_type, flags, data, configfs_fill_super, mnt);
 }
 
 static struct file_system_type configfs_fs_type = {
fs/cramfs/inode.c +4 -3
index 7fe8541..6c62138 100644
@@ -529,10 +529,11 @@ static struct super_operations cramfs_op
 	.statfs		= cramfs_statfs,
 };
 
-static struct super_block *cramfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int cramfs_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, cramfs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, cramfs_fill_super,
+			   mnt);
 }
 
 static struct file_system_type cramfs_fs_type = {
fs/debugfs/inode.c +4 -4
index d4f1a2c..39f25bb 100644
@@ -110,11 +110,11 @@ static int debug_fill_super(struct super
 	return simple_fill_super(sb, DEBUGFS_MAGIC, debug_files);
 }
 
-static struct super_block *debug_get_sb(struct file_system_type *fs_type,
-				        int flags, const char *dev_name,
-					void *data)
+static int debug_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, debug_fill_super);
+	return get_sb_single(fs_type, flags, data, debug_fill_super, mnt);
 }
 
 static struct file_system_type debug_fs_type = {
fs/devfs/base.c +4 -4
index b621521..249e495 100644
@@ -2549,11 +2549,11 @@ static int devfs_fill_super(struct super
 	return -EINVAL;
 }				/*  End Function devfs_fill_super  */
 
-static struct super_block *devfs_get_sb(struct file_system_type *fs_type,
-					int flags, const char *dev_name,
-					void *data)
+static int devfs_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, devfs_fill_super);
+	return get_sb_single(fs_type, flags, data, devfs_fill_super, mnt);
 }
 
 static struct file_system_type devfs_fs_type = {
fs/devpts/inode.c +3 -3
index bfb8a23..9489398 100644
@@ -108,10 +108,10 @@ fail:
 	return -ENOMEM;
 }
 
-static struct super_block *devpts_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int devpts_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, devpts_fill_super);
+	return get_sb_single(fs_type, flags, data, devpts_fill_super, mnt);
 }
 
 static struct file_system_type devpts_fs_type = {
fs/efs/super.c +3 -3
index afc4891..b1b5e31 100644
@@ -18,10 +18,10 @@
 static int efs_statfs(struct super_block *s, struct kstatfs *buf);
 static int efs_fill_super(struct super_block *s, void *d, int silent);
 
-static struct super_block *efs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int efs_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, efs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, efs_fill_super, mnt);
 }
 
 static struct file_system_type efs_fs_type = {
fs/eventpoll.c +7 -6
index 4284cd3..02c20aa 100644
@@ -267,9 +267,9 @@ static int ep_poll(struct eventpoll *ep,
 		   int maxevents, long timeout);
 static int eventpollfs_delete_dentry(struct dentry *dentry);
 static struct inode *ep_eventpoll_inode(void);
-static struct super_block *eventpollfs_get_sb(struct file_system_type *fs_type,
-					      int flags, const char *dev_name,
-					      void *data);
+static int eventpollfs_get_sb(struct file_system_type *fs_type,
+			      int flags, const char *dev_name,
+			      void *data, struct vfsmount *mnt);
 
 /*
  * This semaphore is used to serialize ep_free() and eventpoll_release_file().
@@ -1603,11 +1603,12 @@ eexit_1:
 }
 
 
-static struct super_block *
+static int
 eventpollfs_get_sb(struct file_system_type *fs_type, int flags,
-		   const char *dev_name, void *data)
+		   const char *dev_name, void *data, struct vfsmount *mnt)
 {
-	return get_sb_pseudo(fs_type, "eventpoll:", NULL, EVENTPOLLFS_MAGIC);
+	return get_sb_pseudo(fs_type, "eventpoll:", NULL, EVENTPOLLFS_MAGIC,
+			     mnt);
 }
 
 
fs/ext2/super.c +3 -3
index cb6f9bd..caaf82a 100644
@@ -1089,10 +1089,10 @@ static int ext2_statfs (struct super_blo
 	return 0;
 }
 
-static struct super_block *ext2_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int ext2_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, ext2_fill_super, mnt);
 }
 
 #ifdef CONFIG_QUOTA
fs/ext3/super.c +3 -3
index 56bf765..3d4dcc4 100644
@@ -2650,10 +2650,10 @@ out:
 
 #endif
 
-static struct super_block *ext3_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int ext3_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, ext3_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, ext3_fill_super, mnt);
 }
 
 static struct file_system_type ext3_fs_type = {
fs/freevxfs/vxfs_super.c +4 -3
index 6aa6fbe..a7d8c23 100644
@@ -241,10 +241,11 @@ out:
 /*
  * The usual module blurb.
  */
-static struct super_block *vxfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int vxfs_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, vxfs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, vxfs_fill_super,
+			   mnt);
 }
 
 static struct file_system_type vxfs_fs_type = {
fs/fuse/inode.c +4 -4
index 879e6fb..c9431dd 100644
@@ -604,11 +604,11 @@ static int fuse_fill_super(struct super_
 	return err;
 }
 
-static struct super_block *fuse_get_sb(struct file_system_type *fs_type,
-				       int flags, const char *dev_name,
-				       void *raw_data)
+static int fuse_get_sb(struct file_system_type *fs_type,
+		       int flags, const char *dev_name,
+		       void *raw_data, struct vfsmount *mnt)
 {
-	return get_sb_nodev(fs_type, flags, raw_data, fuse_fill_super);
+	return get_sb_nodev(fs_type, flags, raw_data, fuse_fill_super, mnt);
 }
 
 static struct file_system_type fuse_fs_type = {
fs/hfs/super.c +4 -3
index 1181d11..ee5b80a 100644
@@ -413,10 +413,11 @@ bail:
 	return res;
 }
 
-static struct super_block *hfs_get_sb(struct file_system_type *fs_type,
-				      int flags, const char *dev_name, void *data)
+static int hfs_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, hfs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, hfs_fill_super, mnt);
 }
 
 static struct file_system_type hfs_fs_type = {
fs/hfsplus/super.c +5 -3
index 7843f79..0ed8b7e 100644
@@ -450,10 +450,12 @@ static void hfsplus_destroy_inode(struct
 
 #define HFSPLUS_INODE_SIZE	sizeof(struct hfsplus_inode_info)
 
-static struct super_block *hfsplus_get_sb(struct file_system_type *fs_type,
-					  int flags, const char *dev_name, void *data)
+static int hfsplus_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, hfsplus_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, hfsplus_fill_super,
+			   mnt);
 }
 
 static struct file_system_type hfsplus_fs_type = {
fs/hostfs/hostfs_kern.c +4 -4
index b3ad0bd..ce63271 100644
@@ -993,11 +993,11 @@ static int hostfs_fill_sb_common(struct 
 	return(err);
 }
 
-static struct super_block *hostfs_read_sb(struct file_system_type *type,
-					     int flags, const char *dev_name,
-					     void *data)
+static int hostfs_read_sb(struct file_system_type *type,
+			  int flags, const char *dev_name,
+			  void *data, struct vfsmount *mnt)
 {
-	return(get_sb_nodev(type, flags, data, hostfs_fill_sb_common));
+	return get_sb_nodev(type, flags, data, hostfs_fill_sb_common, mnt);
 }
 
 static struct file_system_type hostfs_type = {
fs/hpfs/super.c +4 -3
index 63e88d7..432eb4b 100644
@@ -661,10 +661,11 @@ bail0:
 	return -EINVAL;
 }
 
-static struct super_block *hpfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int hpfs_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, hpfs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, hpfs_fill_super,
+			   mnt);
 }
 
 static struct file_system_type hpfs_fs_type = {
fs/hppfs/hppfs_kern.c +4 -4
index a44dc58..c0e4b8d 100644
@@ -769,11 +769,11 @@ static int hppfs_fill_super(struct super
 	return(err);
 }
 
-static struct super_block *hppfs_read_super(struct file_system_type *type,
-					     int flags, const char *dev_name,
-					     void *data)
+static int hppfs_read_super(struct file_system_type *type,
+			    int flags, const char *dev_name,
+			    void *data, struct vfsmount *mnt)
 {
-	return(get_sb_nodev(type, flags, data, hppfs_fill_super));
+	return get_sb_nodev(type, flags, data, hppfs_fill_super, mnt);
 }
 
 static struct file_system_type hppfs_type = {
fs/hugetlbfs/inode.c +3 -3
index b351952..fd1c966 100644
@@ -757,10 +757,10 @@ void hugetlb_put_quota(struct address_sp
 	}
 }
 
-static struct super_block *hugetlbfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int hugetlbfs_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, hugetlbfs_fill_super);
+	return get_sb_nodev(fs_type, flags, data, hugetlbfs_fill_super, mnt);
 }
 
 static struct file_system_type hugetlbfs_fs_type = {
fs/inotify.c +3 -3
index 3041503..9559096 100644
@@ -1030,11 +1030,11 @@ out:
 	return ret;
 }
 
-static struct super_block *
+static int
 inotify_get_sb(struct file_system_type *fs_type, int flags,
-	       const char *dev_name, void *data)
+	       const char *dev_name, void *data, struct vfsmount *mnt)
 {
-    return get_sb_pseudo(fs_type, "inotify", NULL, 0xBAD1DEA);
+	return get_sb_pseudo(fs_type, "inotify", NULL, 0xBAD1DEA, mnt);
 }
 
 static struct file_system_type inotify_fs_type = {
fs/isofs/inode.c +4 -3
index 298f08b..9dc0091 100644
@@ -1398,10 +1398,11 @@ struct inode *isofs_iget(struct super_bl
 	return inode;
 }
 
-static struct super_block *isofs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int isofs_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, isofs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, isofs_fill_super,
+			   mnt);
 }
 
 static struct file_system_type iso9660_fs_type = {
fs/jffs/inode-v23.c +4 -3
index fc3855a..e2d9c5c 100644
@@ -1785,10 +1785,11 @@ static struct super_operations jffs_ops 
 	.remount_fs	= jffs_remount,
 };
 
-static struct super_block *jffs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int jffs_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, jffs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, jffs_fill_super,
+			   mnt);
 }
 
 static struct file_system_type jffs_fs_type = {
fs/jffs2/super.c +27 -22

131 lines not on this page · raw message

fs/jfs/super.c +4 -3
index 8d31f13..8f2e8ca 100644
@@ -541,10 +541,11 @@ static void jfs_unlockfs(struct super_bl
 	}
 }
 
-static struct super_block *jfs_get_sb(struct file_system_type *fs_type, 
-	int flags, const char *dev_name, void *data)
+static int jfs_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, jfs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, jfs_fill_super,
+			   mnt);
 }
 
 static int jfs_sync_fs(struct super_block *sb, int wait)
fs/libfs.c +6 -6
index 71fd08f..7a60a20 100644
@@ -194,9 +194,9 @@ struct inode_operations simple_dir_inode
  * Common helper for pseudo-filesystems (sockfs, pipefs, bdev - stuff that
  * will never be mountable)
  */
-struct super_block *
-get_sb_pseudo(struct file_system_type *fs_type, char *name,
-	struct super_operations *ops, unsigned long magic)
+int get_sb_pseudo(struct file_system_type *fs_type, char *name,
+	struct super_operations *ops, unsigned long magic,
+	struct vfsmount *mnt)
 {
 	struct super_block *s = sget(fs_type, NULL, set_anon_super, NULL);
 	static struct super_operations default_ops = {.statfs = simple_statfs};
@@ -205,7 +205,7 @@ get_sb_pseudo(struct file_system_type *f
 	struct qstr d_name = {.name = name, .len = strlen(name)};
 
 	if (IS_ERR(s))
-		return s;
+		return PTR_ERR(s);
 
 	s->s_flags = MS_NOUSER;
 	s->s_maxbytes = ~0ULL;
@@ -230,12 +230,12 @@ get_sb_pseudo(struct file_system_type *f
 	d_instantiate(dentry, root);
 	s->s_root = dentry;
 	s->s_flags |= MS_ACTIVE;
-	return s;
+	return simple_set_mnt(mnt, s);
 
 Enomem:
 	up_write(&s->s_umount);
 	deactivate_super(s);
-	return ERR_PTR(-ENOMEM);
+	return -ENOMEM;
 }
 
 int simple_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry)
fs/minix/inode.c +4 -3
index 790cc0d..6d74911 100644
@@ -558,10 +558,11 @@ void minix_truncate(struct inode * inode
 		V2_minix_truncate(inode);
 }
 
-static struct super_block *minix_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int minix_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, minix_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, minix_fill_super,
+			   mnt);
 }
 
 static struct file_system_type minix_fs_type = {
fs/msdos/namei.c +5 -4
index 626a367..7d71893 100644
@@ -674,11 +674,12 @@ static int msdos_fill_super(struct super
 	return 0;
 }
 
-static struct super_block *msdos_get_sb(struct file_system_type *fs_type,
-					int flags, const char *dev_name,
-					void *data)
+static int msdos_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, msdos_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, msdos_fill_super,
+			   mnt);
 }
 
 static struct file_system_type msdos_fs_type = {
fs/namespace.c +9 -0
index 70bba4b..8f96212 100644
@@ -86,6 +86,15 @@ struct vfsmount *alloc_vfsmnt(const char
 	return mnt;
 }
 
+int simple_set_mnt(struct vfsmount *mnt, struct super_block *sb)
+{
+	mnt->mnt_sb = sb;
+	mnt->mnt_root = dget(sb->s_root);
+	return 0;
+}
+
+EXPORT_SYMBOL(simple_set_mnt);
+
 void free_vfsmnt(struct vfsmount *mnt)
 {
 	kfree(mnt->mnt_devname);
fs/ncpfs/inode.c +3 -3
index d277a58..1115f0b 100644
@@ -956,10 +956,10 @@ out:
 	return result;
 }
 
-static struct super_block *ncp_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int ncp_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, ncp_fill_super);
+	return get_sb_nodev(fs_type, flags, data, ncp_fill_super, mnt);
 }
 
 static struct file_system_type ncp_fs_type = {
fs/nfs/inode.c +49 -33

191 lines not on this page · raw message

fs/nfsd/nfsctl.c +3 -3
index a0871b3..145861b 100644
@@ -494,10 +494,10 @@ static int nfsd_fill_super(struct super_
 	return simple_fill_super(sb, 0x6e667364, nfsd_files);
 }
 
-static struct super_block *nfsd_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int nfsd_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, nfsd_fill_super);
+	return get_sb_single(fs_type, flags, data, nfsd_fill_super, mnt);
 }
 
 static struct file_system_type nfsd_fs_type = {
fs/ntfs/super.c +4 -3
index 368a8ec..35d587a 100644
@@ -3081,10 +3081,11 @@ struct kmem_cache *ntfs_index_ctx_cache;
 /* Driver wide semaphore. */
 DECLARE_MUTEX(ntfs_lock);
 
-static struct super_block *ntfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int ntfs_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, ntfs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, ntfs_fill_super,
+			   mnt);
 }
 
 static struct file_system_type ntfs_fs_type = {
fs/ocfs2/dlm/dlmfs.c +3 -3
index dd2d24d..a5f709f 100644
@@ -574,10 +574,10 @@ static struct inode_operations dlmfs_fil
 	.getattr	= simple_getattr,
 };
 
-static struct super_block *dlmfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int dlmfs_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, dlmfs_fill_super);
+	return get_sb_nodev(fs_type, flags, data, dlmfs_fill_super, mnt);
 }
 
 static struct file_system_type dlmfs_fs_type = {
fs/ocfs2/super.c +7 -5
index 046824b..5949940 100644
@@ -672,12 +672,14 @@ read_super_error:
 	return status;
 }
 
-static struct super_block *ocfs2_get_sb(struct file_system_type *fs_type,
-					int flags,
-					const char *dev_name,
-					void *data)
+static int ocfs2_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, ocfs2_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, ocfs2_fill_super,
+			   mnt);
 }
 
 static struct file_system_type ocfs2_fs_type = {
fs/openpromfs/inode.c +3 -3
index aeb0106..05755f4 100644
@@ -1054,10 +1054,10 @@ out_no_root:
 	return -ENOMEM;
 }
 
-static struct super_block *openprom_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int openprom_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, openprom_fill_super);
+	return get_sb_single(fs_type, flags, data, openprom_fill_super, mnt);
 }
 
 static struct file_system_type openprom_fs_type = {
fs/pipe.c +3 -3
index d722579..fe1e4c8 100644
@@ -805,10 +805,10 @@ no_files:
  * d_name - pipe: will go nicely and kill the special-casing in procfs.
  */
 
-static struct super_block *pipefs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int pipefs_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, "pipe:", NULL, PIPEFS_MAGIC);
+	return get_sb_pseudo(fs_type, "pipe:", NULL, PIPEFS_MAGIC, mnt);
 }
 
 static struct file_system_type pipe_fs_type = {
fs/proc/root.c +3 -3
index c3fd361..9995356 100644
@@ -26,10 +26,10 @@ struct proc_dir_entry *proc_net, *proc_n
 struct proc_dir_entry *proc_sys_root;
 #endif
 
-static struct super_block *proc_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int proc_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, proc_fill_super);
+	return get_sb_single(fs_type, flags, data, proc_fill_super, mnt);
 }
 
 static struct file_system_type proc_fs_type = {
fs/qnx4/inode.c +4 -3
index 80f3291..29eebb3 100644
@@ -560,10 +560,11 @@ static void destroy_inodecache(void)
 		       "qnx4_inode_cache: not all structures were freed\n");
 }
 
-static struct super_block *qnx4_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int qnx4_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, qnx4_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, qnx4_fill_super,
+			   mnt);
 }
 
 static struct file_system_type qnx4_fs_type = {
fs/ramfs/inode.c +7 -6
index cde5d48..27f51cf 100644
@@ -184,16 +184,17 @@ static int ramfs_fill_super(struct super
 	return 0;
 }
 
-struct super_block *ramfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+int ramfs_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, ramfs_fill_super);
+	return get_sb_nodev(fs_type, flags, data, ramfs_fill_super, mnt);
 }
 
-static struct super_block *rootfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int rootfs_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|MS_NOUSER, data, ramfs_fill_super);
+	return get_sb_nodev(fs_type, flags|MS_NOUSER, data, ramfs_fill_super,
+			    mnt);
 }
 
 static struct file_system_type ramfs_fs_type = {
fs/reiserfs/super.c +5 -4
index d63da75..ce9f709 100644
@@ -2248,11 +2248,12 @@ static ssize_t reiserfs_quota_write(stru
 
 #endif
 
-static struct super_block *get_super_block(struct file_system_type *fs_type,
-					   int flags, const char *dev_name,
-					   void *data)
+static int get_super_block(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, reiserfs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, reiserfs_fill_super,
+			   mnt);
 }
 
 static int __init init_reiserfs_fs(void)
fs/relayfs/inode.c +4 -4
index 3835230..20dcb17 100644
@@ -537,11 +537,11 @@ static int relayfs_fill_super(struct sup
 	return 0;
 }
 
-static struct super_block * relayfs_get_sb(struct file_system_type *fs_type,
-					   int flags, const char *dev_name,
-					   void *data)
+static int relayfs_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, relayfs_fill_super);
+	return get_sb_single(fs_type, flags, data, relayfs_fill_super, mnt);
 }
 
 static struct file_system_type relayfs_fs_type = {
fs/romfs/inode.c +4 -3
index 0a13859..3a50126 100644
@@ -606,10 +606,11 @@ static struct super_operations romfs_ops
 	.remount_fs	= romfs_remount,
 };
 
-static struct super_block *romfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int romfs_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, romfs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, romfs_fill_super,
+			   mnt);
 }
 
 static struct file_system_type romfs_fs_type = {
fs/smbfs/inode.c +3 -3
index 02e3e82..2a89b69 100644
@@ -781,10 +781,10 @@ out:
 	return error;
 }
 
-static struct super_block *smb_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int smb_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, smb_fill_super);
+	return get_sb_nodev(fs_type, flags, data, smb_fill_super, mnt);
 }
 
 static struct file_system_type smb_fs_type = {
fs/super.c +50 -42

215 lines not on this page · raw message

fs/sysfs/mount.c +3 -3
index f1117e8..40190c4 100644
@@ -66,10 +66,10 @@ static int sysfs_fill_super(struct super
 	return 0;
 }
 
-static struct super_block *sysfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int sysfs_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, sysfs_fill_super);
+	return get_sb_single(fs_type, flags, data, sysfs_fill_super, mnt);
 }
 
 static struct file_system_type sysfs_fs_type = {
fs/sysv/super.c +7 -6
index 59e76b5..b81675d 100644
@@ -506,16 +506,17 @@ failed:
 
 /* Every kernel module contains stuff like this. */
 
-static struct super_block *sysv_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int sysv_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, sysv_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, sysv_fill_super,
+			   mnt);
 }
 
-static struct super_block *v7_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int v7_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, v7_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, v7_fill_super, mnt);
 }
 
 static struct file_system_type sysv_fs_type = {
fs/udf/super.c +3 -3
index 4a6f49a..4e86784 100644
@@ -94,10 +94,10 @@ static unsigned int udf_count_free(struc
 static int udf_statfs(struct super_block *, struct kstatfs *);
 
 /* UDF filesystem type */
-static struct super_block *udf_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int udf_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, udf_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, udf_fill_super, mnt);
 }
 
 static struct file_system_type udf_fstype = {
fs/ufs/super.c +3 -3
index e9055ef..8792cdf 100644
@@ -1310,10 +1310,10 @@ out:
 
 #endif
 
-static struct super_block *ufs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int ufs_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, ufs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, ufs_fill_super, mnt);
 }
 
 static struct file_system_type ufs_fs_type = {
fs/vfat/namei.c +5 -4
index ef46939..45601e7 100644
@@ -1041,11 +1041,12 @@ static int vfat_fill_super(struct super_
 	return 0;
 }
 
-static struct super_block *vfat_get_sb(struct file_system_type *fs_type,
-				       int flags, const char *dev_name,
-				       void *data)
+static int vfat_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, vfat_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, vfat_fill_super,
+			   mnt);
 }
 
 static struct file_system_type vfat_fs_type = {
fs/xfs/linux-2.6/xfs_super.c +5 -3
index f22e426..85405ab 100644
@@ -909,14 +909,16 @@ fail_vfsop:
 	return -error;
 }
 
-STATIC struct super_block *
+STATIC int
 linvfs_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, linvfs_fill_super);
+	return get_sb_bdev(fs_type, flags, dev_name, data, linvfs_fill_super,
+			   mnt);
 }
 
 STATIC struct super_operations linvfs_sops = {
include/linux/fs.h +15 -10
index 4652e42..074d776 100644
@@ -1236,23 +1236,26 @@ find_exported_dentry(struct super_block 
 struct file_system_type {
 	const char *name;
 	int fs_flags;
-	struct super_block *(*get_sb) (struct file_system_type *, int,
-				       const char *, void *);
+	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;
 	struct list_head fs_supers;
 };
 
-struct super_block *get_sb_bdev(struct file_system_type *fs_type,
+extern int get_sb_bdev(struct file_system_type *fs_type,
 	int flags, const char *dev_name, void *data,
-	int (*fill_super)(struct super_block *, void *, int));
-struct super_block *get_sb_single(struct file_system_type *fs_type,
+	int (*fill_super)(struct super_block *, void *, int),
+	struct vfsmount *mnt);
+extern int get_sb_single(struct file_system_type *fs_type,
 	int flags, void *data,
-	int (*fill_super)(struct super_block *, void *, int));
-struct super_block *get_sb_nodev(struct file_system_type *fs_type,
+	int (*fill_super)(struct super_block *, void *, int),
+	struct vfsmount *mnt);
+extern int get_sb_nodev(struct file_system_type *fs_type,
 	int flags, void *data,
-	int (*fill_super)(struct super_block *, void *, int));
+	int (*fill_super)(struct super_block *, void *, int),
+	struct vfsmount *mnt);
 void generic_shutdown_super(struct super_block *sb);
 void kill_block_super(struct super_block *sb);
 void kill_anon_super(struct super_block *sb);
@@ -1263,8 +1266,10 @@ struct super_block *sget(struct file_sys
 			int (*test)(struct super_block *,void *),
 			int (*set)(struct super_block *,void *),
 			void *data);
-struct super_block *get_sb_pseudo(struct file_system_type *, char *,
-			struct super_operations *ops, unsigned long);
+extern int get_sb_pseudo(struct file_system_type *, char *,
+	struct super_operations *ops, unsigned long,
+	struct vfsmount *mnt);
+extern int simple_set_mnt(struct vfsmount *mnt, struct super_block *sb);
 int __put_super(struct super_block *sb);
 int __put_super_and_need_restart(struct super_block *sb);
 void unnamed_dev_init(void);
include/linux/ramfs.h +2 -2
index 953b6df..8454a94 100644
@@ -2,8 +2,8 @@
 #define _LINUX_RAMFS_H
 
 struct inode *ramfs_get_inode(struct super_block *sb, int mode, dev_t dev);
-struct super_block *ramfs_get_sb(struct file_system_type *fs_type,
-	 int flags, const char *dev_name, void *data);
+extern int ramfs_get_sb(struct file_system_type *fs_type,
+	 int flags, const char *dev_name, void *data, struct vfsmount *mnt);
 
 #ifndef CONFIG_MMU
 extern unsigned long ramfs_nommu_get_unmapped_area(struct file *file,
ipc/mqueue.c +4 -4
index fd2e26b..7ef43ca 100644
@@ -201,11 +201,11 @@ static int mqueue_fill_super(struct supe
 	return 0;
 }
 
-static struct super_block *mqueue_get_sb(struct file_system_type *fs_type,
-					 int flags, const char *dev_name,
-					 void *data)
+static int mqueue_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, mqueue_fill_super);
+	return get_sb_single(fs_type, flags, data, mqueue_fill_super, mnt);
 }
 
 static void init_once(void *foo, kmem_cache_t * cachep, unsigned long flags)
kernel/cpuset.c +4 -4
index 12815d3..1bd9a96 100644
@@ -384,11 +384,11 @@ static int cpuset_fill_super(struct supe
 	return 0;
 }
 
-static struct super_block *cpuset_get_sb(struct file_system_type *fs_type,
-					int flags, const char *unused_dev_name,
-					void *data)
+static int cpuset_get_sb(struct file_system_type *fs_type,
+			 int flags, const char *unused_dev_name,
+			 void *data, struct vfsmount *mnt)
 {
-	return get_sb_single(fs_type, flags, data, cpuset_fill_super);
+	return get_sb_single(fs_type, flags, data, cpuset_fill_super, mnt);
 }
 
 static struct file_system_type cpuset_fs_type = {
kernel/futex.c +4 -4
index 5efa2f9..b8a2882 100644
@@ -884,11 +884,11 @@ asmlinkage long sys_futex(u32 __user *ua
 			(unsigned long)uaddr2, val2, val3);
 }
 
-static struct super_block *
-futexfs_get_sb(struct file_system_type *fs_type,
-	       int flags, const char *dev_name, void *data)
+static int futexfs_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, "futex", NULL, 0xBAD1DEA);
+	return get_sb_pseudo(fs_type, "futex", NULL, 0xBAD1DEA, mnt);
 }
 
 static struct file_system_type futex_fs_type = {
mm/shmem.c +3 -3
index 7c455fb..5ba9c1f 100644
@@ -2227,10 +2227,10 @@ static struct vm_operations_struct shmem
 };
 
 
-static struct super_block *shmem_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int shmem_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, shmem_fill_super);
+	return get_sb_nodev(fs_type, flags, data, shmem_fill_super, mnt);
 }
 
 static struct file_system_type tmpfs_fs_type = {
net/socket.c +4 -3
index a00851f..1d23df7 100644
@@ -327,10 +327,11 @@ static struct super_operations sockfs_op
 	.statfs =	simple_statfs,
 };
 
-static struct super_block *sockfs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data)
+static int sockfs_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, "socket:", &sockfs_ops, SOCKFS_MAGIC);
+	return get_sb_pseudo(fs_type, "socket:", &sockfs_ops, SOCKFS_MAGIC,
+			     mnt);
 }
 
 static struct vfsmount *sock_mnt __read_mostly;
net/sunrpc/rpc_pipe.c +3 -3
index 72b2217..1de7916 100644
@@ -813,11 +813,11 @@ out:
 	return -ENOMEM;
 }
 
-static struct super_block *
+static int
 rpc_get_sb(struct file_system_type *fs_type,
-		int flags, const char *dev_name, void *data)
+		int flags, const char *dev_name, void *data, struct vfsmount *mnt)
 {
-	return get_sb_single(fs_type, flags, data, rpc_fill_super);
+	return get_sb_single(fs_type, flags, data, rpc_fill_super, mnt);
 }
 
 static struct file_system_type rpc_pipe_fs_type = {
security/inode.c +4 -4
index 0f77b02..e6fc29a 100644
@@ -135,11 +135,11 @@ static int fill_super(struct super_block
 	return simple_fill_super(sb, SECURITYFS_MAGIC, files);
 }
 
-static struct super_block *get_sb(struct file_system_type *fs_type,
-				        int flags, const char *dev_name,
-					void *data)
+static int 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, fill_super);
+	return get_sb_single(fs_type, flags, data, fill_super, mnt);
 }
 
 static struct file_system_type fs_type = {
security/selinux/selinuxfs.c +4 -3
index b5fa02d..24eb26a 100644
@@ -1280,10 +1280,11 @@ out:
 	return -ENOMEM;
 }
 
-static struct super_block *sel_get_sb(struct file_system_type *fs_type,
-				      int flags, const char *dev_name, void *data)
+static int sel_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, sel_fill_super);
+	return get_sb_single(fs_type, flags, data, sel_fill_super, mnt);
 }
 
 static struct file_system_type sel_fs_type = {
author maintainer
#1 lore raw

PATCH 2/6 9p: Fix error handling on superblock alloc failure [try #7]

The attached patch adds error handling and cleanup in the case that sget() fails, lest a memory leak occur.

---

 fs/9p/vfs_super.c |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)
fs/9p/vfs_super.c +4 -1
index bca1efb..17dee27 100644
@@ -134,8 +134,11 @@ static int v9fs_get_sb(struct file_syste
 	}
 
 	sb = sget(fs_type, NULL, v9fs_set_super, v9ses);
-	if (IS_ERR(sb))
+	if (IS_ERR(sb)) {
+		v9fs_session_close(v9ses);
+		kfree(v9ses);
 		return PTR_ERR(sb);
+	}
 
 	v9fs_fill_super(sb, v9ses, flags);
 
#2 lore raw

Re: [PATCH 2/6] 9p: Fix error handling on superblock alloc failure [try #7]

Quoted 2 lines · David Howells

The attached patch adds error handling and cleanup in the case that sget() fails, lest a memory leak occur.

this is a small and simple fix that should go into 2.6.16 after the 9p maintainers ACKed it. Of course that means redoing it to apply before the get_sb prototype change.

author maintainer
#3 lore raw

PATCH 3/6 NFS: Abstract out namespace initialisation [try #7]

In reply to David Howells

The attached patch abstracts out the namespace initialisation so that temporary namespaces can be set up elsewhere.

---

 fs/namespace.c            |    8 +-------
 include/linux/namespace.h |   15 +++++++++++++++
 2 files changed, 16 insertions(+), 7 deletions(-)
fs/namespace.c +1 -7
index 8f96212..6f13481 100644
@@ -1726,13 +1726,7 @@ static void __init init_mount_tree(void)
 	namespace = kmalloc(sizeof(*namespace), GFP_KERNEL);
 	if (!namespace)
 		panic("Can't allocate initial namespace");
-	atomic_set(&namespace->count, 1);
-	INIT_LIST_HEAD(&namespace->list);
-	init_waitqueue_head(&namespace->poll);
-	namespace->event = 0;
-	list_add(&mnt->mnt_list, &namespace->list);
-	namespace->root = mnt;
-	mnt->mnt_namespace = namespace;
+	init_namespace(namespace, mnt);
 
 	init_task.namespace = namespace;
 	read_lock(&tasklist_lock);
include/linux/namespace.h +15 -0
index 3abc8e3..ea6fd62 100644
@@ -17,6 +17,21 @@ extern int copy_namespace(int, struct ta
 extern void __put_namespace(struct namespace *namespace);
 extern struct namespace *dup_namespace(struct task_struct *, struct fs_struct *);
 
+static inline void init_namespace(struct namespace *namespace,
+				  struct vfsmount *mnt)
+{
+	atomic_set(&namespace->count, 1);
+	INIT_LIST_HEAD(&namespace->list);
+	init_waitqueue_head(&namespace->poll);
+	namespace->event = 0;
+	namespace->root = mnt;
+
+	if (mnt) {
+		list_add(&mnt->mnt_list, &namespace->list);
+		mnt->mnt_namespace = namespace;
+	}
+}
+
 static inline void put_namespace(struct namespace *namespace)
 {
 	if (atomic_dec_and_lock(&namespace->count, &vfsmount_lock))
#4 lore raw

Re: [PATCH 3/6] NFS: Abstract out namespace initialisation [try #7]

Quoted 2 lines · David Howells

The attached patch abstracts out the namespace initialisation so that temporary namespaces can be set up elsewhere.

this _still_ should't be inline.

author maintainer
#5 lore raw

PATCH 1/1 NFS: Make namespace initialisation out of line

In reply to David Howells

The attached patch makes the abstracted-out namespace initialisation out of line.

---
warthog>diffstat -p1 /tmp/initns.diff

 fs/namespace.c            |   20 ++++++++++++++++++++
 include/linux/namespace.h |   16 +---------------
 2 files changed, 21 insertions(+), 15 deletions(-)
fs/namespace.c +20 -0
index 6f13481..3466a11 100644
@@ -1714,6 +1714,26 @@ out3:
 	goto out2;
 }
 
+/*
+ * initialise a namespace, rooting it at the given specified mountpoint if one
+ * is given
+ */
+void init_namespace(struct namespace *namespace, struct vfsmount *mnt)
+{
+	atomic_set(&namespace->count, 1);
+	INIT_LIST_HEAD(&namespace->list);
+	init_waitqueue_head(&namespace->poll);
+	namespace->event = 0;
+	namespace->root = mnt;
+
+	if (mnt) {
+		list_add(&mnt->mnt_list, &namespace->list);
+		mnt->mnt_namespace = namespace;
+	}
+}
+
+EXPORT_SYMBOL_GPL(init_namespace);
+
 static void __init init_mount_tree(void)
 {
 	struct vfsmount *mnt;
include/linux/namespace.h +1 -15
index ea6fd62..c371a30 100644
@@ -16,21 +16,7 @@ struct namespace {
 extern int copy_namespace(int, struct task_struct *);
 extern void __put_namespace(struct namespace *namespace);
 extern struct namespace *dup_namespace(struct task_struct *, struct fs_struct *);
-
-static inline void init_namespace(struct namespace *namespace,
-				  struct vfsmount *mnt)
-{
-	atomic_set(&namespace->count, 1);
-	INIT_LIST_HEAD(&namespace->list);
-	init_waitqueue_head(&namespace->poll);
-	namespace->event = 0;
-	namespace->root = mnt;
-
-	if (mnt) {
-		list_add(&mnt->mnt_list, &namespace->list);
-		mnt->mnt_namespace = namespace;
-	}
-}
+extern void init_namespace(struct namespace *, struct vfsmount *);
 
 static inline void put_namespace(struct namespace *namespace)
 {
author maintainer
#6 lore raw

PATCH 4/6 NFS: Add dentry materialisation op [try #7]

In reply to David Howells

The attached patch adds a new directory cache management function that prepares a disconnected anonymous function to be connected into the dentry tree. The anonymous dentry is transferred the name and parentage from another dentry.

The following changes were made in [try #2]:

(*) d_materialise_dentry() now switches the parentage of the two nodes around
correctly when one or other of them is self-referential.

The following changes were made in [try #7]:

(*) d_instantiate_unique() has had the interior part split out as function
__d_instantiate_unique(). Callers of this latter function must be holding the appropriate locks.

(*) _d_rehash() has been added as a wrapper around __d_rehash() to call it
with the most obvious hash list (the one from the name). d_rehash() now calls _d_rehash().

(*) d_materialise_dentry() is now __d_materialise_dentry() and is static.

(*) d_materialise_unique() added to perform the combination of d_find_alias(),
d_materialise_dentry() and d_add_unique() that the NFS client was doing twice, all within a single dcache_lock critical section. This reduces the number of times two different spinlocks were being accessed.

---

 fs/dcache.c            |  152 ++++++++++++++++++++++++++++++++++++++++++++----
 include/linux/dcache.h |    1 
 2 files changed, 139 insertions(+), 14 deletions(-)
fs/dcache.c +138 -14
index a173bba..9eb2698 100644
@@ -819,17 +819,19 @@ void d_instantiate(struct dentry *entry,
  * (or otherwise set) by the caller to indicate that it is now
  * in use by the dcache.
  */
-struct dentry *d_instantiate_unique(struct dentry *entry, struct inode *inode)
+static struct dentry *__d_instantiate_unique(struct dentry *entry,
+					     struct inode *inode)
 {
 	struct dentry *alias;
 	int len = entry->d_name.len;
 	const char *name = entry->d_name.name;
 	unsigned int hash = entry->d_name.hash;
 
-	BUG_ON(!list_empty(&entry->d_alias));
-	spin_lock(&dcache_lock);
-	if (!inode)
-		goto do_negative;
+	if (!inode) {
+		entry->d_inode = NULL;
+		return NULL;
+	}
+
 	list_for_each_entry(alias, &inode->i_dentry, d_alias) {
 		struct qstr *qstr = &alias->d_name;
 
@@ -842,18 +844,34 @@ struct dentry *d_instantiate_unique(stru
 		if (memcmp(qstr->name, name, len))
 			continue;
 		dget_locked(alias);
-		spin_unlock(&dcache_lock);
-		BUG_ON(!d_unhashed(alias));
-		iput(inode);
 		return alias;
 	}
+
 	list_add(&entry->d_alias, &inode->i_dentry);
-do_negative:
 	entry->d_inode = inode;
-	spin_unlock(&dcache_lock);
-	security_d_instantiate(entry, inode);
 	return NULL;
 }
+
+struct dentry *d_instantiate_unique(struct dentry *entry, struct inode *inode)
+{
+	struct dentry *result;
+
+	BUG_ON(!list_empty(&entry->d_alias));
+
+	spin_lock(&dcache_lock);
+	result = __d_instantiate_unique(entry, inode);
+	spin_unlock(&dcache_lock);
+
+	if (!result) {
+		security_d_instantiate(entry, inode);
+		return NULL;
+	}
+
+	BUG_ON(!d_unhashed(result));
+	iput(inode);
+	return result;
+}
+
 EXPORT_SYMBOL(d_instantiate_unique);
 
 /**
@@ -1194,6 +1212,11 @@ static void __d_rehash(struct dentry * e
  	hlist_add_head_rcu(&entry->d_hash, list);
 }
 
+static void _d_rehash(struct dentry * entry)
+{
+	__d_rehash(entry, d_hash(entry->d_parent, entry->d_name.hash));
+}
+
 /**
  * d_rehash	- add an entry back to the hash
  * @entry: dentry to add to the hash
@@ -1203,11 +1226,9 @@ static void __d_rehash(struct dentry * e
  
 void d_rehash(struct dentry * entry)
 {
-	struct hlist_head *list = d_hash(entry->d_parent, entry->d_name.hash);
-
 	spin_lock(&dcache_lock);
 	spin_lock(&entry->d_lock);
-	__d_rehash(entry, list);
+	_d_rehash(entry);
 	spin_unlock(&entry->d_lock);
 	spin_unlock(&dcache_lock);
 }
@@ -1344,6 +1365,108 @@ already_unhashed:
 	spin_unlock(&dcache_lock);
 }
 
+/*
+ * Prepare an anonymous dentry for life in the superblock's dentry tree as a
+ * named dentry in place of the dentry to be replaced.
+ */
+static void __d_materialise_dentry(struct dentry *dentry, struct dentry *anon)
+{
+	struct dentry *dparent, *aparent;
+
+	switch_names(dentry, anon);
+	do_switch(dentry->d_name.len, anon->d_name.len);
+	do_switch(dentry->d_name.hash, anon->d_name.hash);
+
+	dparent = dentry->d_parent;
+	aparent = anon->d_parent;
+	dentry->d_parent = (aparent == anon) ? dentry : aparent;
+	anon->d_parent = (dparent == dentry) ? anon : dparent;
+
+	anon->d_flags &= ~DCACHE_DISCONNECTED;
+}
+
+/**
+ * d_materialise_unique - introduce an inode into the tree
+ * @dentry: candidate dentry
+ * @inode: inode to bind to the dentry, to which aliases may be attached
+ *
+ * Introduces an dentry into the tree, substituting an extant disconnected
+ * root directory alias in its place if there is one
+ */
+struct dentry *d_materialise_unique(struct dentry *dentry, struct inode *inode)
+{
+	struct dentry *alias, *actual;
+
+	BUG_ON(!d_unhashed(dentry));
+
+	spin_lock(&dcache_lock);
+
+	if (!inode) {
+		actual = dentry;
+		dentry->d_inode = NULL;
+		goto found_lock;
+	}
+
+	/* See if a disconnected directory already exists as an anonymous root
+	 * that we should splice into the tree instead */
+	if (S_ISDIR(inode->i_mode) && (alias = __d_find_alias(inode, 1))) {
+		spin_lock(&alias->d_lock);
+
+		/* Is this a mountpoint that we could splice into our tree? */
+		if (IS_ROOT(alias))
+			goto connect_mountpoint;
+
+		if (alias->d_name.len == dentry->d_name.len &&
+		    alias->d_parent == dentry->d_parent &&
+		    memcmp(alias->d_name.name,
+			   dentry->d_name.name,
+			   dentry->d_name.len) == 0)
+			goto replace_with_alias;
+
+		spin_unlock(&alias->d_lock);
+
+		/* Doh! Seem to be aliasing directories for some reason... */
+		dput(alias);
+	}
+
+	/* Add a unique reference */
+	actual = __d_instantiate_unique(dentry, inode);
+	if (!actual)
+		actual = dentry;
+	else if (unlikely(!d_unhashed(actual)))
+		goto shouldnt_be_hashed;
+
+found_lock:
+	spin_lock(&actual->d_lock);
+found:
+	_d_rehash(actual);
+	spin_unlock(&actual->d_lock);
+	spin_unlock(&dcache_lock);
+
+	if (actual == dentry) {
+		security_d_instantiate(dentry, inode);
+		return NULL;
+	}
+
+	iput(inode);
+	return actual;
+
+	/* Convert the anonymous/root alias into an ordinary dentry */
+connect_mountpoint:
+	__d_materialise_dentry(dentry, alias);
+
+	/* Replace the candidate dentry with the alias in the tree */
+replace_with_alias:
+	__d_drop(alias);
+	actual = alias;
+	goto found;
+
+shouldnt_be_hashed:
+	spin_unlock(&dcache_lock);
+	BUG();
+	goto shouldnt_be_hashed;
+}
+
 /**
  * d_path - return the path of a dentry
  * @dentry: dentry to report
@@ -1755,6 +1878,7 @@ EXPORT_SYMBOL(d_instantiate);
 EXPORT_SYMBOL(d_invalidate);
 EXPORT_SYMBOL(d_lookup);
 EXPORT_SYMBOL(d_move);
+EXPORT_SYMBOL_GPL(d_materialise_unique);
 EXPORT_SYMBOL(d_path);
 EXPORT_SYMBOL(d_prune_aliases);
 EXPORT_SYMBOL(d_rehash);
include/linux/dcache.h +1 -0
index 4361f37..379e7f2 100644
@@ -207,6 +207,7 @@ static inline int dname_external(struct 
  */
 extern void d_instantiate(struct dentry *, struct inode *);
 extern struct dentry * d_instantiate_unique(struct dentry *, struct inode *);
+extern struct dentry * d_materialise_unique(struct dentry *, struct inode *);
 extern void d_delete(struct dentry *);
 
 /* allocate/de-allocate */
#7 lore raw

Re: [PATCH 4/6] NFS: Add dentry materialisation op [try #7]

Quoted 3 lines

(*) _d_rehash() has been added as a wrapper around __d_rehash() to call it
with the most obvious hash list (the one from the name). d_rehash() now calls _d_rehash().

We're not really kean on single _ prefix functions. Could you give it a saner name instead (or a new one for __d_rehash and rename _d_reshash to that)

author maintainer
#8 lore raw

PATCH 5/6 NFS: Unify NFS superblocks per-protocol per-server [try #7]

In reply to David Howells

The attached patch makes NFS share superblocks between mounts from the same server over the same protocol.

It does this by creating each superblock with a false root and returning the real root dentry through the new argument added to get_sb() in patch 1/5. The root dentry returned starts off as an anonymous dentry if we don't already have the dentry for its inode, otherwise it simply returns the dentry we already have.

We may thus end up with several trees of dentries in the superblock, and if at some later point one of anonymous tree roots is discovered by normal filesystem activity to be located in another tree within the superblock, the anonymous root is named and materialises attached to the second tree at the appropriate point.

Why modify get_sb() in this way? Why not pass an extra argument to the mount() syscall to indicate the subpath and then pathwalk from the server root to the desired directory? You can't guarantee this will work for two reasons:

(1) The root and intervening nodes may not be accessible to the client.

With NFS2 and NFS3, for instance, mountd is called on the server to get the filehandle for the tip of a path. mountd won't give us handles for anything we don't have permission to access, and so we can't set up NFS inodes for such nodes, and so can't easily set up dentries (we'd have to have ghost inodes or something).

With this patch we don't actually create dentries until we get handles from the server that we can use to set up their inodes, and we don't actually bind them into the tree until we know for sure where they go.

(2) Inaccessible symbolic links.

If we're asked to mount two exports from the server, eg:

mount warthog:/warthog/aaa/xxx /mmm
mount warthog:/warthog/bbb/yyy /nnn

We may not be able to access anything nearer the root than xxx and yyy, but we may find out later that /mmm/www/yyy, say, is actually the same directory as the one mounted on /nnn. What we might then find out, for example, is that /warthog/bbb was actually a symbolic link to /warthog/aaa/xxx/www, but we can't actually determine that by talking to the server until /warthog is made available by NFS.

This would lead to having constructed an errneous dentry tree which we can't easily fix. We can end up with a dentry marked as a directory when it should actually be a symlink, or we could end up with an apparently hardlinked directory.

With this patch we need not make assumptions about the type of a dentry for which we can't retrieve information, nor need we assume we know its place in the grand scheme of things until we actually see that place.

This patch reduces the possibility of aliasing in the inode and page caches for inodes that may be accessed by more than one NFS export. It also reduces the number of superblocks required for NFS where there are many NFS exports being used from a server (home directory server + autofs for example).

This in turn makes it simpler to do local caching of network filesystems, as it can then be guaranteed that there won't be links from multiple inodes in separate superblocks to the same cache file.

Obviously, cache aliasing between different levels of NFS protocol is still a problem, but at least that gives us another key to use when indexing the cache.

This patch also cleans up the error handling to make sure the NFS iostats are released correctly upon error. The server construction and destruction have been abstracted into functions as the first is done twice and the second four times.

And finally, the patch exports some functions required from the core kernel.

The following changes were made in [try #2]:

(*) Whilst invoking link_path_walk(), the nfs4_get_root() routine now
temporarily overrides the FS settings of the process to prevent absolute symbolic links from walking into the wrong namespace.

(*) nfs*_get_sb() instantiate the supplied vfsmount directly by assigning to
its mnt_root and mnt_sb members.

(*) nfs_readdir_lookup() creates dentries for each of the entries readdir()
returns; this function now attaches disconnected trees from alternate roots that happen to be discovered attached to a directory being read (in the same way nfs_lookup() is made to do for lookup ops).

Further changes [try #3] that have been made:

(*) The patches are now against Trond's NFS git tree, so won't apply to
Linus's tree.

(*) The server record construction/destruction has been abstracted out into
its own pair of functions to make things easier to get right.

Further changes [try #7] that have been made:

(*) The new d_materialise_unique() function is now used in place of the
combination of d_find_alias(), d_materialise_dentry() and d_add_unique() in nfs_lookup() and nfs_readdir_lookup().

This permits the combination to be done under one set of locks, thus avoiding any race between mount and lookup operations on the same directory.

---

 fs/namei.c                |    2 
 fs/namespace.c            |    4 +
 fs/nfs/Makefile           |    4 -
 fs/nfs/dir.c              |    7 +
 fs/nfs/getroot.c          |  239 +++++++++++++++++++++++++++++++++++++++++
 fs/nfs/inode.c            |  265 ++++++++++++++++++++++++---------------------
 fs/nfs/internal.h         |   30 +++++
 fs/nfs/nfs3proc.c         |    2 
 fs/nfs/nfs4proc.c         |   59 +---------
 fs/nfs/nfs4state.c        |    2 
 include/linux/nfs_fs_sb.h |    2 
 11 files changed, 428 insertions(+), 188 deletions(-)
fs/namei.c +2 -0
index 557dcf3..81bc997 100644
@@ -992,6 +992,8 @@ int fastcall link_path_walk(const char *
 	return result;
 }
 
+EXPORT_SYMBOL_GPL(link_path_walk);
+
 int fastcall path_walk(const char * name, struct nameidata *nd)
 {
 	current->total_link_count = 0;
fs/namespace.c +4 -0
index 6f13481..ee3b146 100644
@@ -86,6 +86,8 @@ struct vfsmount *alloc_vfsmnt(const char
 	return mnt;
 }
 
+EXPORT_SYMBOL_GPL(alloc_vfsmnt);
+
 int simple_set_mnt(struct vfsmount *mnt, struct super_block *sb)
 {
 	mnt->mnt_sb = sb;
@@ -101,6 +103,8 @@ void free_vfsmnt(struct vfsmount *mnt)
 	kmem_cache_free(mnt_cache, mnt);
 }
 
+EXPORT_SYMBOL_GPL(free_vfsmnt);
+
 /*
  * find the first or last mount at @dentry on vfsmount @mnt depending on
  * @dir. If @dir is set return the first mount else return the last mount.
fs/nfs/Makefile +2 -2
index ec61fd5..43ce192 100644
@@ -4,8 +4,8 @@
 
 obj-$(CONFIG_NFS_FS) += nfs.o
 
-nfs-y 			:= dir.o file.o inode.o nfs2xdr.o pagelist.o \
-			   proc.o read.o symlink.o unlink.o write.o
+nfs-y 			:= dir.o file.o getroot.o inode.o nfs2xdr.o \
+			   pagelist.o proc.o read.o symlink.o unlink.o write.o
 nfs-$(CONFIG_ROOT_NFS)	+= nfsroot.o mount_clnt.o      
 nfs-$(CONFIG_NFS_V3)	+= nfs3proc.o nfs3xdr.o
 nfs-$(CONFIG_NFS_V3_ACL)	+= nfs3acl.o
fs/nfs/dir.c +5 -2
index 609185a..639896d 100644
@@ -905,8 +905,9 @@ static struct dentry *nfs_lookup(struct 
 	inode = nfs_fhget(dentry->d_sb, &fhandle, &fattr);
 	if (!inode)
 		goto out_unlock;
+
 no_entry:
-	res = d_add_unique(dentry, inode);
+	res = d_materialise_unique(dentry, inode);
 	if (res != NULL)
 		dentry = res;
 	nfs_renew_times(dentry);
@@ -1100,11 +1101,13 @@ static struct dentry *nfs_readdir_lookup
 		dput(dentry);
 		return NULL;
 	}
-	alias = d_add_unique(dentry, inode);
+
+	alias = d_materialise_unique(dentry, inode);
 	if (alias != NULL) {
 		dput(dentry);
 		dentry = alias;
 	}
+
 	nfs_renew_times(dentry);
 	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
 	return dentry;
fs/nfs/getroot.c new file +239 -0

239 lines not on this page · raw message

fs/nfs/inode.c +141 -124

443 lines not on this page · raw message

fs/nfs/internal.h new file +30 -0
new file mode 100644
index 0000000..25f554f
@@ -0,0 +1,30 @@
+/* internal.h: internal NFS definitions
+ *
+ * Copyright (C) 2005 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells ([email protected])
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+
+/*
+ * getroot.c
+ */
+typedef void (*nfs_set_params_func)(struct super_block *sb,
+				    struct nfs_fh *mntfh,
+				    struct nfs_fsinfo *fsinfo);
+
+extern struct dentry *nfs_get_root(struct super_block *sb,
+				   struct nfs_fh *mntfh,
+				   nfs_set_params_func set_params);
+
+#ifdef CONFIG_NFS_V4
+
+extern struct dentry *nfs4_get_root(struct super_block *sb,
+				    const char *mntpath,
+				    nfs_set_params_func set_params);
+
+#endif
fs/nfs/nfs3proc.c +1 -1
index cf186f0..fdcee05 100644
@@ -82,7 +82,7 @@ do_proc_get_root(struct rpc_clnt *client
 }
 
 /*
- * Bare-bones access to getattr: this is for nfs_read_super.
+ * Bare-bones access to getattr: this is for nfs_get_root/nfs_get_sb
  */
 static int
 nfs3_proc_get_root(struct nfs_server *server, struct nfs_fh *fhandle,
fs/nfs/nfs4proc.c +4 -55

74 lines not on this page · raw message

fs/nfs/nfs4state.c +0 -2
index afad025..dbe3adf 100644
@@ -69,8 +69,6 @@ init_nfsv4_state(struct nfs_server *serv
 void
 destroy_nfsv4_state(struct nfs_server *server)
 {
-	kfree(server->mnt_path);
-	server->mnt_path = NULL;
 	if (server->nfs4_state) {
 		nfs4_put_client(server->nfs4_state);
 		server->nfs4_state = NULL;
include/linux/nfs_fs_sb.h +0 -2
index 65dec21..691a6cc 100644
@@ -33,7 +33,6 @@ struct nfs_server {
 	unsigned int		retrans_count;	/* number of retransmit tries */
 	unsigned int		namelen;
 	char *			hostname;	/* remote hostname */
-	struct nfs_fh		fh;
 	struct sockaddr_in	addr;
 	unsigned long		mount_time;	/* when this fs was mounted */
 #ifdef CONFIG_NFS_V4
@@ -41,7 +40,6 @@ struct nfs_server {
 	 * This is used to generate the clientid, and the callback address.
 	 */
 	char			ip_addr[16];
-	char *			mnt_path;
 	struct nfs4_client *	nfs4_state;	/* all NFSv4 state starts here */
 	struct list_head	nfs4_siblings;	/* List of other nfs_server structs
 						 * that share the same clientid
#9 lore raw

Re: [PATCH 5/6] NFS: Unify NFS superblocks per-protocol per-server [try #7]

Everything in nfs4_get_root past the "create a mount to represent the NFS.." comment in nfs4_get_root is generic and not nfs-specific. this should move into a helper in namei.c instead of adding all these odd exports.

what it's doing looks a little fishy aswell, but I'll come back to that later.

#10 lore raw

Re: [PATCH 5/6] NFS: Unify NFS superblocks per-protocol per-server [try #7]

In reply to David Howells
Quoted 2 lines · David Howells

The attached patch makes NFS share superblocks between mounts from the same server over the same protocol.

We want to make NFS share superblocks on a per-filesystem basis, rather than forcing it into a per-server basis. Cachefs may like the latter, but POSIX does not like a filesystem where inode numbers are not guaranteed to be unique.
A unique per-server superblock also makes it hard to support features like failover onto replicated filesystems and/or migration of individual filesystems onto another server.

Cheers,
Trond

author maintainer
#11 lore raw

Re: [PATCH 5/6] NFS: Unify NFS superblocks per-protocol per-server [try #7]

Quoted 5 lines · Trond Myklebust
Quoted 2 lines

The attached patch makes NFS share superblocks between mounts from the same server over the same protocol.

We want to make NFS share superblocks on a per-filesystem basis, rather than forcing it into a per-server basis.

By "per-filesystem", I presume you mean per-server-mounted-filesystem? Something like what the kAFS client tries to do. As it happens the current NFS code doesn't do this either, but it has occurred to me that this would be useful, just not necessarily easy...

Quoted 2 lines

Cachefs may like the latter, but POSIX does not like a filesystem where inode numbers are not guaranteed to be unique.

I don't like it either, but the situation is already present.

Note that CacheFS would be happy with the former too, just as long as the keys it is given don't end up aliasing.

Quoted 3 lines

A unique per-server superblock also makes it hard to support features like failover onto replicated filesystems and/or migration of individual filesystems onto another server.

So you want these patches dropping?

Note that you can't necessarily do what you want with the current code either.

David

author maintainer
#12 lore raw

PATCH 6/6 Optimise d_find_alias() [try #7]

In reply to David Howells

The attached patch optimises d_find_alias() to only take the spinlock if there's anything in the the inode's alias list. If there isn't, it returns NULL immediately.

The following changes were made in [try #7]:

(*) The memory barrier was removed.

---

 fs/dcache.c |   10 ++++++----
 1 files changed, 6 insertions(+), 4 deletions(-)
fs/dcache.c +6 -4
index 9eb2698..bb309f5 100644
@@ -325,10 +325,12 @@ static struct dentry * __d_find_alias(st
 
 struct dentry * d_find_alias(struct inode *inode)
 {
-	struct dentry *de;
-	spin_lock(&dcache_lock);
-	de = __d_find_alias(inode, 0);
-	spin_unlock(&dcache_lock);
+	struct dentry *de = NULL;
+	if (!list_empty(&inode->i_dentry)) {
+		spin_lock(&dcache_lock);
+		de = __d_find_alias(inode, 0);
+		spin_unlock(&dcache_lock);
+	}
 	return de;
 }