afs: Support the AFS dynamic root

Support the AFS dynamic root which is a pseudo-volume that doesn't connect
to any server resource, but rather is just a root directory that
dynamically creates mountpoint directories where the name of such a
directory is the name of the cell.

Such a mount can be created thus:

	mount -t afs none /afs -o dyn

Dynamic root superblocks aren't shared except by bind mounts and
propagation.  Cell root volumes can then be mounted by referring to them by
name, e.g.:

	ls /afs/grand.central.org/
	ls /afs/.grand.central.org/

The kernel will upcall to consult the DNS if the address wasn't supplied
directly.

Signed-off-by: David Howells <[email protected]>

Documentation/filesystems/afs.txt

@@ -7,6 +7,7 @@ Contents:
  - Overview.
  - Usage.
  - Mountpoints.
+ - Dynamic root.
  - Proc filesystem.
  - The cell database.
  - Security.
@@ -127,6 +128,22 @@ mounted on /afs in one go by doing:
 	umount /afs
 
 
+============
+DYNAMIC ROOT
+============
+
+A mount option is available to create a serverless mount that is only usable
+for dynamic lookup.  Creating such a mount can be done by, for example:
+
+	mount -t afs none /afs -o dyn
+
+This creates a mount that just has an empty directory at the root.  Attempting
+to look up a name in this directory will cause a mountpoint to be created that
+looks up a cell of the same name, for example:
+
+	ls /afs/grand.central.org/
+
+
 ===============
 PROC FILESYSTEM
 ===============

fs/afs/dir.c

@@ -17,10 +17,13 @@
 #include <linux/pagemap.h>
 #include <linux/ctype.h>
 #include <linux/sched.h>
+#include <linux/dns_resolver.h>
 #include "internal.h"
 
 static struct dentry *afs_lookup(struct inode *dir, struct dentry *dentry,
 				 unsigned int flags);
+static struct dentry *afs_dynroot_lookup(struct inode *dir, struct dentry *dentry,
+					 unsigned int flags);
 static int afs_dir_open(struct inode *inode, struct file *file);
 static int afs_readdir(struct file *file, struct dir_context *ctx);
 static int afs_d_revalidate(struct dentry *dentry, unsigned int flags);
@@ -64,6 +67,17 @@ const struct inode_operations afs_dir_inode_operations = {
 	.listxattr	= afs_listxattr,
 };
 
+const struct file_operations afs_dynroot_file_operations = {
+	.open		= dcache_dir_open,
+	.release	= dcache_dir_close,
+	.iterate_shared	= dcache_readdir,
+	.llseek		= dcache_dir_lseek,
+};
+
+const struct inode_operations afs_dynroot_inode_operations = {
+	.lookup		= afs_dynroot_lookup,
+};
+
 const struct dentry_operations afs_fs_dentry_operations = {
 	.d_revalidate	= afs_d_revalidate,
 	.d_delete	= afs_d_delete,
@@ -467,26 +481,59 @@ static int afs_do_lookup(struct inode *dir, struct dentry *dentry,
 	return 0;
 }
 
+/*
+ * Probe to see if a cell may exist.  This prevents positive dentries from
+ * being created unnecessarily.
+ */
+static int afs_probe_cell_name(struct dentry *dentry)
+{
+	struct afs_cell *cell;
+	const char *name = dentry->d_name.name;
+	size_t len = dentry->d_name.len;
+	int ret;
+
+	/* Names prefixed with a dot are R/W mounts. */
+	if (name[0] == '.') {
+		if (len == 1)
+			return -EINVAL;
+		name++;
+		len--;
+	}
+
+	cell = afs_lookup_cell_rcu(afs_d2net(dentry), name, len);
+	if (!IS_ERR(cell)) {
+		afs_put_cell(afs_d2net(dentry), cell);
+		return 0;
+	}
+
+	ret = dns_query("afsdb", name, len, "ipv4", NULL, NULL);
+	if (ret == -ENODATA)
+		ret = -EDESTADDRREQ;
+	return ret;
+}
+
 /*
  * Try to auto mount the mountpoint with pseudo directory, if the autocell
  * operation is setted.
  */
-static struct inode *afs_try_auto_mntpt(
-	int ret, struct dentry *dentry, struct inode *dir, struct key *key,
-	struct afs_fid *fid)
+static struct inode *afs_try_auto_mntpt(struct dentry *dentry,
+					struct inode *dir, struct afs_fid *fid)
 {
-	const char *devname = dentry->d_name.name;
 	struct afs_vnode *vnode = AFS_FS_I(dir);
 	struct inode *inode;
+	int ret = -ENOENT;
 
-	_enter("%d, %p{%pd}, {%x:%u}, %p",
-	       ret, dentry, dentry, vnode->fid.vid, vnode->fid.vnode, key);
+	_enter("%p{%pd}, {%x:%u}",
+	       dentry, dentry, vnode->fid.vid, vnode->fid.vnode);
+
+	if (!test_bit(AFS_VNODE_AUTOCELL, &vnode->flags))
+		goto out;
 
-	if (ret != -ENOENT ||
-	    !test_bit(AFS_VNODE_AUTOCELL, &vnode->flags))
+	ret = afs_probe_cell_name(dentry);
+	if (ret < 0)
 		goto out;
 
-	inode = afs_iget_autocell(dir, devname, strlen(devname), key);
+	inode = afs_iget_pseudo_dir(dir->i_sb, false);
 	if (IS_ERR(inode)) {
 		ret = PTR_ERR(inode);
 		goto out;
@@ -545,13 +592,16 @@ static struct dentry *afs_lookup(struct inode *dir, struct dentry *dentry,
 
 	ret = afs_do_lookup(dir, dentry, &fid, key);
 	if (ret < 0) {
-		inode = afs_try_auto_mntpt(ret, dentry, dir, key, &fid);
-		if (!IS_ERR(inode)) {
-			key_put(key);
-			goto success;
+		if (ret == -ENOENT) {
+			inode = afs_try_auto_mntpt(dentry, dir, &fid);
+			if (!IS_ERR(inode)) {
+				key_put(key);
+				goto success;
+			}
+
+			ret = PTR_ERR(inode);
 		}
 
-		ret = PTR_ERR(inode);
 		key_put(key);
 		if (ret == -ENOENT) {
 			d_add(dentry, NULL);
@@ -582,13 +632,54 @@ static struct dentry *afs_lookup(struct inode *dir, struct dentry *dentry,
 	return NULL;
 }
 
+/*
+ * Look up an entry in a dynroot directory.
+ */
+static struct dentry *afs_dynroot_lookup(struct inode *dir, struct dentry *dentry,
+					 unsigned int flags)
+{
+	struct afs_vnode *vnode;
+	struct afs_fid fid;
+	struct inode *inode;
+	int ret;
+
+	vnode = AFS_FS_I(dir);
+
+	_enter("%pd", dentry);
+
+	ASSERTCMP(d_inode(dentry), ==, NULL);
+
+	if (dentry->d_name.len >= AFSNAMEMAX) {
+		_leave(" = -ENAMETOOLONG");
+		return ERR_PTR(-ENAMETOOLONG);
+	}
+
+	inode = afs_try_auto_mntpt(dentry, dir, &fid);
+	if (IS_ERR(inode)) {
+		ret = PTR_ERR(inode);
+		if (ret == -ENOENT) {
+			d_add(dentry, NULL);
+			_leave(" = NULL [negative]");
+			return NULL;
+		}
+		_leave(" = %d [do]", ret);
+		return ERR_PTR(ret);
+	}
+
+	d_add(dentry, inode);
+	_leave(" = 0 { ino=%lu v=%u }",
+	       d_inode(dentry)->i_ino, d_inode(dentry)->i_generation);
+	return NULL;
+}
+
 /*
  * check that a dentry lookup hit has found a valid entry
  * - NOTE! the hit can be a negative hit too, so we can't assume we have an
  *   inode
  */
 static int afs_d_revalidate(struct dentry *dentry, unsigned int flags)
 {
+	struct afs_super_info *as = dentry->d_sb->s_fs_info;
 	struct afs_vnode *vnode, *dir;
 	struct afs_fid uninitialized_var(fid);
 	struct dentry *parent;
@@ -600,6 +691,9 @@ static int afs_d_revalidate(struct dentry *dentry, unsigned int flags)
 	if (flags & LOOKUP_RCU)
 		return -ECHILD;
 
+	if (as->dyn_root)
+		return 1;
+
 	if (d_really_is_positive(dentry)) {
 		vnode = AFS_FS_I(d_inode(dentry));
 		_enter("{v={%x:%u} n=%pd fl=%lx},",

fs/afs/inode.c

@@ -147,7 +147,7 @@ int afs_iget5_test(struct inode *inode, void *opaque)
  *
  * These pseudo inodes don't match anything.
  */
-static int afs_iget5_autocell_test(struct inode *inode, void *opaque)
+static int afs_iget5_pseudo_dir_test(struct inode *inode, void *opaque)
 {
 	return 0;
 }
@@ -169,31 +169,34 @@ static int afs_iget5_set(struct inode *inode, void *opaque)
 }
 
 /*
- * inode retrieval for autocell
+ * Create an inode for a dynamic root directory or an autocell dynamic
+ * automount dir.
  */
-struct inode *afs_iget_autocell(struct inode *dir, const char *dev_name,
-				int namesz, struct key *key)
+struct inode *afs_iget_pseudo_dir(struct super_block *sb, bool root)
 {
 	struct afs_iget_data data;
 	struct afs_super_info *as;
 	struct afs_vnode *vnode;
-	struct super_block *sb;
 	struct inode *inode;
 	static atomic_t afs_autocell_ino;
 
-	_enter("{%x:%u},%*.*s,",
-	       AFS_FS_I(dir)->fid.vid, AFS_FS_I(dir)->fid.vnode,
-	       namesz, namesz, dev_name ?: "");
+	_enter("");
 
-	sb = dir->i_sb;
 	as = sb->s_fs_info;
-	data.volume = as->volume;
-	data.fid.vid = as->volume->vid;
-	data.fid.unique = 0;
-	data.fid.vnode = 0;
+	if (as->volume) {
+		data.volume = as->volume;
+		data.fid.vid = as->volume->vid;
+	}
+	if (root) {
+		data.fid.vnode = 1;
+		data.fid.unique = 1;
+	} else {
+		data.fid.vnode = atomic_inc_return(&afs_autocell_ino);
+		data.fid.unique = 0;
+	}
 
-	inode = iget5_locked(sb, atomic_inc_return(&afs_autocell_ino),
-			     afs_iget5_autocell_test, afs_iget5_set,
+	inode = iget5_locked(sb, data.fid.vnode,
+			     afs_iget5_pseudo_dir_test, afs_iget5_set,
 			     &data);
 	if (!inode) {
 		_leave(" = -ENOMEM");
@@ -211,7 +214,12 @@ struct inode *afs_iget_autocell(struct inode *dir, const char *dev_name,
 
 	inode->i_size		= 0;
 	inode->i_mode		= S_IFDIR | S_IRUGO | S_IXUGO;
-	inode->i_op		= &afs_autocell_inode_operations;
+	if (root) {
+		inode->i_op	= &afs_dynroot_inode_operations;
+		inode->i_fop	= &afs_dynroot_file_operations;
+	} else {
+		inode->i_op	= &afs_autocell_inode_operations;
+	}
 	set_nlink(inode, 2);
 	inode->i_uid		= GLOBAL_ROOT_UID;
 	inode->i_gid		= GLOBAL_ROOT_GID;
@@ -223,8 +231,12 @@ struct inode *afs_iget_autocell(struct inode *dir, const char *dev_name,
 	inode->i_generation	= 0;
 
 	set_bit(AFS_VNODE_PSEUDODIR, &vnode->flags);
-	set_bit(AFS_VNODE_MOUNTPOINT, &vnode->flags);
-	inode->i_flags |= S_AUTOMOUNT | S_NOATIME;
+	if (!root) {
+		set_bit(AFS_VNODE_MOUNTPOINT, &vnode->flags);
+		inode->i_flags |= S_AUTOMOUNT;
+	}
+
+	inode->i_flags |= S_NOATIME;
 	unlock_new_inode(inode);
 	_leave(" = %p", inode);
 	return inode;

fs/afs/internal.h

@@ -36,6 +36,7 @@ struct afs_mount_params {
 	bool			rwpath;		/* T if the parent should be considered R/W */
 	bool			force;		/* T to force cell type */
 	bool			autocell;	/* T if set auto mount operation */
+	bool			dyn_root;	/* T if dynamic root */
 	afs_voltype_t		type;		/* type of volume requested */
 	int			volnamesz;	/* size of volume name */
 	const char		*volname;	/* name of volume to mount */
@@ -186,6 +187,7 @@ struct afs_super_info {
 	struct afs_net		*net;		/* Network namespace */
 	struct afs_cell		*cell;		/* The cell in which the volume resides */
 	struct afs_volume	*volume;	/* volume record */
+	bool			dyn_root;	/* True if dynamic root */
 };
 
 static inline struct afs_super_info *AFS_FS_S(struct super_block *sb)
@@ -634,10 +636,13 @@ extern bool afs_cm_incoming_call(struct afs_call *);
 /*
  * dir.c
  */
-extern bool afs_dir_check_page(struct inode *, struct page *);
+extern const struct file_operations afs_dir_file_operations;
 extern const struct inode_operations afs_dir_inode_operations;
+extern const struct file_operations afs_dynroot_file_operations;
+extern const struct inode_operations afs_dynroot_inode_operations;
 extern const struct dentry_operations afs_fs_dentry_operations;
-extern const struct file_operations afs_dir_file_operations;
+
+extern bool afs_dir_check_page(struct inode *, struct page *);
 
 /*
  * file.c
@@ -695,8 +700,7 @@ extern int afs_fs_get_capabilities(struct afs_net *, struct afs_server *,
  */
 extern int afs_fetch_status(struct afs_vnode *, struct key *);
 extern int afs_iget5_test(struct inode *, void *);
-extern struct inode *afs_iget_autocell(struct inode *, const char *, int,
-				       struct key *);
+extern struct inode *afs_iget_pseudo_dir(struct super_block *, bool);
 extern struct inode *afs_iget(struct super_block *, struct key *,
 			      struct afs_fid *, struct afs_file_status *,
 			      struct afs_callback *,

fs/afs/mntpt.c

@@ -72,7 +72,7 @@ static int afs_mntpt_open(struct inode *inode, struct file *file)
  */
 static struct vfsmount *afs_mntpt_do_automount(struct dentry *mntpt)
 {
-	struct afs_super_info *super;
+	struct afs_super_info *as;
 	struct vfsmount *mnt;
 	struct afs_vnode *vnode;
 	struct page *page;
@@ -104,13 +104,13 @@ static struct vfsmount *afs_mntpt_do_automount(struct dentry *mntpt)
 			goto error_no_page;
 
 		if (mntpt->d_name.name[0] == '.') {
-			devname[0] = '#';
-			memcpy(devname + 1, mntpt->d_name.name, size - 1);
+			devname[0] = '%';
+			memcpy(devname + 1, mntpt->d_name.name + 1, size - 1);
 			memcpy(devname + size, afs_root_cell,
 			       sizeof(afs_root_cell));
 			rwpath = true;
 		} else {
-			devname[0] = '%';
+			devname[0] = '#';
 			memcpy(devname + 1, mntpt->d_name.name, size);
 			memcpy(devname + size + 1, afs_root_cell,
 			       sizeof(afs_root_cell));
@@ -142,11 +142,13 @@ static struct vfsmount *afs_mntpt_do_automount(struct dentry *mntpt)
 	}
 
 	/* work out what options we want */
-	super = AFS_FS_S(mntpt->d_sb);
-	memcpy(options, "cell=", 5);
-	strcpy(options + 5, super->volume->cell->name);
-	if (super->volume->type == AFSVL_RWVOL || rwpath)
-		strcat(options, ",rwpath");
+	as = AFS_FS_S(mntpt->d_sb);
+	if (as->cell) {
+		memcpy(options, "cell=", 5);
+		strcpy(options + 5, as->cell->name);
+		if ((as->volume && as->volume->type == AFSVL_RWVOL) || rwpath)
+			strcat(options, ",rwpath");
+	}
 
 	/* try and do the mount */
 	_debug("--- attempting mount %s -o %s ---", devname, options);