[RFC PATCH] add filesystem subtype support

[RFC PATCH] add filesystem subtype support

[Miklos Szeredi]

There's a slight problem with filesystem type representation in fuse
based filesystems.

>From the kernel's view, there are just two filesystem types: fuse and
fuseblk.  From the user's view there are lots of different filesystem
types.  The user is not even much concerned if the filesystem is fuse
based or not.  So there's a conflict of interest in how this should be
represented in fstab, mtab and /proc/mounts.

The current scheme is to encode the real filesystem type in the mount
source.  So an sshfs mount looks like this:

  sshfs#user@server:/   /mnt/server    fuse   rw,nosuid,nodev,...

This url-ish syntax works OK for sshfs and similar filesystems.
However for block device based filesystems (ntfs-3g, zfs) it doesn't
work, since the kernel expects the mount source to be a real device
name.

A possibly better scheme would be to encode the real type in the type
field as "type.subtype".  So fuse mounts would look like this:

  /dev/hda1       /mnt/windows   fuseblk.ntfs-3g   rw,...
  user@server:/   /mnt/server    fuse.sshfs        rw,nosuid,nodev,...

This patch adds the necessary code to the kernel so that this can be
correctly displayed in /proc/mounts.

---
Index: linux/fs/filesystems.c
===================================================================
--- linux.orig/fs/filesystems.c	2007-02-12 12:42:55.000000000 +0100
+++ linux/fs/filesystems.c	2007-02-12 12:43:00.000000000 +0100
@@ -42,11 +42,12 @@ void put_filesystem(struct file_system_t
 	module_put(fs->owner);
 }
 
-static struct file_system_type **find_filesystem(const char *name)
+static struct file_system_type **find_filesystem(const char *name, unsigned len)
 {
 	struct file_system_type **p;
 	for (p=&file_systems; *p; p=&(*p)->next)
-		if (strcmp((*p)->name,name) == 0)
+		if (strlen((*p)->name) == len &&
+		    strncmp((*p)->name, name, len) == 0)
 			break;
 	return p;
 }
@@ -69,11 +70,12 @@ int register_filesystem(struct file_syst
 	int res = 0;
 	struct file_system_type ** p;
 
+	BUG_ON(strchr(fs->name, '.'));
 	if (fs->next)
 		return -EBUSY;
 	INIT_LIST_HEAD(&fs->fs_supers);
 	write_lock(&file_systems_lock);
-	p = find_filesystem(fs->name);
+	p = find_filesystem(fs->name, strlen(fs->name));
 	if (*p)
 		res = -EBUSY;
 	else
@@ -216,19 +218,26 @@ int get_filesystem_list(char * buf)
 struct file_system_type *get_fs_type(const char *name)
 {
 	struct file_system_type *fs;
+	const char *dot = strchr(name, '.');
+	unsigned len = dot ? dot - name : strlen(name);
 
 	read_lock(&file_systems_lock);
-	fs = *(find_filesystem(name));
+	fs = *(find_filesystem(name, len));
 	if (fs && !try_module_get(fs->owner))
 		fs = NULL;
 	read_unlock(&file_systems_lock);
-	if (!fs && (request_module("%s", name) == 0)) {
+	if (!fs && (request_module("%.*s", len, name) == 0)) {
 		read_lock(&file_systems_lock);
-		fs = *(find_filesystem(name));
+		fs = *(find_filesystem(name, len));
 		if (fs && !try_module_get(fs->owner))
 			fs = NULL;
 		read_unlock(&file_systems_lock);
 	}
+
+	if (dot && fs && !(fs->fs_flags & FS_HAS_SUBTYPE)) {
+		put_filesystem(fs);
+		fs = NULL;
+	}
 	return fs;
 }
 
Index: linux/fs/fuse/inode.c
===================================================================
--- linux.orig/fs/fuse/inode.c	2007-02-12 12:42:55.000000000 +0100
+++ linux/fs/fuse/inode.c	2007-02-12 12:43:00.000000000 +0100
@@ -634,6 +634,7 @@ static int fuse_get_sb(struct file_syste
 static struct file_system_type fuse_fs_type = {
 	.owner		= THIS_MODULE,
 	.name		= "fuse",
+	.fs_flags	= FS_HAS_SUBTYPE,
 	.get_sb		= fuse_get_sb,
 	.kill_sb	= kill_anon_super,
 };
@@ -650,6 +651,7 @@ static int fuse_get_sb_blk(struct file_s
 static struct file_system_type fuseblk_fs_type = {
 	.owner		= THIS_MODULE,
 	.name		= "fuseblk",
+	.fs_flags	= FS_HAS_SUBTYPE,
 	.get_sb		= fuse_get_sb_blk,
 	.kill_sb	= kill_block_super,
 	.fs_flags	= FS_REQUIRES_DEV,
Index: linux/fs/namespace.c
===================================================================
--- linux.orig/fs/namespace.c	2007-02-12 12:42:55.000000000 +0100
+++ linux/fs/namespace.c	2007-02-12 12:43:00.000000000 +0100
@@ -378,6 +378,10 @@ static int show_vfsmnt(struct seq_file *
 	seq_path(m, mnt, mnt->mnt_root, " \t\n\\");
 	seq_putc(m, ' ');
 	mangle(m, mnt->mnt_sb->s_type->name);
+	if (mnt->mnt_sb->s_subtype && mnt->mnt_sb->s_subtype[0]) {
+		seq_putc(m, '.');
+		mangle(m, mnt->mnt_sb->s_subtype);
+	}
 	seq_puts(m, mnt->mnt_sb->s_flags & MS_RDONLY ? " ro" : " rw");
 	for (fs_infop = fs_info; fs_infop->flag; fs_infop++) {
 		if (mnt->mnt_sb->s_flags & fs_infop->flag)
Index: linux/fs/super.c
===================================================================
--- linux.orig/fs/super.c	2007-02-12 12:42:55.000000000 +0100
+++ linux/fs/super.c	2007-02-12 12:43:00.000000000 +0100
@@ -107,6 +107,7 @@ out:
 static inline void destroy_super(struct super_block *s)
 {
 	security_sb_free(s);
+	kfree(s->s_subtype);
 	kfree(s);
 }
 
@@ -919,6 +920,29 @@ out:
 
 EXPORT_SYMBOL_GPL(vfs_kern_mount);
 
+static struct vfsmount *fs_set_subtype(struct vfsmount *mnt, const char *fstype)
+{
+	int err;
+	const char *subtype = strchr(fstype, '.');
+	if (subtype) {
+		subtype++;
+		err = -EINVAL;
+		if (!subtype[0])
+			goto err;
+	} else
+		subtype = "";
+
+	mnt->mnt_sb->s_subtype = kstrdup(subtype, GFP_KERNEL);
+	err = -ENOMEM;
+	if (!mnt->mnt_sb->s_subtype)
+		goto err;
+	return mnt;
+
+ err:
+	mntput(mnt);
+	return ERR_PTR(err);
+}
+
 struct vfsmount *
 do_kern_mount(const char *fstype, int flags, const char *name, void *data)
 {
@@ -927,6 +951,9 @@ do_kern_mount(const char *fstype, int fl
 	if (!type)
 		return ERR_PTR(-ENODEV);
 	mnt = vfs_kern_mount(type, flags, name, data);
+	if (!IS_ERR(mnt) && (type->fs_flags & FS_HAS_SUBTYPE) &&
+	    !mnt->mnt_sb->s_subtype)
+		mnt = fs_set_subtype(mnt, fstype);
 	put_filesystem(type);
 	return mnt;
 }
Index: linux/include/linux/fs.h
===================================================================
--- linux.orig/include/linux/fs.h	2007-02-12 12:42:55.000000000 +0100
+++ linux/include/linux/fs.h	2007-02-12 12:43:00.000000000 +0100
@@ -91,6 +91,7 @@ extern int dir_notify_enable;
 /* public flags for file_system_type */
 #define FS_REQUIRES_DEV 1 
 #define FS_BINARY_MOUNTDATA 2
+#define FS_HAS_SUBTYPE 4
 #define FS_REVAL_DOT	16384	/* Check the paths ".", ".." for staleness */
 #define FS_RENAME_DOES_D_MOVE	32768	/* FS will handle d_move()
 					 * during rename() internally.
@@ -950,6 +951,12 @@ struct super_block {
 	/* Granularity of c/m/atime in ns.
 	   Cannot be worse than a second */
 	u32		   s_time_gran;
+
+	/*
+	 * Filesystem subtype.  If non-empty the filesystem type field
+	 * in /proc/mounts will be "type.subtype"
+	 */
+	char *s_subtype;
 };
 
 extern struct timespec current_fs_time(struct super_block *sb);

Re: [RFC PATCH] add filesystem subtype support

[Szakacsits Szabolcs]

Hi,

On Mon, 12 Feb 2007, Miklos Szeredi wrote:

> There's a slight problem with filesystem type representation in fuse
> based filesystems.
> 
> >From the kernel's view, there are just two filesystem types: fuse and
> fuseblk.  From the user's view there are lots of different filesystem
> types.  The user is not even much concerned if the filesystem is fuse
> based or not.  

Yes. Those who are concerned with the fstype and mount like

    mount -t fstype device mountpoint

apparently expect mount/fstab line like

    device mountpoint fstype ...

Of course the fstype could be fuse.subtype or fuseblk.subtype but that 
would add a needless complexity (also, for example ntfs-3g uses both and 
it decides run-time which one to use).

> So there's a conflict of interest in how this should be
> represented in fstab, mtab and /proc/mounts.
> 
> The current scheme is to encode the real filesystem type in the mount
> source.  So an sshfs mount looks like this:
> 
>   sshfs#user@server:/   /mnt/server    fuse   rw,nosuid,nodev,...
> 
> This url-ish syntax works OK for sshfs and similar filesystems.
> However for block device based filesystems (ntfs-3g, zfs) it doesn't
> work, since the kernel expects the mount source to be a real device
> name.
> 
> A possibly better scheme would be to encode the real type in the type
> field as "type.subtype".  So fuse mounts would look like this:
> 
>   /dev/hda1       /mnt/windows   fuseblk.ntfs-3g   rw,...
>   user@server:/   /mnt/server    fuse.sshfs        rw,nosuid,nodev,...

I think it's definitely an improvement because it solves real problems, 
though perhaps not the way users would expect.

	Szaka

Re: [RFC PATCH] add filesystem subtype support

[Miklos Szeredi]

> > There's a slight problem with filesystem type representation in fuse
> > based filesystems.
> > 
> > >From the kernel's view, there are just two filesystem types: fuse and
> > fuseblk.  From the user's view there are lots of different filesystem
> > types.  The user is not even much concerned if the filesystem is fuse
> > based or not.  
> 
> Yes. Those who are concerned with the fstype and mount like
> 
>     mount -t fstype device mountpoint
> 
> apparently expect mount/fstab line like
> 
>     device mountpoint fstype ...

The problem with supporting 

  /dev/hda1   /mnt/windows   ntfs-3g   rw,...

type syntax in /proc/mounts is that it would add much more complexity
to the kernel.  The same could be achieved in userspace by teaching
the mount utility about special filesystems like ntfs-3g.

This patch aims for the minium footprint for a feature, that really
doesn't concern the kernel at all.

Miklos

Re: [RFC PATCH] add filesystem subtype support

[Jan Engelhardt]

Hi,

On Feb 12 2007 12:50, Miklos Szeredi wrote:
>Index: linux/fs/filesystems.c
>===================================================================
>--- linux.orig/fs/filesystems.c	2007-02-12 12:42:55.000000000 +0100
>+++ linux/fs/filesystems.c	2007-02-12 12:43:00.000000000 +0100
>@@ -42,11 +42,12 @@ void put_filesystem(struct file_system_t
> 	module_put(fs->owner);
> }
> 
>-static struct file_system_type **find_filesystem(const char *name)
>+static struct file_system_type **find_filesystem(const char *name, unsigned len)
> {
> 	struct file_system_type **p;
> 	for (p=&file_systems; *p; p=&(*p)->next)
>-		if (strcmp((*p)->name,name) == 0)
>+		if (strlen((*p)->name) == len &&
>+		    strncmp((*p)->name, name, len) == 0)
> 			break;
> 	return p;
> }

Question btw, why does this function return a struct file_system_type ** at
all? Would not struct file_system_type * suffice?


I like the idea. Then the "real filesystem type" (e.g. sshfs) pops up in the
fstype field in getmntent, and `df`, for example.


Jan
-- 
ft: http://freshmeat.net/p/chaostables/

Re: [RFC PATCH] add filesystem subtype support

[Miklos Szeredi]

> >-static struct file_system_type **find_filesystem(const char *name)
> >+static struct file_system_type **find_filesystem(const char *name, unsigned len)
> > {
> > 	struct file_system_type **p;
> > 	for (p=&file_systems; *p; p=&(*p)->next)
> >-		if (strcmp((*p)->name,name) == 0)
> >+		if (strlen((*p)->name) == len &&
> >+		    strncmp((*p)->name, name, len) == 0)
> > 			break;
> > 	return p;
> > }
> 
> Question btw, why does this function return a struct file_system_type ** at
> all? Would not struct file_system_type * suffice?

It's used in register_filesystem() to get the end of the list pointer.

Miklos