lock_mount.md

lock_mount()

Given a path as the parameter, this function is to get the deepest mount at path.

static struct mountpoint *lock_mount(struct path *path)
{
	struct vfsmount *mnt;
	struct dentry *dentry = path->dentry;
retry:
	inode_lock(dentry->d_inode);
	if (unlikely(cant_mount(dentry))) {
		inode_unlock(dentry->d_inode);
		return ERR_PTR(-ENOENT);
	}
	namespace_lock();
	mnt = lookup_mnt(path);
	if (likely(!mnt)) {
		struct mountpoint *mp = get_mountpoint(dentry);
		if (IS_ERR(mp)) {
			namespace_unlock();
			inode_unlock(dentry->d_inode);
			return mp;
		}
		return mp;
	}
	namespace_unlock();
	inode_unlock(path->dentry->d_inode);
	path_put(path);
	path->mnt = mnt;
	dentry = path->dentry = dget(mnt->mnt_root);
	goto retry;
}

It's a loop, every iteration lookup_mnt() find a mount at path->dentry and then update the path->mnt and path->dentry, this loops until there is no mount, which means we reaches the root dentry of the last mount at that directory. After that, it creates a new struct mountpoint and return it.

lookup_mnt()

/*
 * lookup_mnt - Return the first child mount mounted at path
 *
 * "First" means first mounted chronologically.  If you create the
 * following mounts:
 *
 * mount /dev/sda1 /mnt
 * mount /dev/sda2 /mnt
 * mount /dev/sda3 /mnt
 *
 * Then lookup_mnt() on the base /mnt dentry in the root mount will
 * return successively the root dentry and vfsmount of /dev/sda1, then
 * /dev/sda2, then /dev/sda3, then NULL.
 *
 * lookup_mnt takes a reference to the found vfsmount.
 */
struct vfsmount *lookup_mnt(const struct path *path)
{
	struct mount *child_mnt;
	struct vfsmount *m;
	unsigned seq;

	rcu_read_lock();
	do {
		seq = read_seqbegin(&mount_lock);
		child_mnt = __lookup_mnt(path->mnt, path->dentry);
		m = child_mnt ? &child_mnt->mnt : NULL;
	} while (!legitimize_mnt(m, seq));
	rcu_read_unlock();
	return m;
}

From the comments above, we can clearly see what this function does. The work of look up a mount at path is given to __lookup_mnt(), and it is in a seqlock reader section. What legitimize_mnt() does is checking if the seq changes when we read the data and if unluckly so, recover the state(undo the change we have done).

__lookup_mnt()

/*
 * find the first mount at @dentry on vfsmount @mnt.
 * call under rcu_read_lock()
 */
struct mount *__lookup_mnt(struct vfsmount *mnt, struct dentry *dentry)
{
	struct hlist_head *head = m_hash(mnt, dentry);
	struct mount *p;

	hlist_for_each_entry_rcu(p, head, mnt_hash)
		if (&p->mnt_parent->mnt == mnt && p->mnt_mountpoint == dentry)
			return p;
	return NULL;
}

This one is simple, just try to find the mount by path->mnt and path->dentry in the global mount_hashtable.

get_mountpoint()

At last, we reaches the root dentry of the last mount, we create a new mountpoint for later use.

static struct mountpoint *get_mountpoint(struct dentry *dentry)
{
	struct mountpoint *mp, *new = NULL;
	int ret;

	if (d_mountpoint(dentry)) {
		/* might be worth a WARN_ON() */
		if (d_unlinked(dentry))
			return ERR_PTR(-ENOENT);
mountpoint:
		read_seqlock_excl(&mount_lock);
		mp = lookup_mountpoint(dentry);
		read_sequnlock_excl(&mount_lock);
		if (mp)
			goto done;
	}

	if (!new)
		new = kmalloc(sizeof(struct mountpoint), GFP_KERNEL);
	if (!new)
		return ERR_PTR(-ENOMEM);


	/* Exactly one processes may set d_mounted */
	ret = d_set_mounted(dentry);

	/* Someone else set d_mounted? */
	if (ret == -EBUSY)
		goto mountpoint;

	/* The dentry is not available as a mountpoint? */
	mp = ERR_PTR(ret);
	if (ret)
		goto done;

	/* Add the new mountpoint to the hash table */
	read_seqlock_excl(&mount_lock);
	new->m_dentry = dget(dentry);
	new->m_count = 1;
	hlist_add_head(&new->m_hash, mp_hash(dentry));
	INIT_HLIST_HEAD(&new->m_list);
	read_sequnlock_excl(&mount_lock);

	mp = new;
	new = NULL;
done:
	kfree(new);
	return mp;
}

Since we call this function when dentry is not a mount point, the first if is bypassed. Let's look at the rest part. It basically does these things:

• alloc a new struct mountpoint
• set dentry->d_flags |= DCACHE_MOUNTED
• set new->m_dentry = dget(dentry)
• set new->m_count = 1
• add new to the global mountpoint_hashtable by new->m_hash the key of mountpoint_hashtable is the dentry where the mountpoint locates.
• return new (the struct mountpoint)