Re: [RFC][PATCH 1/8] namespace: use global namespace semaphore

[Mike Waychison <mike@waychison.com>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Ian Kent wrote:
> On Mon, 30 May 2005, Miklos Szeredi wrote:
> 
> 
>>This patch removes the per-namespace semaphore in favor of a global
>>semaphore.  This can have an effect on scalability, however it is
>>necessary for the following patches.
>>
>>A similar scheme can be added later if needed.
> 
> 
> Which, together with patch 3, effectively removes the ability for code in 
> a module to safely find out about vfsmounts in its namespace.
> 
> Given the assumption in link_path_walk, that a vfsmount will not be 
> followed prior to a call to follow_link this stops me from doing 
> autofs direct mount via this method.
> 
> Any ideas on how I might cope with this?

No, the removal of struct namespace shouldn't keep you from walking
vfsmounts from modules.  In fact, in Miklos' 1/8 patch, he moved
namespace->sem to a global semaphore, which isn't needed by the end of
the patch set.  namespace->sem only ever protected the namespace->list
field.

If you which to walk a mounttree, you still only ever require grabbing
vfsmount_lock and use mntget/mntput.

Overall, this means autofs4 stuff will need to be updated to reflect
these changes, but they are minimal at best.

> 
> 
>>Signed-off-by: Miklos Szeredi <miklos@szeredi.hu>
>>
>>Index: linux/include/linux/namespace.h
>>===================================================================
>>--- linux.orig/include/linux/namespace.h	2005-05-29 13:34:30.000000000 +0200
>>+++ linux/include/linux/namespace.h	2005-05-29 13:46:23.000000000 +0200
>>@@ -9,7 +9,6 @@ struct namespace {
>> 	atomic_t		count;
>> 	struct vfsmount *	root;
>> 	struct list_head	list;
>>-	struct rw_semaphore	sem;
>> };
>> 
>> extern int copy_namespace(int, struct task_struct *);
>>Index: linux/fs/namespace.c
>>===================================================================
>>--- linux.orig/fs/namespace.c	2005-05-29 13:42:33.000000000 +0200
>>+++ linux/fs/namespace.c	2005-05-29 14:06:50.000000000 +0200
>>@@ -42,6 +42,7 @@ static inline int sysfs_init(void)
>> static struct list_head *mount_hashtable;
>> static int hash_mask, hash_bits;
>> static kmem_cache_t *mnt_cache; 
>>+static struct rw_semaphore namespace_sem;
>> 
>> static inline unsigned long hash(struct vfsmount *mnt, struct dentry *dentry)
>> {
>>@@ -189,7 +190,7 @@ static void *m_start(struct seq_file *m,
>> 	struct list_head *p;
>> 	loff_t l = *pos;
>> 
>>-	down_read(&n->sem);
>>+	down_read(&namespace_sem);
>> 	list_for_each(p, &n->list)
>> 		if (!l--)
>> 			return list_entry(p, struct vfsmount, mnt_list);
>>@@ -206,8 +207,7 @@ static void *m_next(struct seq_file *m, 
>> 
>> static void m_stop(struct seq_file *m, void *v)
>> {
>>-	struct namespace *n = m->private;
>>-	up_read(&n->sem);
>>+	up_read(&namespace_sem);
>> }
>> 
>> static inline void mangle(struct seq_file *m, const char *s)
>>@@ -432,7 +432,7 @@ static int do_umount(struct vfsmount *mn
>> 		return retval;
>> 	}
>> 
>>-	down_write(&current->namespace->sem);
>>+	down_write(&namespace_sem);
>> 	spin_lock(&vfsmount_lock);
>> 
>> 	if (atomic_read(&sb->s_active) == 1) {
>>@@ -454,7 +454,7 @@ static int do_umount(struct vfsmount *mn
>> 	spin_unlock(&vfsmount_lock);
>> 	if (retval)
>> 		security_sb_umount_busy(mnt);
>>-	up_write(&current->namespace->sem);
>>+	up_write(&namespace_sem);
>> 	return retval;
>> }
>> 
>>@@ -632,7 +632,7 @@ static int do_loopback(struct nameidata 
>> 	if (err)
>> 		return err;
>> 
>>-	down_write(&current->namespace->sem);
>>+	down_write(&namespace_sem);
>> 	err = -EINVAL;
>> 	if (check_mnt(nd->mnt) && (!recurse || check_mnt(old_nd.mnt))) {
>> 		err = -ENOMEM;
>>@@ -657,7 +657,7 @@ static int do_loopback(struct nameidata 
>> 			mntput(mnt);
>> 	}
>> 
>>-	up_write(&current->namespace->sem);
>>+	up_write(&namespace_sem);
>> 	path_release(&old_nd);
>> 	return err;
>> }
>>@@ -706,7 +706,7 @@ static int do_move_mount(struct nameidat
>> 	if (err)
>> 		return err;
>> 
>>-	down_write(&current->namespace->sem);
>>+	down_write(&namespace_sem);
>> 	while(d_mountpoint(nd->dentry) && follow_down(&nd->mnt, &nd->dentry))
>> 		;
>> 	err = -EINVAL;
>>@@ -750,7 +750,7 @@ out2:
>> out1:
>> 	up(&nd->dentry->d_inode->i_sem);
>> out:
>>-	up_write(&current->namespace->sem);
>>+	up_write(&namespace_sem);
>> 	if (!err)
>> 		path_release(&parent_nd);
>> 	path_release(&old_nd);
>>@@ -789,7 +789,7 @@ int do_add_mount(struct vfsmount *newmnt
>> {
>> 	int err;
>> 
>>-	down_write(&current->namespace->sem);
>>+	down_write(&namespace_sem);
>> 	/* Something was mounted here while we slept */
>> 	while(d_mountpoint(nd->dentry) && follow_down(&nd->mnt, &nd->dentry))
>> 		;
>>@@ -819,7 +819,7 @@ int do_add_mount(struct vfsmount *newmnt
>> 	}
>> 
>> unlock:
>>-	up_write(&current->namespace->sem);
>>+	up_write(&namespace_sem);
>> 	mntput(newmnt);
>> 	return err;
>> }
>>@@ -875,7 +875,6 @@ static void expire_mount(struct vfsmount
>>  */
>> void mark_mounts_for_expiry(struct list_head *mounts)
>> {
>>-	struct namespace *namespace;
>> 	struct vfsmount *mnt, *next;
>> 	LIST_HEAD(graveyard);
>> 
>>@@ -909,20 +908,12 @@ void mark_mounts_for_expiry(struct list_
>> 		mnt = list_entry(graveyard.next, struct vfsmount, mnt_expire);
>> 		list_del_init(&mnt->mnt_expire);
>> 
>>-		/* don't do anything if the namespace is dead - all the
>>-		 * vfsmounts from it are going away anyway */
>>-		namespace = mnt->mnt_namespace;
>>-		if (!namespace || !namespace->root)
>>-			continue;
>>-		get_namespace(namespace);
>>-
>> 		spin_unlock(&vfsmount_lock);
>>-		down_write(&namespace->sem);
>>+		down_write(&namespace_sem);
>> 		expire_mount(mnt, mounts);
>>-		up_write(&namespace->sem);
>>+		up_write(&namespace_sem);
>> 
>> 		mntput(mnt);
>>-		put_namespace(namespace);
>> 
>> 		spin_lock(&vfsmount_lock);
>> 	}
>>@@ -1087,14 +1078,13 @@ int copy_namespace(int flags, struct tas
>> 		goto out;
>> 
>> 	atomic_set(&new_ns->count, 1);
>>-	init_rwsem(&new_ns->sem);
>> 	INIT_LIST_HEAD(&new_ns->list);
>> 
>>-	down_write(&tsk->namespace->sem);
>>+	down_write(&namespace_sem);
>> 	/* First pass: copy the tree topology */
>> 	new_ns->root = copy_tree(namespace->root, namespace->root->mnt_root);
>> 	if (!new_ns->root) {
>>-		up_write(&tsk->namespace->sem);
>>+		up_write(&namespace_sem);
>> 		kfree(new_ns);
>> 		goto out;
>> 	}
>>@@ -1128,7 +1118,7 @@ int copy_namespace(int flags, struct tas
>> 		p = next_mnt(p, namespace->root);
>> 		q = next_mnt(q, new_ns->root);
>> 	}
>>-	up_write(&tsk->namespace->sem);
>>+	up_write(&namespace_sem);
>> 
>> 	tsk->namespace = new_ns;
>> 
>>@@ -1310,7 +1300,7 @@ asmlinkage long sys_pivot_root(const cha
>> 	user_nd.mnt = mntget(current->fs->rootmnt);
>> 	user_nd.dentry = dget(current->fs->root);
>> 	read_unlock(&current->fs->lock);
>>-	down_write(&current->namespace->sem);
>>+	down_write(&namespace_sem);
>> 	down(&old_nd.dentry->d_inode->i_sem);
>> 	error = -EINVAL;
>> 	if (!check_mnt(user_nd.mnt))
>>@@ -1356,7 +1346,7 @@ asmlinkage long sys_pivot_root(const cha
>> 	path_release(&parent_nd);
>> out2:
>> 	up(&old_nd.dentry->d_inode->i_sem);
>>-	up_write(&current->namespace->sem);
>>+	up_write(&namespace_sem);
>> 	path_release(&user_nd);
>> 	path_release(&old_nd);
>> out1:
>>@@ -1383,7 +1373,6 @@ static void __init init_mount_tree(void)
>> 		panic("Can't allocate initial namespace");
>> 	atomic_set(&namespace->count, 1);
>> 	INIT_LIST_HEAD(&namespace->list);
>>-	init_rwsem(&namespace->sem);
>> 	list_add(&mnt->mnt_list, &namespace->list);
>> 	namespace->root = mnt;
>> 	mnt->mnt_namespace = namespace;
>>@@ -1406,6 +1395,8 @@ void __init mnt_init(unsigned long mempa
>> 	unsigned int nr_hash;
>> 	int i;
>> 
>>+	init_rwsem(&namespace_sem);
>>+
>> 	mnt_cache = kmem_cache_create("mnt_cache", sizeof(struct vfsmount),
>> 			0, SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL, NULL);
>> 
>>@@ -1454,10 +1445,10 @@ void __put_namespace(struct namespace *n
>> 	struct vfsmount *root = namespace->root;
>> 	namespace->root = NULL;
>> 	spin_unlock(&vfsmount_lock);
>>-	down_write(&namespace->sem);
>>+	down_write(&namespace_sem);
>> 	spin_lock(&vfsmount_lock);
>> 	umount_tree(root);
>> 	spin_unlock(&vfsmount_lock);
>>-	up_write(&namespace->sem);
>>+	up_write(&namespace_sem);
>> 	kfree(namespace);
>> }
>>-
>>To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
>>the body of a message to majordomo@vger.kernel.org
>>More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>
> 
> 

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFCm9i3dQs4kOxk3/MRAlESAJ9rItxr1YbOfOJCWvArhlSmlNUM0QCfezPN
GyXzPaNSzLAUx/sInVOhRkU=
=v1Ix
-----END PGP SIGNATURE-----