Unverified Commit 7d7d1649 authored by Christian Brauner's avatar Christian Brauner
Browse files

mnt: support ns lookup 



Move the mount namespace to the generic ns lookup infrastructure.
This allows us to drop a bunch of members from struct mnt_namespace.

Signed-off-by: default avatarChristian Brauner <brauner@kernel.org>
parent 7914f15c

fs/mount.h

+2 −8
Original line number Diff line number Diff line
@@ -17,11 +17,7 @@ struct mnt_namespace { 
	};

	struct user_namespace	*user_ns;

	struct ucounts		*ucounts;

	u64			seq;	/* Sequence number to prevent loops */

	union {

	wait_queue_head_t	poll;

		struct rcu_head		mnt_ns_rcu;

	};

	u64			seq_origin; /* Sequence number of origin mount namespace */

	u64 event;

#ifdef CONFIG_FSNOTIFY
@@ -30,8 +26,6 @@ struct mnt_namespace { 
#endif

	unsigned int		nr_mounts; /* # of mounts in the namespace */

	unsigned int		pending_mounts;

	struct rb_node		mnt_ns_tree_node; /* node in the mnt_ns_tree */

	struct list_head	mnt_ns_list; /* entry in the sequential list of mounts namespace */

	refcount_t		passive; /* number references not pinning @mounts */

} __randomize_layout;
@@ -173,7 +167,7 @@ static inline bool is_local_mountpoint(const struct dentry *dentry) 


static inline bool is_anon_ns(struct mnt_namespace *ns)

{

	return ns->seq == 0;

	return ns->ns.ns_id == 0;

}



static inline bool anon_ns_root(const struct mount *m)

fs/namespace.c

+29 −110
Original line number Diff line number Diff line
@@ -33,6 +33,7 @@ 
#include <linux/shmem_fs.h>

#include <linux/mnt_idmapping.h>

#include <linux/pidfs.h>

#include <linux/nstree.h>



#include "pnode.h"

#include "internal.h"
@@ -80,13 +81,10 @@ static DECLARE_RWSEM(namespace_sem); 
static HLIST_HEAD(unmounted);	/* protected by namespace_sem */

static LIST_HEAD(ex_mountpoints); /* protected by namespace_sem */

static struct mnt_namespace *emptied_ns; /* protected by namespace_sem */

static DEFINE_SEQLOCK(mnt_ns_tree_lock);



#ifdef CONFIG_FSNOTIFY

LIST_HEAD(notify_list); /* protected by namespace_sem */

#endif

static struct rb_root mnt_ns_tree = RB_ROOT; /* protected by mnt_ns_tree_lock */

static LIST_HEAD(mnt_ns_list); /* protected by mnt_ns_tree_lock */



enum mount_kattr_flags_t {

	MOUNT_KATTR_RECURSE		= (1 << 0),
@@ -119,53 +117,12 @@ __cacheline_aligned_in_smp DEFINE_SEQLOCK(mount_lock); 


static inline struct mnt_namespace *node_to_mnt_ns(const struct rb_node *node)

{

	struct ns_common *ns;



	if (!node)

		return NULL;

	return rb_entry(node, struct mnt_namespace, mnt_ns_tree_node);

}



static int mnt_ns_cmp(struct rb_node *a, const struct rb_node *b)

{

	struct mnt_namespace *ns_a = node_to_mnt_ns(a);

	struct mnt_namespace *ns_b = node_to_mnt_ns(b);

	u64 seq_a = ns_a->seq;

	u64 seq_b = ns_b->seq;



	if (seq_a < seq_b)

		return -1;

	if (seq_a > seq_b)

		return 1;

	return 0;

}



static inline void mnt_ns_tree_write_lock(void)

{

	write_seqlock(&mnt_ns_tree_lock);

}



static inline void mnt_ns_tree_write_unlock(void)

{

	write_sequnlock(&mnt_ns_tree_lock);

}



static void mnt_ns_tree_add(struct mnt_namespace *ns)

{

	struct rb_node *node, *prev;



	mnt_ns_tree_write_lock();

	node = rb_find_add_rcu(&ns->mnt_ns_tree_node, &mnt_ns_tree, mnt_ns_cmp);

	/*

	 * If there's no previous entry simply add it after the

	 * head and if there is add it after the previous entry.

	 */

	prev = rb_prev(&ns->mnt_ns_tree_node);

	if (!prev)

		list_add_rcu(&ns->mnt_ns_list, &mnt_ns_list);

	else

		list_add_rcu(&ns->mnt_ns_list, &node_to_mnt_ns(prev)->mnt_ns_list);

	mnt_ns_tree_write_unlock();



	WARN_ON_ONCE(node);

	ns = rb_entry(node, struct ns_common, ns_tree_node);

	return container_of(ns, struct mnt_namespace, ns);

}



static void mnt_ns_release(struct mnt_namespace *ns)
@@ -181,32 +138,16 @@ DEFINE_FREE(mnt_ns_release, struct mnt_namespace *, if (_T) mnt_ns_release(_T)) 


static void mnt_ns_release_rcu(struct rcu_head *rcu)

{

	mnt_ns_release(container_of(rcu, struct mnt_namespace, mnt_ns_rcu));

	mnt_ns_release(container_of(rcu, struct mnt_namespace, ns.ns_rcu));

}



static void mnt_ns_tree_remove(struct mnt_namespace *ns)

{

	/* remove from global mount namespace list */

	if (!RB_EMPTY_NODE(&ns->mnt_ns_tree_node)) {

		mnt_ns_tree_write_lock();

		rb_erase(&ns->mnt_ns_tree_node, &mnt_ns_tree);

		list_bidir_del_rcu(&ns->mnt_ns_list);

		mnt_ns_tree_write_unlock();

	}



	call_rcu(&ns->mnt_ns_rcu, mnt_ns_release_rcu);

}

	if (ns_tree_active(ns))

		ns_tree_remove(ns);



static int mnt_ns_find(const void *key, const struct rb_node *node)

{

	const u64 mnt_ns_id = *(u64 *)key;

	const struct mnt_namespace *ns = node_to_mnt_ns(node);



	if (mnt_ns_id < ns->seq)

		return -1;

	if (mnt_ns_id > ns->seq)

		return 1;

	return 0;

	call_rcu(&ns->ns.ns_rcu, mnt_ns_release_rcu);

}



/*
@@ -225,28 +166,21 @@ static int mnt_ns_find(const void *key, const struct rb_node *node) 
 */

static struct mnt_namespace *lookup_mnt_ns(u64 mnt_ns_id)

{

	struct mnt_namespace *ns;

	struct rb_node *node;

	unsigned int seq;

	struct mnt_namespace *mnt_ns;

	struct ns_common *ns;



	guard(rcu)();

	do {

		seq = read_seqbegin(&mnt_ns_tree_lock);

		node = rb_find_rcu(&mnt_ns_id, &mnt_ns_tree, mnt_ns_find);

		if (node)

			break;

	} while (read_seqretry(&mnt_ns_tree_lock, seq));



	if (!node)

	ns = ns_tree_lookup_rcu(mnt_ns_id, CLONE_NEWNS);

	if (!ns)

		return NULL;



	/*

	 * The last reference count is put with RCU delay so we can

	 * unconditonally acquire a reference here.

	 */

	ns = node_to_mnt_ns(node);

	refcount_inc(&ns->passive);

	return ns;

	mnt_ns = container_of(ns, struct mnt_namespace, ns);

	refcount_inc(&mnt_ns->passive);

	return mnt_ns;

}



static inline void lock_mount_hash(void)
@@ -1017,7 +951,7 @@ static inline bool check_anonymous_mnt(struct mount *mnt) 
		return false;



	seq = mnt->mnt_ns->seq_origin;

	return !seq || (seq == current->nsproxy->mnt_ns->seq);

	return !seq || (seq == current->nsproxy->mnt_ns->ns.ns_id);

}



/*
@@ -2152,19 +2086,16 @@ struct ns_common *from_mnt_ns(struct mnt_namespace *mnt) 


struct mnt_namespace *get_sequential_mnt_ns(struct mnt_namespace *mntns, bool previous)

{

	struct ns_common *ns;



	guard(rcu)();



	for (;;) {

		struct list_head *list;



		if (previous)

			list = rcu_dereference(list_bidir_prev_rcu(&mntns->mnt_ns_list));

		else

			list = rcu_dereference(list_next_rcu(&mntns->mnt_ns_list));

		if (list_is_head(list, &mnt_ns_list))

			return ERR_PTR(-ENOENT);

		ns = ns_tree_adjoined_rcu(mntns, previous);

		if (IS_ERR(ns))

			return ERR_CAST(ns);



		mntns = list_entry_rcu(list, struct mnt_namespace, mnt_ns_list);

		mntns = to_mnt_ns(ns);



		/*

		 * The last passive reference count is put with RCU
@@ -2204,7 +2135,7 @@ static bool mnt_ns_loop(struct dentry *dentry) 
	if (!mnt_ns)

		return false;



	return current->nsproxy->mnt_ns->seq >= mnt_ns->seq;

	return current->nsproxy->mnt_ns->ns.ns_id >= mnt_ns->ns.ns_id;

}



struct mount *copy_tree(struct mount *src_root, struct dentry *dentry,
@@ -3080,7 +3011,7 @@ static struct file *open_detached_copy(struct path *path, bool recursive) 
		if (is_anon_ns(src_mnt_ns))

			ns->seq_origin = src_mnt_ns->seq_origin;

		else

			ns->seq_origin = src_mnt_ns->seq;

			ns->seq_origin = src_mnt_ns->ns.ns_id;

	}



	mnt = __do_loopback(path, recursive);
@@ -4156,15 +4087,6 @@ static void free_mnt_ns(struct mnt_namespace *ns) 
	mnt_ns_tree_remove(ns);

}



/*

 * Assign a sequence number so we can detect when we attempt to bind

 * mount a reference to an older mount namespace into the current

 * mount namespace, preventing reference counting loops.  A 64bit

 * number incrementing at 10Ghz will take 12,427 years to wrap which

 * is effectively never, so we can ignore the possibility.

 */

static atomic64_t mnt_ns_seq = ATOMIC64_INIT(1);



static struct mnt_namespace *alloc_mnt_ns(struct user_namespace *user_ns, bool anon)

{

	struct mnt_namespace *new_ns;
@@ -4188,11 +4110,9 @@ static struct mnt_namespace *alloc_mnt_ns(struct user_namespace *user_ns, bool a 
		return ERR_PTR(ret);

	}

	if (!anon)

		new_ns->seq = atomic64_inc_return(&mnt_ns_seq);

		ns_tree_gen_id(&new_ns->ns);

	refcount_set(&new_ns->passive, 1);

	new_ns->mounts = RB_ROOT;

	INIT_LIST_HEAD(&new_ns->mnt_ns_list);

	RB_CLEAR_NODE(&new_ns->mnt_ns_tree_node);

	init_waitqueue_head(&new_ns->poll);

	new_ns->user_ns = get_user_ns(user_ns);

	new_ns->ucounts = ucounts;
@@ -4278,7 +4198,7 @@ struct mnt_namespace *copy_mnt_ns(unsigned long flags, struct mnt_namespace *ns, 
	if (pwdmnt)

		mntput(pwdmnt);



	mnt_ns_tree_add(new_ns);

	ns_tree_add_raw(new_ns);

	return new_ns;

}
@@ -5397,7 +5317,7 @@ static int statmount_sb_source(struct kstatmount *s, struct seq_file *seq) 
static void statmount_mnt_ns_id(struct kstatmount *s, struct mnt_namespace *ns)

{

	s->sm.mask |= STATMOUNT_MNT_NS_ID;

	s->sm.mnt_ns_id = ns->seq;

	s->sm.mnt_ns_id = ns->ns.ns_id;

}



static int statmount_mnt_opts(struct kstatmount *s, struct seq_file *seq)
@@ -6102,7 +6022,6 @@ static void __init init_mount_tree(void) 
	ns = alloc_mnt_ns(&init_user_ns, true);

	if (IS_ERR(ns))

		panic("Can't allocate initial namespace");

	ns->seq = atomic64_inc_return(&mnt_ns_seq);

	ns->ns.inum = PROC_MNT_INIT_INO;

	m = real_mount(mnt);

	ns->root = m;
@@ -6117,7 +6036,7 @@ static void __init init_mount_tree(void) 
	set_fs_pwd(current->fs, &root);

	set_fs_root(current->fs, &root);



	mnt_ns_tree_add(ns);

	ns_tree_add(ns);

}



void __init mnt_init(void)

fs/nsfs.c

+2 −2
Original line number Diff line number Diff line
@@ -139,7 +139,7 @@ static int copy_ns_info_to_user(const struct mnt_namespace *mnt_ns, 
	 * the size value will be set to the size the kernel knows about.

	 */

	kinfo->size		= min(usize, sizeof(*kinfo));

	kinfo->mnt_ns_id	= mnt_ns->seq;

	kinfo->mnt_ns_id	= mnt_ns->ns.ns_id;

	kinfo->nr_mounts	= READ_ONCE(mnt_ns->nr_mounts);

	/* Subtract the root mount of the mount namespace. */

	if (kinfo->nr_mounts)
@@ -221,7 +221,7 @@ static long ns_ioctl(struct file *filp, unsigned int ioctl, 


		mnt_ns = container_of(ns, struct mnt_namespace, ns);

		idp = (__u64 __user *)arg;

		id = mnt_ns->seq;

		id = mnt_ns->ns.ns_id;

		return put_user(id, idp);

	}

	case NS_GET_PID_FROM_PIDNS: