[PATCH 17/17] RCU'd vfsmounts

From: Al Viro
Date: Thu Oct 03 2013 - 02:20:25 EST


_very_ preliminary, barely tested.

Signed-off-by: Al Viro <viro@xxxxxxxxxxxxxxxxxx>

diff --git a/fs/adfs/super.c b/fs/adfs/super.c
index 0ff4bae..3b79d15 100644
--- a/fs/adfs/super.c
+++ b/fs/adfs/super.c
@@ -123,6 +123,7 @@ static void adfs_put_super(struct super_block *sb)
for (i = 0; i < asb->s_map_size; i++)
brelse(asb->s_map[i].dm_bh);
kfree(asb->s_map);
+ synchronize_rcu();
kfree(asb);
sb->s_fs_info = NULL;
}
diff --git a/fs/autofs4/inode.c b/fs/autofs4/inode.c
index b104726..07599e2 100644
--- a/fs/autofs4/inode.c
+++ b/fs/autofs4/inode.c
@@ -62,6 +62,7 @@ void autofs4_kill_sb(struct super_block *sb)
/* Free wait queues, close pipe */
autofs4_catatonic_mode(sbi);

+ synchronize_rcu();
sb->s_fs_info = NULL;
kfree(sbi);

diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
index a279ffc..e0305ae 100644
--- a/fs/cifs/connect.c
+++ b/fs/cifs/connect.c
@@ -3779,6 +3779,7 @@ cifs_umount(struct cifs_sb_info *cifs_sb)

bdi_destroy(&cifs_sb->bdi);
kfree(cifs_sb->mountdata);
+ synchronize_rcu();
unload_nls(cifs_sb->local_nls);
kfree(cifs_sb);
}
diff --git a/fs/dcache.c b/fs/dcache.c
index d888223..ae74923 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -1075,116 +1075,6 @@ void shrink_dcache_sb(struct super_block *sb)
EXPORT_SYMBOL(shrink_dcache_sb);

/*
- * destroy a single subtree of dentries for unmount
- * - see the comments on shrink_dcache_for_umount() for a description of the
- * locking
- */
-static void shrink_dcache_for_umount_subtree(struct dentry *dentry)
-{
- struct dentry *parent;
-
- BUG_ON(!IS_ROOT(dentry));
-
- for (;;) {
- /* descend to the first leaf in the current subtree */
- while (!list_empty(&dentry->d_subdirs))
- dentry = list_entry(dentry->d_subdirs.next,
- struct dentry, d_u.d_child);
-
- /* consume the dentries from this leaf up through its parents
- * until we find one with children or run out altogether */
- do {
- struct inode *inode;
-
- /*
- * inform the fs that this dentry is about to be
- * unhashed and destroyed.
- */
- if ((dentry->d_flags & DCACHE_OP_PRUNE) &&
- !d_unhashed(dentry))
- dentry->d_op->d_prune(dentry);
-
- dentry_lru_del(dentry);
- __d_shrink(dentry);
-
- if (dentry->d_lockref.count != 0) {
- printk(KERN_ERR
- "BUG: Dentry %p{i=%lx,n=%s}"
- " still in use (%d)"
- " [unmount of %s %s]\n",
- dentry,
- dentry->d_inode ?
- dentry->d_inode->i_ino : 0UL,
- dentry->d_name.name,
- dentry->d_lockref.count,
- dentry->d_sb->s_type->name,
- dentry->d_sb->s_id);
- BUG();
- }
-
- if (IS_ROOT(dentry)) {
- parent = NULL;
- list_del(&dentry->d_u.d_child);
- } else {
- parent = dentry->d_parent;
- parent->d_lockref.count--;
- list_del(&dentry->d_u.d_child);
- }
-
- inode = dentry->d_inode;
- if (inode) {
- dentry->d_inode = NULL;
- hlist_del_init(&dentry->d_alias);
- if (dentry->d_op && dentry->d_op->d_iput)
- dentry->d_op->d_iput(dentry, inode);
- else
- iput(inode);
- }
-
- d_free(dentry);
-
- /* finished when we fall off the top of the tree,
- * otherwise we ascend to the parent and move to the
- * next sibling if there is one */
- if (!parent)
- return;
- dentry = parent;
- } while (list_empty(&dentry->d_subdirs));
-
- dentry = list_entry(dentry->d_subdirs.next,
- struct dentry, d_u.d_child);
- }
-}
-
-/*
- * destroy the dentries attached to a superblock on unmounting
- * - we don't need to use dentry->d_lock because:
- * - the superblock is detached from all mountings and open files, so the
- * dentry trees will not be rearranged by the VFS
- * - s_umount is write-locked, so the memory pressure shrinker will ignore
- * any dentries belonging to this superblock that it comes across
- * - the filesystem itself is no longer permitted to rearrange the dentries
- * in this superblock
- */
-void shrink_dcache_for_umount(struct super_block *sb)
-{
- struct dentry *dentry;
-
- if (down_read_trylock(&sb->s_umount))
- BUG();
-
- dentry = sb->s_root;
- sb->s_root = NULL;
- dentry->d_lockref.count--;
- shrink_dcache_for_umount_subtree(dentry);
-
- while (!hlist_bl_empty(&sb->s_anon)) {
- dentry = hlist_bl_entry(hlist_bl_first(&sb->s_anon), struct dentry, d_hash);
- shrink_dcache_for_umount_subtree(dentry);
- }
-}
-
-/*
* This tries to ascend one level of parenthood, but
* we can race with renaming, so we need to re-check
* the parenthood after dropping the lock and check
@@ -1478,6 +1368,90 @@ void shrink_dcache_parent(struct dentry *parent)
}
EXPORT_SYMBOL(shrink_dcache_parent);

+static enum d_walk_ret umount_collect(void *_data, struct dentry *dentry)
+{
+ struct select_data *data = _data;
+ enum d_walk_ret ret = D_WALK_CONTINUE;
+
+ if (dentry->d_lockref.count) {
+ dentry_lru_del(dentry);
+ if (likely(!list_empty(&dentry->d_subdirs)))
+ goto out;
+ if (dentry == data->start && dentry->d_lockref.count == 1)
+ goto out;
+ printk(KERN_ERR
+ "BUG: Dentry %p{i=%lx,n=%s}"
+ " still in use (%d)"
+ " [unmount of %s %s]\n",
+ dentry,
+ dentry->d_inode ?
+ dentry->d_inode->i_ino : 0UL,
+ dentry->d_name.name,
+ dentry->d_lockref.count,
+ dentry->d_sb->s_type->name,
+ dentry->d_sb->s_id);
+ BUG();
+ } else if (!(dentry->d_flags & DCACHE_SHRINK_LIST)) {
+ /*
+ * We can't use d_lru_shrink_move() because we
+ * need to get the global LRU lock and do the
+ * LRU accounting.
+ */
+ d_lru_del(dentry);
+ d_shrink_add(dentry, &data->dispose);
+ data->found++;
+ ret = D_WALK_NORETRY;
+ }
+out:
+ if (data->found && need_resched())
+ ret = D_WALK_QUIT;
+ return ret;
+}
+
+/*
+ * destroy the dentries attached to a superblock on unmounting
+ */
+void shrink_dcache_for_umount(struct super_block *sb)
+{
+ struct dentry *dentry;
+
+ if (down_read_trylock(&sb->s_umount))
+ BUG();
+
+ dentry = sb->s_root;
+ sb->s_root = NULL;
+ for (;;) {
+ struct select_data data;
+
+ INIT_LIST_HEAD(&data.dispose);
+ data.start = dentry;
+ data.found = 0;
+
+ d_walk(dentry, &data, umount_collect, NULL);
+ if (!data.found)
+ break;
+
+ shrink_dentry_list(&data.dispose);
+ cond_resched();
+ }
+ d_drop(dentry);
+ dput(dentry);
+
+ while (!hlist_bl_empty(&sb->s_anon)) {
+ struct select_data data;
+ dentry = hlist_bl_entry(hlist_bl_first(&sb->s_anon), struct dentry, d_hash);
+
+ INIT_LIST_HEAD(&data.dispose);
+ data.start = NULL;
+ data.found = 0;
+
+ d_walk(dentry, &data, umount_collect, NULL);
+ if (data.found)
+ shrink_dentry_list(&data.dispose);
+ cond_resched();
+ }
+}
+
static enum d_walk_ret check_and_collect(void *_data, struct dentry *dentry)
{
struct select_data *data = _data;
@@ -2885,24 +2859,28 @@ static int prepend_path(const struct path *path,
struct vfsmount *vfsmnt = path->mnt;
struct mount *mnt = real_mount(vfsmnt);
int error = 0;
- unsigned seq = 0;
+ unsigned seq, m_seq = 0;
char *bptr;
int blen;

- br_read_lock(&vfsmount_lock);
rcu_read_lock();
+restart_mnt:
+ read_seqbegin_or_lock(&mount_lock, &m_seq);
+ seq = 0;
restart:
bptr = *buffer;
blen = *buflen;
+ error = 0;
read_seqbegin_or_lock(&rename_lock, &seq);
while (dentry != root->dentry || vfsmnt != root->mnt) {
struct dentry * parent;

if (dentry == vfsmnt->mnt_root || IS_ROOT(dentry)) {
+ struct mount *parent = ACCESS_ONCE(mnt->mnt_parent);
/* Global root? */
- if (mnt_has_parent(mnt)) {
- dentry = mnt->mnt_mountpoint;
- mnt = mnt->mnt_parent;
+ if (mnt != parent) {
+ dentry = ACCESS_ONCE(mnt->mnt_mountpoint);
+ mnt = parent;
vfsmnt = &mnt->mnt;
continue;
}
@@ -2936,7 +2914,11 @@ restart:
goto restart;
}
done_seqretry(&rename_lock, seq);
- br_read_unlock(&vfsmount_lock);
+ if (need_seqretry(&mount_lock, m_seq)) {
+ m_seq = 1;
+ goto restart_mnt;
+ }
+ done_seqretry(&mount_lock, m_seq);

if (error >= 0 && bptr == *buffer) {
if (--blen < 0)
diff --git a/fs/fat/inode.c b/fs/fat/inode.c
index 0062da2..3d297e6 100644
--- a/fs/fat/inode.c
+++ b/fs/fat/inode.c
@@ -557,6 +557,7 @@ static void fat_put_super(struct super_block *sb)
iput(sbi->fsinfo_inode);
iput(sbi->fat_inode);

+ synchronize_rcu();
unload_nls(sbi->nls_disk);
unload_nls(sbi->nls_io);

diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c
index a8ce6da..2dfd2b4 100644
--- a/fs/fuse/inode.c
+++ b/fs/fuse/inode.c
@@ -387,6 +387,7 @@ static void fuse_put_super(struct super_block *sb)
mutex_unlock(&fuse_mutex);
fuse_bdi_destroy(fc);

+ synchronize_rcu();
fuse_conn_put(fc);
}

diff --git a/fs/hpfs/super.c b/fs/hpfs/super.c
index 4334cda..2946c6b 100644
--- a/fs/hpfs/super.c
+++ b/fs/hpfs/super.c
@@ -109,6 +109,7 @@ static void hpfs_put_super(struct super_block *s)
unmark_dirty(s);
hpfs_unlock(s);

+ synchronize_rcu();
kfree(sbi->sb_cp_table);
kfree(sbi->sb_bmp_dir);
s->s_fs_info = NULL;
diff --git a/fs/mount.h b/fs/mount.h
index f086607..d64c594 100644
--- a/fs/mount.h
+++ b/fs/mount.h
@@ -1,7 +1,6 @@
#include <linux/mount.h>
#include <linux/seq_file.h>
#include <linux/poll.h>
-#include <linux/lglock.h>

struct mnt_namespace {
atomic_t count;
@@ -30,6 +29,7 @@ struct mount {
struct mount *mnt_parent;
struct dentry *mnt_mountpoint;
struct vfsmount mnt;
+ struct rcu_head mnt_rcu;
#ifdef CONFIG_SMP
struct mnt_pcp __percpu *mnt_pcp;
#else
@@ -80,21 +80,23 @@ static inline int is_mounted(struct vfsmount *mnt)
extern struct mount *__lookup_mnt(struct vfsmount *, struct dentry *);
extern struct mount *__lookup_mnt_last(struct vfsmount *, struct dentry *);

+extern bool legitimize_mnt(struct vfsmount *, unsigned);
+
static inline void get_mnt_ns(struct mnt_namespace *ns)
{
atomic_inc(&ns->count);
}

-extern struct lglock vfsmount_lock;
+extern seqlock_t mount_lock;

static inline void lock_mount_hash(void)
{
- br_write_lock(&vfsmount_lock);
+ write_seqlock(&mount_lock);
}

static inline void unlock_mount_hash(void)
{
- br_write_unlock(&vfsmount_lock);
+ write_sequnlock(&mount_lock);
}

struct proc_mounts {
diff --git a/fs/namei.c b/fs/namei.c
index 1f844fb..4b4310a 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -482,18 +482,6 @@ EXPORT_SYMBOL(path_put);
* to restart the path walk from the beginning in ref-walk mode.
*/

-static inline void lock_rcu_walk(void)
-{
- br_read_lock(&vfsmount_lock);
- rcu_read_lock();
-}
-
-static inline void unlock_rcu_walk(void)
-{
- rcu_read_unlock();
- br_read_unlock(&vfsmount_lock);
-}
-
/**
* unlazy_walk - try to switch to ref-walk mode.
* @nd: nameidata pathwalk data
@@ -512,26 +500,23 @@ static int unlazy_walk(struct nameidata *nd, struct dentry *dentry)
BUG_ON(!(nd->flags & LOOKUP_RCU));

/*
- * Get a reference to the parent first: we're
- * going to make "path_put(nd->path)" valid in
- * non-RCU context for "terminate_walk()".
- *
- * If this doesn't work, return immediately with
- * RCU walking still active (and then we will do
- * the RCU walk cleanup in terminate_walk()).
+ * After legitimizing the bastards, terminate_walk()
+ * will do the right thing for non-RCU mode, and all our
+ * subsequent exit cases should rcu_read_unlock()
+ * before returning. Do vfsmount first; if dentry
+ * can't be legitimized, just set nd->path.dentry to NULL
+ * and rely on dput(NULL) being a no-op.
*/
- if (!lockref_get_not_dead(&parent->d_lockref))
+ if (!legitimize_mnt(nd->path.mnt, nd->m_seq))
return -ECHILD;
-
- /*
- * After the mntget(), we terminate_walk() will do
- * the right thing for non-RCU mode, and all our
- * subsequent exit cases should unlock_rcu_walk()
- * before returning.
- */
- mntget(nd->path.mnt);
nd->flags &= ~LOOKUP_RCU;

+ if (!lockref_get_not_dead(&parent->d_lockref)) {
+ nd->path.dentry = NULL;
+ rcu_read_unlock();
+ return -ECHILD;
+ }
+
/*
* For a negative lookup, the lookup sequence point is the parents
* sequence point, and it only needs to revalidate the parent dentry.
@@ -566,17 +551,17 @@ static int unlazy_walk(struct nameidata *nd, struct dentry *dentry)
spin_unlock(&fs->lock);
}

- unlock_rcu_walk();
+ rcu_read_unlock();
return 0;

unlock_and_drop_dentry:
spin_unlock(&fs->lock);
drop_dentry:
- unlock_rcu_walk();
+ rcu_read_unlock();
dput(dentry);
goto drop_root_mnt;
out:
- unlock_rcu_walk();
+ rcu_read_unlock();
drop_root_mnt:
if (!(nd->flags & LOOKUP_ROOT))
nd->root.mnt = NULL;
@@ -608,17 +593,22 @@ static int complete_walk(struct nameidata *nd)
if (!(nd->flags & LOOKUP_ROOT))
nd->root.mnt = NULL;

+ if (!legitimize_mnt(nd->path.mnt, nd->m_seq)) {
+ rcu_read_unlock();
+ return -ECHILD;
+ }
if (unlikely(!lockref_get_not_dead(&dentry->d_lockref))) {
- unlock_rcu_walk();
+ rcu_read_unlock();
+ mntput(nd->path.mnt);
return -ECHILD;
}
if (read_seqcount_retry(&dentry->d_seq, nd->seq)) {
- unlock_rcu_walk();
+ rcu_read_unlock();
dput(dentry);
+ mntput(nd->path.mnt);
return -ECHILD;
}
- mntget(nd->path.mnt);
- unlock_rcu_walk();
+ rcu_read_unlock();
}

if (likely(!(nd->flags & LOOKUP_JUMPED)))
@@ -909,15 +899,15 @@ int follow_up(struct path *path)
struct mount *parent;
struct dentry *mountpoint;

- br_read_lock(&vfsmount_lock);
+ read_seqlock_excl(&mount_lock);
parent = mnt->mnt_parent;
if (parent == mnt) {
- br_read_unlock(&vfsmount_lock);
+ read_sequnlock_excl(&mount_lock);
return 0;
}
mntget(&parent->mnt);
mountpoint = dget(mnt->mnt_mountpoint);
- br_read_unlock(&vfsmount_lock);
+ read_sequnlock_excl(&mount_lock);
dput(path->dentry);
path->dentry = mountpoint;
mntput(path->mnt);
@@ -1048,8 +1038,8 @@ static int follow_managed(struct path *path, unsigned flags)

/* Something is mounted on this dentry in another
* namespace and/or whatever was mounted there in this
- * namespace got unmounted before we managed to get the
- * vfsmount_lock */
+ * namespace got unmounted before lookup_mnt() could
+ * get it */
}

/* Handle an automount point */
@@ -1174,7 +1164,7 @@ failed:
nd->flags &= ~LOOKUP_RCU;
if (!(nd->flags & LOOKUP_ROOT))
nd->root.mnt = NULL;
- unlock_rcu_walk();
+ rcu_read_unlock();
return -ECHILD;
}

@@ -1501,7 +1491,7 @@ static void terminate_walk(struct nameidata *nd)
nd->flags &= ~LOOKUP_RCU;
if (!(nd->flags & LOOKUP_ROOT))
nd->root.mnt = NULL;
- unlock_rcu_walk();
+ rcu_read_unlock();
}
}

@@ -1862,7 +1852,7 @@ static int path_init(int dfd, const char *name, unsigned int flags,
nd->path = nd->root;
nd->inode = inode;
if (flags & LOOKUP_RCU) {
- lock_rcu_walk();
+ rcu_read_lock();
nd->seq = __read_seqcount_begin(&nd->path.dentry->d_seq);
} else {
path_get(&nd->path);
@@ -1872,9 +1862,10 @@ static int path_init(int dfd, const char *name, unsigned int flags,

nd->root.mnt = NULL;

+ nd->m_seq = read_seqbegin(&mount_lock);
if (*name=='/') {
if (flags & LOOKUP_RCU) {
- lock_rcu_walk();
+ rcu_read_lock();
set_root_rcu(nd);
} else {
set_root(nd);
@@ -1886,7 +1877,7 @@ static int path_init(int dfd, const char *name, unsigned int flags,
struct fs_struct *fs = current->fs;
unsigned seq;

- lock_rcu_walk();
+ rcu_read_lock();

do {
seq = read_seqcount_begin(&fs->seq);
@@ -1918,7 +1909,7 @@ static int path_init(int dfd, const char *name, unsigned int flags,
if (f.need_put)
*fp = f.file;
nd->seq = __read_seqcount_begin(&nd->path.dentry->d_seq);
- lock_rcu_walk();
+ rcu_read_lock();
} else {
path_get(&nd->path);
fdput(f);
diff --git a/fs/namespace.c b/fs/namespace.c
index 8ae16b9f..1711536 100644
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -53,7 +53,7 @@ EXPORT_SYMBOL_GPL(fs_kobj);
* It should be taken for write in all cases where the vfsmount
* tree or hash is modified or when a vfsmount structure is modified.
*/
-DEFINE_BRLOCK(vfsmount_lock);
+__cacheline_aligned_in_smp DEFINE_SEQLOCK(mount_lock);

static inline unsigned long hash(struct vfsmount *mnt, struct dentry *dentry)
{
@@ -547,16 +547,38 @@ static void free_vfsmnt(struct mount *mnt)
kmem_cache_free(mnt_cache, mnt);
}

+/* call under rcu_read_lock */
+bool legitimize_mnt(struct vfsmount *bastard, unsigned seq)
+{
+ struct mount *mnt;
+ if (read_seqretry(&mount_lock, seq))
+ return false;
+ if (bastard == NULL)
+ return true;
+ mnt = real_mount(bastard);
+ mnt_add_count(mnt, 1);
+ if (likely(!read_seqretry(&mount_lock, seq)))
+ return true;
+ if (bastard->mnt_flags & MNT_SYNC_UMOUNT) {
+ mnt_add_count(mnt, -1);
+ return false;
+ }
+ rcu_read_unlock();
+ mntput(bastard);
+ rcu_read_lock();
+ return false;
+}
+
/*
* find the first mount at @dentry on vfsmount @mnt.
- * vfsmount_lock must be held for read or write.
+ * call under rcu_read_lock()
*/
struct mount *__lookup_mnt(struct vfsmount *mnt, struct dentry *dentry)
{
struct list_head *head = mount_hashtable + hash(mnt, dentry);
struct mount *p;

- list_for_each_entry(p, head, mnt_hash)
+ list_for_each_entry_rcu(p, head, mnt_hash)
if (&p->mnt_parent->mnt == mnt && p->mnt_mountpoint == dentry)
return p;
return NULL;
@@ -564,7 +586,7 @@ struct mount *__lookup_mnt(struct vfsmount *mnt, struct dentry *dentry)

/*
* find the last mount at @dentry on vfsmount @mnt.
- * vfsmount_lock must be held for read or write.
+ * mount_lock must be held.
*/
struct mount *__lookup_mnt_last(struct vfsmount *mnt, struct dentry *dentry)
{
@@ -596,17 +618,17 @@ struct mount *__lookup_mnt_last(struct vfsmount *mnt, struct dentry *dentry)
struct vfsmount *lookup_mnt(struct path *path)
{
struct mount *child_mnt;
+ struct vfsmount *m;
+ unsigned seq;

- br_read_lock(&vfsmount_lock);
- child_mnt = __lookup_mnt(path->mnt, path->dentry);
- if (child_mnt) {
- mnt_add_count(child_mnt, 1);
- br_read_unlock(&vfsmount_lock);
- return &child_mnt->mnt;
- } else {
- br_read_unlock(&vfsmount_lock);
- return NULL;
- }
+ 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;
}

static struct mountpoint *new_mountpoint(struct dentry *dentry)
@@ -874,38 +896,47 @@ static struct mount *clone_mnt(struct mount *old, struct dentry *root,
return ERR_PTR(err);
}

+static void delayed_free(struct rcu_head *head)
+{
+ struct mount *mnt = container_of(head, struct mount, mnt_rcu);
+ kfree(mnt->mnt_devname);
+#ifdef CONFIG_SMP
+ free_percpu(mnt->mnt_pcp);
+#endif
+ kmem_cache_free(mnt_cache, mnt);
+}
+
static void mntput_no_expire(struct mount *mnt)
{
put_again:
-#ifdef CONFIG_SMP
- br_read_lock(&vfsmount_lock);
- if (likely(mnt->mnt_ns)) {
- /* shouldn't be the last one */
- mnt_add_count(mnt, -1);
- br_read_unlock(&vfsmount_lock);
+ rcu_read_lock();
+ mnt_add_count(mnt, -1);
+ smp_mb();
+ if (likely(mnt->mnt_ns)) { /* shouldn't be the last one */
+ rcu_read_unlock();
return;
}
- br_read_unlock(&vfsmount_lock);
-
lock_mount_hash();
- mnt_add_count(mnt, -1);
if (mnt_get_count(mnt)) {
+ rcu_read_unlock();
unlock_mount_hash();
return;
}
-#else
- mnt_add_count(mnt, -1);
- if (likely(mnt_get_count(mnt)))
- return;
- lock_mount_hash();
-#endif
if (unlikely(mnt->mnt_pinned)) {
mnt_add_count(mnt, mnt->mnt_pinned + 1);
mnt->mnt_pinned = 0;
+ rcu_read_unlock();
unlock_mount_hash();
acct_auto_close_mnt(&mnt->mnt);
goto put_again;
}
+ if (unlikely(mnt->mnt.mnt_flags & MNT_DOOMED)) {
+ rcu_read_unlock();
+ unlock_mount_hash();
+ return;
+ }
+ mnt->mnt.mnt_flags |= MNT_DOOMED;
+ rcu_read_unlock();

list_del(&mnt->mnt_instance);
unlock_mount_hash();
@@ -924,7 +955,8 @@ put_again:
fsnotify_vfsmount_delete(&mnt->mnt);
dput(mnt->mnt.mnt_root);
deactivate_super(mnt->mnt.mnt_sb);
- free_vfsmnt(mnt);
+ mnt_free_id(mnt);
+ call_rcu(&mnt->mnt_rcu, delayed_free);
}

void mntput(struct vfsmount *mnt)
@@ -1137,6 +1169,8 @@ static void namespace_unlock(void)
list_splice_init(&unmounted, &head);
up_write(&namespace_sem);

+ synchronize_rcu();
+
while (!list_empty(&head)) {
mnt = list_first_entry(&head, struct mount, mnt_hash);
list_del_init(&mnt->mnt_hash);
@@ -1152,10 +1186,13 @@ static inline void namespace_lock(void)
}

/*
- * vfsmount lock must be held for write
+ * mount_lock must be held
* namespace_sem must be held for write
+ * how = 0 => just this tree, don't propagate
+ * how = 1 => propagate; we know that nobody else has reference to any victims
+ * how = 2 => lazy umount
*/
-void umount_tree(struct mount *mnt, int propagate)
+void umount_tree(struct mount *mnt, int how)
{
LIST_HEAD(tmp_list);
struct mount *p;
@@ -1163,7 +1200,7 @@ void umount_tree(struct mount *mnt, int propagate)
for (p = mnt; p; p = next_mnt(p, mnt))
list_move(&p->mnt_hash, &tmp_list);

- if (propagate)
+ if (how)
propagate_umount(&tmp_list);

list_for_each_entry(p, &tmp_list, mnt_hash) {
@@ -1171,6 +1208,8 @@ void umount_tree(struct mount *mnt, int propagate)
list_del_init(&p->mnt_list);
__touch_mnt_namespace(p->mnt_ns);
p->mnt_ns = NULL;
+ if (how < 2)
+ p->mnt.mnt_flags |= MNT_SYNC_UMOUNT;
list_del_init(&p->mnt_child);
if (mnt_has_parent(p)) {
put_mountpoint(p->mnt_mp);
@@ -1262,14 +1301,18 @@ static int do_umount(struct mount *mnt, int flags)
lock_mount_hash();
event++;

- if (!(flags & MNT_DETACH))
- shrink_submounts(mnt);
-
- retval = -EBUSY;
- if (flags & MNT_DETACH || !propagate_mount_busy(mnt, 2)) {
+ if (flags & MNT_DETACH) {
if (!list_empty(&mnt->mnt_list))
- umount_tree(mnt, 1);
+ umount_tree(mnt, 2);
retval = 0;
+ } else {
+ shrink_submounts(mnt);
+ retval = -EBUSY;
+ if (!propagate_mount_busy(mnt, 2)) {
+ if (!list_empty(&mnt->mnt_list))
+ umount_tree(mnt, 1);
+ retval = 0;
+ }
}
unlock_mount_hash();
namespace_unlock();
@@ -1955,7 +1998,7 @@ static int do_add_mount(struct mount *newmnt, struct path *path, int mnt_flags)
struct mount *parent;
int err;

- mnt_flags &= ~(MNT_SHARED | MNT_WRITE_HOLD | MNT_INTERNAL);
+ mnt_flags &= ~(MNT_SHARED | MNT_WRITE_HOLD | MNT_INTERNAL | MNT_DOOMED | MNT_SYNC_UMOUNT);

mp = lock_mount(path);
if (IS_ERR(mp))
@@ -2172,7 +2215,7 @@ resume:
* process a list of expirable mountpoints with the intent of discarding any
* submounts of a specific parent mountpoint
*
- * vfsmount_lock must be held for write
+ * mount_lock must be held for write
*/
static void shrink_submounts(struct mount *mnt)
{
@@ -2558,7 +2601,7 @@ out_type:
/*
* Return true if path is reachable from root
*
- * namespace_sem or vfsmount_lock is held
+ * namespace_sem or mount_lock is held
*/
bool is_path_reachable(struct mount *mnt, struct dentry *dentry,
const struct path *root)
@@ -2573,9 +2616,9 @@ bool is_path_reachable(struct mount *mnt, struct dentry *dentry,
int path_is_under(struct path *path1, struct path *path2)
{
int res;
- br_read_lock(&vfsmount_lock);
+ read_seqlock_excl(&mount_lock);
res = is_path_reachable(real_mount(path1->mnt), path1->dentry, path2);
- br_read_unlock(&vfsmount_lock);
+ read_sequnlock_excl(&mount_lock);
return res;
}
EXPORT_SYMBOL(path_is_under);
@@ -2748,8 +2791,6 @@ void __init mnt_init(void)
for (u = 0; u < HASH_SIZE; u++)
INIT_LIST_HEAD(&mountpoint_hashtable[u]);

- br_lock_init(&vfsmount_lock);
-
err = sysfs_init();
if (err)
printk(KERN_WARNING "%s: sysfs_init error: %d\n",
@@ -2783,6 +2824,7 @@ struct vfsmount *kern_mount_data(struct file_system_type *type, void *data)
* we unmount before file sys is unregistered
*/
real_mount(mnt)->mnt_ns = MNT_NS_INTERNAL;
+ smp_wmb();
}
return mnt;
}
@@ -2792,9 +2834,8 @@ void kern_unmount(struct vfsmount *mnt)
{
/* release long term mount so mount point can be released */
if (!IS_ERR_OR_NULL(mnt)) {
- lock_mount_hash();
real_mount(mnt)->mnt_ns = NULL;
- unlock_mount_hash();
+ smp_mb();
mntput(mnt);
}
}
diff --git a/fs/ncpfs/inode.c b/fs/ncpfs/inode.c
index 4659da6..5539b5b 100644
--- a/fs/ncpfs/inode.c
+++ b/fs/ncpfs/inode.c
@@ -792,6 +792,7 @@ static void ncp_put_super(struct super_block *sb)

ncp_stop_tasks(server);

+ synchronize_rcu();
#ifdef CONFIG_NCPFS_NLS
/* unload the NLS charsets */
unload_nls(server->nls_vol);
diff --git a/fs/proc/root.c b/fs/proc/root.c
index 87dbcbe..8d1e094 100644
--- a/fs/proc/root.c
+++ b/fs/proc/root.c
@@ -148,6 +148,7 @@ static void proc_kill_sb(struct super_block *sb)
if (ns->proc_self)
dput(ns->proc_self);
kill_anon_super(sb);
+ synchronize_rcu(); /* might be an overkill */
put_pid_ns(ns);
}

diff --git a/include/linux/mount.h b/include/linux/mount.h
index 38cd98f..371d346 100644
--- a/include/linux/mount.h
+++ b/include/linux/mount.h
@@ -49,6 +49,8 @@ struct mnt_namespace;

#define MNT_LOCK_READONLY 0x400000
#define MNT_LOCKED 0x800000
+#define MNT_DOOMED 0x1000000
+#define MNT_SYNC_UMOUNT 0x2000000

struct vfsmount {
struct dentry *mnt_root; /* root of the mounted tree */
diff --git a/include/linux/namei.h b/include/linux/namei.h
index 8e47bc7..492de72 100644
--- a/include/linux/namei.h
+++ b/include/linux/namei.h
@@ -16,7 +16,7 @@ struct nameidata {
struct path root;
struct inode *inode; /* path.dentry.d_inode */
unsigned int flags;
- unsigned seq;
+ unsigned seq, m_seq;
int last_type;
unsigned depth;
char *saved_names[MAX_NESTED_LINKS + 1];
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/