diff --git a/Documentation/filesystems/porting b/Documentation/filesystems/porting
index 266d2059b9b8d29a387ce2ef057fd995057542e8..dfbcd1b00b0ae3eac43f0744646f44a88f81cfcc 100644
--- a/Documentation/filesystems/porting
+++ b/Documentation/filesystems/porting
@@ -365,8 +365,8 @@ must be done in the RCU callback.
[recommended]
vfs now tries to do path walking in "rcu-walk mode", which avoids
atomic operations and scalability hazards on dentries and inodes (see
-Documentation/filesystems/path-walk.txt). d_hash and d_compare changes (above)
-are examples of the changes required to support this. For more complex
+Documentation/filesystems/path-lookup.txt). d_hash and d_compare changes
+(above) are examples of the changes required to support this. For more complex
filesystem callbacks, the vfs drops out of rcu-walk mode before the fs call, so
no changes are required to the filesystem. However, this is costly and loses
the benefits of rcu-walk mode. We will begin to add filesystem callbacks that
@@ -383,8 +383,8 @@ Documentation/filesystems/vfs.txt for more details.
permission and check_acl are inode permission checks that are called
on many or all directory inodes on the way down a path walk (to check for
-exec permission). These must now be rcu-walk aware (flags & IPERM_RCU). See
-Documentation/filesystems/vfs.txt for more details.
+exec permission). These must now be rcu-walk aware (flags & IPERM_FLAG_RCU).
+See Documentation/filesystems/vfs.txt for more details.
--
[mandatory]
diff --git a/Documentation/filesystems/vfs.txt b/Documentation/filesystems/vfs.txt
index fbb324e2bd43e6d1926a13f344ea5d55f6ce9360..cae6d27c9f5bfa006f3ee8a9b086e9627a6d0f2c 100644
--- a/Documentation/filesystems/vfs.txt
+++ b/Documentation/filesystems/vfs.txt
@@ -415,8 +415,8 @@ otherwise noted.
permission: called by the VFS to check for access rights on a POSIX-like
filesystem.
- May be called in rcu-walk mode (flags & IPERM_RCU). If in rcu-walk
- mode, the filesystem must check the permission without blocking or
+ May be called in rcu-walk mode (flags & IPERM_FLAG_RCU). If in rcu-walk
+ mode, the filesystem must check the permission without blocking or
storing to the inode.
If a situation is encountered that rcu-walk cannot handle, return
diff --git a/fs/namei.c b/fs/namei.c
index 0b14f6910fc68363cfd6da4799d9a0782b342be0..86643302079e33e010d29ae18d2fb392723c57bb 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -479,6 +479,14 @@ static int nameidata_dentry_drop_rcu(struct nameidata *nd, struct dentry *dentry
struct fs_struct *fs = current->fs;
struct dentry *parent = nd->path.dentry;
+ /*
+ * It can be possible to revalidate the dentry that we started
+ * the path walk with. force_reval_path may also revalidate the
+ * dentry already committed to the nameidata.
+ */
+ if (unlikely(parent == dentry))
+ return nameidata_drop_rcu(nd);
+
BUG_ON(!(nd->flags & LOOKUP_RCU));
if (nd->root.mnt) {
spin_lock(&fs->lock);
@@ -583,6 +591,13 @@ void release_open_intent(struct nameidata *nd)
fput(nd->intent.open.file);
}
+/*
+ * Call d_revalidate and handle filesystems that request rcu-walk
+ * to be dropped. This may be called and return in rcu-walk mode,
+ * regardless of success or error. If -ECHILD is returned, the caller
+ * must return -ECHILD back up the path walk stack so path walk may
+ * be restarted in ref-walk mode.
+ */
static int d_revalidate(struct dentry *dentry, struct nameidata *nd)
{
int status;
@@ -673,6 +688,9 @@ force_reval_path(struct path *path, struct nameidata *nd)
return 0;
if (!status) {
+ /* Don't d_invalidate in rcu-walk mode */
+ if (nameidata_drop_rcu(nd))
+ return -ECHILD;
d_invalidate(dentry);
status = -ESTALE;
}
@@ -2105,11 +2123,13 @@ static struct file *do_last(struct nameidata *nd, struct path *path,
dir = nd->path.dentry;
case LAST_DOT:
if (need_reval_dot(dir)) {
- error = d_revalidate(nd->path.dentry, nd);
- if (!error)
- error = -ESTALE;
- if (error < 0)
+ int status = d_revalidate(nd->path.dentry, nd);
+ if (!status)
+ status = -ESTALE;
+ if (status < 0) {
+ error = status;
goto exit;
+ }
}
/* fallthrough */
case LAST_ROOT:
diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 64ee240f3c80776748936db8929344426d970079..df8c03a02161c45270d1ff45c53945308fcd8208 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -1406,11 +1406,15 @@ static struct dentry *nfs_atomic_lookup(struct inode *dir, struct dentry *dentry
static int nfs_open_revalidate(struct dentry *dentry, struct nameidata *nd)
{
struct dentry *parent = NULL;
- struct inode *inode = dentry->d_inode;
+ struct inode *inode;
struct inode *dir;
struct nfs_open_context *ctx;
int openflags, ret = 0;
+ if (nd->flags & LOOKUP_RCU)
+ return -ECHILD;
+
+ inode = dentry->d_inode;
if (!is_atomic_open(nd) || d_mountpoint(dentry))
goto no_open;
diff --git a/include/linux/list_bl.h b/include/linux/list_bl.h
index 9ee97e7f2be4711901720fffb55d713b702bc252..b2adbb4b2f731f49d3b5f54918ef5b4a4508aa71 100644
--- a/include/linux/list_bl.h
+++ b/include/linux/list_bl.h
@@ -16,7 +16,7 @@
* some fast and compact auxiliary data.
*/
-#if defined(CONFIG_SMP) || defined(CONFIG_DEBUG_SPINLOCK)
+#if defined(CONFIG_SMP)
#define LIST_BL_LOCKMASK 1UL
#else
#define LIST_BL_LOCKMASK 0UL
@@ -62,7 +62,8 @@ static inline void hlist_bl_set_first(struct hlist_bl_head *h,
struct hlist_bl_node *n)
{
LIST_BL_BUG_ON((unsigned long)n & LIST_BL_LOCKMASK);
- LIST_BL_BUG_ON(!((unsigned long)h->first & LIST_BL_LOCKMASK));
+ LIST_BL_BUG_ON(((unsigned long)h->first & LIST_BL_LOCKMASK) !=
+ LIST_BL_LOCKMASK);
h->first = (struct hlist_bl_node *)((unsigned long)n | LIST_BL_LOCKMASK);
}
diff --git a/include/linux/rculist_bl.h b/include/linux/rculist_bl.h
index b872b493724d3360f46be7c0bc16cd5793214842..cf1244fbf3b66cc444ad80854d7e0126d45cfb09 100644
--- a/include/linux/rculist_bl.h
+++ b/include/linux/rculist_bl.h
@@ -11,7 +11,8 @@ static inline void hlist_bl_set_first_rcu(struct hlist_bl_head *h,
struct hlist_bl_node *n)
{
LIST_BL_BUG_ON((unsigned long)n & LIST_BL_LOCKMASK);
- LIST_BL_BUG_ON(!((unsigned long)h->first & LIST_BL_LOCKMASK));
+ LIST_BL_BUG_ON(((unsigned long)h->first & LIST_BL_LOCKMASK) !=
+ LIST_BL_LOCKMASK);
rcu_assign_pointer(h->first,
(struct hlist_bl_node *)((unsigned long)n | LIST_BL_LOCKMASK));
}