[PATCH] VFS: allow ->d_manage() to declare -EISDIR in rcu_walk mode.

From: NeilBrown
Date: Wed Jul 30 2014 - 02:08:52 EST

In REF-walk mode, ->d_manage can return -EISDIR to indicate
that the dentry is not really a mount trap (or even a mount point)
and that any mounts or any DCACHE_NEED_AUTOMOUNT flag should be

RCU-walk mode doesn't currently support this, so if there is a dentry
with DCACHE_NEED_AUTOMOUNT set but which shouldn't be a mount-trap,
lookup_fast() will always drop in REF-walk mode.

With this patch, an -EISDIR from ->d_manage will always cause mounts
and automounts to be ignored, both in REF-walk and RCU-walk.

Cc: Ian Kent <raven@xxxxxxxxxx>
Signed-off-by: NeilBrown <neilb@xxxxxxx>

Hi Al,
this patch is needed before I can make autofs4 fully support RCU-walk.
There are cases currently were directories have DCACHE_NEED_AUTOMOUNT but for which
no automount is required.


diff --git a/Documentation/filesystems/vfs.txt b/Documentation/filesystems/vfs.txt
index a1d0d7a30165..61d65cc65c54 100644
--- a/Documentation/filesystems/vfs.txt
+++ b/Documentation/filesystems/vfs.txt
@@ -1053,7 +1053,8 @@ struct dentry_operations {
If the 'rcu_walk' parameter is true, then the caller is doing a
pathwalk in RCU-walk mode. Sleeping is not permitted in this mode,
and the caller can be asked to leave it and call again by returning
+ -ECHILD. -EISDIR may also be returned to tell pathwalk to
+ ignore d_automount or any mounts.

This function is only used if DCACHE_MANAGE_TRANSIT is set on the
dentry being transited from.
diff --git a/fs/namei.c b/fs/namei.c
index 985c6f368485..0abfea5697e0 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -1091,10 +1091,10 @@ int follow_down_one(struct path *path)

-static inline bool managed_dentry_might_block(struct dentry *dentry)
+static inline int managed_dentry_rcu(struct dentry *dentry)
- return (dentry->d_flags & DCACHE_MANAGE_TRANSIT &&
- dentry->d_op->d_manage(dentry, true) < 0);
+ return (dentry->d_flags & DCACHE_MANAGE_TRANSIT) ?
+ dentry->d_op->d_manage(dentry, true) : 0;

@@ -1110,11 +1110,16 @@ static bool __follow_mount_rcu(struct nameidata *nd, struct path *path,
* Don't forget we might have a non-mountpoint managed dentry
* that wants to block transit.
- if (unlikely(managed_dentry_might_block(path->dentry)))
+ switch (managed_dentry_rcu(path->dentry)) {
+ case -ECHILD:
+ default:
return false;
+ case -EISDIR:
+ return true;
+ }

if (!d_mountpoint(path->dentry))
- return true;
+ return !(path->dentry->d_flags & DCACHE_NEED_AUTOMOUNT);

mounted = __lookup_mnt(path->mnt, path->dentry);
if (!mounted)
@@ -1130,7 +1135,8 @@ static bool __follow_mount_rcu(struct nameidata *nd, struct path *path,
*inode = path->dentry->d_inode;
- return read_seqretry(&mount_lock, nd->m_seq);
+ return read_seqretry(&mount_lock, nd->m_seq) &&
+ !(path->dentry->d_flags & DCACHE_NEED_AUTOMOUNT);

static int follow_dotdot_rcu(struct nameidata *nd)
@@ -1402,11 +1408,8 @@ static int lookup_fast(struct nameidata *nd,
path->mnt = mnt;
path->dentry = dentry;
- if (unlikely(!__follow_mount_rcu(nd, path, inode)))
- goto unlazy;
- if (unlikely(path->dentry->d_flags & DCACHE_NEED_AUTOMOUNT))
- goto unlazy;
- return 0;
+ if (likely(__follow_mount_rcu(nd, path, inode)))
+ return 0;
if (unlazy_walk(nd, dentry))
return -ECHILD;

Attachment: signature.asc
Description: PGP signature