sysfs: fail dentry revalidation after namespace change

When we change the namespace tag of a sysfs entry, the associated dentry
is still kept around. readdir() will work correctly and not display the
old entries, but open() will still succeed, so will reads and writes.

This will no longer happen if sysfs is remounted, hinting that this is a
cache-related problem.

I am using the following sequence to demonstrate that:

shell1:
ip link add type veth
unshare -nm

shell2:
ip link set veth1 <pid_of_shell_1>
cat /sys/devices/virtual/net/veth1/ifindex

Before that patch, this will succeed (fail to fail). After it, it will
correctly return an error. Differently from a normal rename, which we
handle fine, changing the object namespace will keep it's path intact.
So this check seems necessary as well.

[ v2: get type from parent, as suggested by Eric Biederman ]

Signed-off-by: Glauber Costa <glommer@parallels.com>
CC: Tejun Heo <tj@kernel.org>
Reviewed-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

fs/sysfs/dir.c

@@ -307,6 +307,7 @@ static int sysfs_dentry_revalidate(struct dentry *dentry, struct nameidata *nd)
 {
 	struct sysfs_dirent *sd;
 	int is_dir;
+	int type;
 
 	if (nd->flags & LOOKUP_RCU)
 		return -ECHILD;
@@ -326,6 +327,13 @@ static int sysfs_dentry_revalidate(struct dentry *dentry, struct nameidata *nd)
 	if (strcmp(dentry->d_name.name, sd->s_name) != 0)
 		goto out_bad;
 
+	/* The sysfs dirent has been moved to a different namespace */
+	type = KOBJ_NS_TYPE_NONE;
+	if (sd->s_parent)
+		type = sysfs_ns_type(sd->s_parent);
+	if (type && (sysfs_info(dentry->d_sb)->ns[type] != sd->s_ns))
+		goto out_bad;
+
 	mutex_unlock(&sysfs_mutex);
 out_valid:
 	return 1;