Ban ecryptfs over ecryptfs

(cherry picked from commit 4403158ba295c8e36f6736b1bb12d0f7e1923dac)

This is a seriously simplified patch from Eric Sandeen; copy of
rationale follows:
===
  mounting stacked ecryptfs on ecryptfs has been shown to lead to bugs
  in testing.  For crypto info in xattr, there is no mechanism for handling
  this at all, and for normal file headers, we run into other trouble:

  BUG: unable to handle kernel NULL pointer dereference at 0000000000000008
  IP: [<ffffffffa015b0b3>] ecryptfs_d_revalidate+0x43/0xa0 [ecryptfs]
  ...

  There doesn't seem to be any good usecase for this, so I'd suggest just
  disallowing the configuration.

  Based on a patch originally, I believe, from Mike Halcrow.
===

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

diff --git a/fs/ecryptfs/main.c b/fs/ecryptfs/main.c
index c6ac85d6c701f5b7d82e069c9a3f82a67a51d0b3..f6ccb4f6442e5ae9d1c6aa55a9d7a7514e40e75b 100644 (file)
--- a/fs/ecryptfs/main.c
+++ b/fs/ecryptfs/main.c
@@ -487,6 +487,7 @@ out:
 }
 
 struct kmem_cache *ecryptfs_sb_info_cache;
+static struct file_system_type ecryptfs_fs_type;
 
 /**
  * ecryptfs_fill_super
@@ -561,6 +562,13 @@ static int ecryptfs_read_super(struct super_block *sb, const char *dev_name)
                ecryptfs_printk(KERN_WARNING, "path_lookup() failed\n");
                goto out;
        }
+       if (path.dentry->d_sb->s_type == &ecryptfs_fs_type) {
+               rc = -EINVAL;
+               printk(KERN_ERR "Mount on filesystem of type "
+                       "eCryptfs explicitly disallowed due to "
+                       "known incompatibilities\n");
+               goto out_free;
+       }
        ecryptfs_set_superblock_lower(sb, path.dentry->d_sb);
        sb->s_maxbytes = path.dentry->d_sb->s_maxbytes;
        sb->s_blocksize = path.dentry->d_sb->s_blocksize;