We need to get the underlying dentry of parent; sure, absent the races
it is the parent of underlying dentry, but there's nothing to prevent
losing a timeslice to preemtion in the middle of evaluation of
lower_dentry->d_parent->d_inode, having another process move lower_dentry
around and have its (ex)parent not pinned anymore and freed on memory
pressure.  Then we regain CPU and try to fetch ->d_inode from memory
that is freed by that point.

dentry->d_parent *is* stable here - it's an argument of ->lookup() and
we are guaranteed that it won't be moved anywhere until we feed it
to d_add/d_splice_alias.  So we safely go that way to get to its
underlying dentry.

Cc: stable@vger.kernel.org # since 2009 or so
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

lower_dentry can't go from positive to negative (we have it pinned),
but it *can* go from negative to positive.  So fetching ->d_inode
into a local variable, doing a blocking allocation, checking that
now ->d_inode is non-NULL and feeding the value we'd fetched
earlier to a function that won't accept NULL is not a good idea.

Cc: stable@vger.kernel.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

A problem similar to the one caught in commit 74dd7c97ea2a ("ecryptfs_rename():
verify that lower dentries are still OK after lock_rename()") exists for
unlink/rmdir as well.

Instead of playing with dget_parent() of underlying dentry of victim
and hoping it's the same as underlying dentry of our directory,
do the following:
        * find the underlying dentry of victim
        * find the underlying directory of victim's parent (stable
since the victim is ecryptfs dentry and inode of its parent is
held exclusive by the caller).
        * lock the inode of dentry underlying the victim's parent
        * check that underlying dentry of victim is still hashed and
has the right parent - it can be moved, but it can't be moved to/from
the directory we are holding exclusive.  So while ->d_parent itself
might not be stable, the result of comparison is.

If the check passes, everything is fine - underlying directory is locked,
underlying victim is still a child of that directory and we can go ahead
and feed them to vfs_unlink().  As in the current mainline we need to
pin the underlying dentry of victim, so that it wouldn't go negative under
us, but that's the only temporary reference that needs to be grabbed there.
Underlying dentry of parent won't go away (it's pinned by the parent,
which is held by caller), so there's no need to grab it.

The same problem (with the same solution) exists for rmdir.  Moreover,
rename gets simpler and more robust with the same "don't bother with
dget_parent()" approach.

Fixes: 74dd7c97ea2 "ecryptfs_rename(): verify that lower dentries are still OK after lock_rename()"
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

Pull eCryptfs updates from Tyler Hicks:

 - Fix error handling when ecryptfs_read_lower() encounters an error

 - Fix read-only file creation when the eCryptfs mount is configured to
   store metadata in xattrs

 - Minor code cleanups

* tag 'ecryptfs-5.3-rc1-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tyhicks/ecryptfs:
  ecryptfs: Change return type of ecryptfs_process_flags
  ecryptfs: Make ecryptfs_xattr_handler static
  ecryptfs: remove unnessesary null check in ecryptfs_keyring_auth_tok_for_sig
  ecryptfs: use print_hex_dump_bytes for hexdump
  eCryptfs: fix permission denied with ecryptfs_xattr mount option when create readonly file
  ecryptfs: re-order a condition for static checkers
  eCryptfs: fix a couple type promotion bugs

Fix sparse warning:

fs/ecryptfs/inode.c:1138:28: warning:
 symbol 'ecryptfs_xattr_handler' was not declared. Should it be static?

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Tyler Hicks <tyhicks@canonical.com>

Based on 1 normalized pattern(s):

  this program is free software you can redistribute it and or modify
  it under the terms of the gnu general public license as published by
  the free software foundation either version 2 of the license or at
  your option any later version this program is distributed in the
  hope that it will be useful but without any warranty without even
  the implied warranty of merchantability or fitness for a particular
  purpose see the gnu general public license for more details you
  should have received a copy of the gnu general public license along
  with this program if not write to the free software foundation inc
  59 temple place suite 330 boston ma 02111 1307 usa

extracted by the scancode license scanner the SPDX license identifier

  GPL-2.0-or-later

has been chosen to replace the boilerplate/reference in 1334 file(s).

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Allison Randal <allison@lohutok.net>
Reviewed-by: Richard Fontana <rfontana@redhat.com>
Cc: linux-spdx@vger.kernel.org
Link: https://lkml.kernel.org/r/20190527070033.113240726@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

We get lower layer dentries, find their parents, do lock_rename() and
proceed to vfs_rename().  However, we do not check that dentries still
have the same parents and are not unlinked.  Need to check that...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

Pull vfs fixes from Al Viro:
 "Assorted fixes all over the place"

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  aio: fix io_destroy(2) vs. lookup_ioctx() race
  ext2: fix a block leak
  nfsd: vfs_mkdir() might succeed leaving dentry negative unhashed
  cachefiles: vfs_mkdir() might succeed leaving dentry negative unhashed
  unfuck sysfs_mount()
  kernfs: deal with kernfs_fill_super() failures
  cramfs: Fix IS_ENABLED typo
  befs_lookup(): use d_splice_alias()
  affs_lookup: switch to d_splice_alias()
  affs_lookup(): close a race with affs_remove_link()
  fix breakage caused by d_find_alias() semantics change
  fs: don't scan the inode cache before SB_BORN is set
  do d_instantiate/unlock_new_inode combinations safely
  iov_iter: fix memory leak in pipe_get_pages_alloc()
  iov_iter: fix return type of __pipe_get_pages()

For anything NFS-exported we do _not_ want to unlock new inode
before it has grown an alias; original set of fixes got the
ordering right, but missed the nasty complication in case of
lockdep being enabled - unlock_new_inode() does
	lockdep_annotate_inode_mutex_key(inode)
which can only be done before anyone gets a chance to touch
->i_mutex.  Unfortunately, flipping the order and doing
unlock_new_inode() before d_instantiate() opens a window when
mkdir can race with open-by-fhandle on a guessed fhandle, leading
to multiple aliases for a directory inode and all the breakage
that follows from that.

	Correct solution: a new primitive (d_instantiate_new())
combining these two in the right order - lockdep annotate, then
d_instantiate(), then the rest of unlock_new_inode().  All
combinations of d_instantiate() with unlock_new_inode() should
be converted to that.

Cc: stable@kernel.org	# 2.6.29 and later
Tested-by: Mike Marshall <hubcap@omnibond.com>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

mount_crypt_stat is assigned to
&ecryptfs_superblock_to_private(ecryptfs_dentry->d_sb)->mount_crypt_stat,
and mount_crypt_stat is not the first object in struct ecryptfs_sb_info.
mount_crypt_stat is therefore never NULL. At the same time, no crash
in ecryptfs_lookup() has been reported, and the lookup functions in
other file systems don't check if d_sb is NULL either.
Given that, remove the NULL check.

Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Tyler Hicks <tyhicks@canonical.com>