When I moved the RCU walk termination into unlazy_walk(), I didn't copy
quite all of it: for the successful RCU termination we properly add the
necessary reference counts to our temporary copy of the root path, but
for the failure case we need to make sure that any temporary root path
information is cleared out (since it does _not_ have the proper
reference counts from the RCU lookup).

We could clean up this mess by just always dropping the temporary root
information, but Al points out that that would mean that a single lookup
through symlinks could see multiple different root entries if it races
with another thread doing chroot.  Not that I think we should really
care (we had that before too, back before we had a copy of the root path
in the nameidata).

Al says he has a cunning plan.  In the meantime, this is the minimal fix
for the problem, even if it's not all that pretty.
Reported-by: NMace Moneta <moneta.mace@gmail.com>
Acked-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This is the fix that the last two commits indirectly led up to - making
sure that we don't call dput() in a bad context on the dentries we've
looked up in RCU mode after the sequence count validation fails.

This basically expands d_rcu_to_refcount() into the callers, and then
fixes the callers to delay the dput() in the failure case until _after_
we've dropped all locks and are no longer in an RCU-locked region.

The case of 'complete_walk()' was trivial, since its failure case did
the unlock_rcu_walk() directly after the call to d_rcu_to_refcount(),
and as such that is just a pure expansion of the function with a trivial
movement of the resulting dput() to after 'unlock_rcu_walk()'.

In contrast, the unlazy_walk() case was much more complicated, because
not only does convert two different dentries from RCU to be reference
counted, but it used to not call unlock_rcu_walk() at all, and instead
just returned an error and let the caller clean everything up in
"terminate_walk()".

Happily, one of the dentries in question (called "parent" inside
unlazy_walk()) is the dentry of "nd->path", which terminate_walk() wants
a refcount to anyway for the non-RCU case.

So what the new and improved unlazy_walk() does is to first turn that
dentry into a refcounted one, and once that is set up, the error cases
can continue to use the terminate_walk() helper for cleanup, but for the
non-RCU case.  Which makes it possible to drop out of RCU mode if we
actually hit the sequence number failure case.
Acked-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This simplifies the RCU to refcounting code in particular.

I was originally intending to leave this for later, but walking through
all the dput() logic (see previous commit), I realized that the dput()
"might_sleep()" check was misleadingly weak.  And I removed it as
misleading, both for performance profiling and for debugging.

However, the might_sleep() debugging case is actually true: the final
dput() can indeed sleep, if the inode of the dentry that you are
releasing ends up sleeping at iput time (see dentry_iput()).  So the
problem with the might_sleep() in dput() wasn't that it wasn't true, it
was that it wasn't actually testing and triggering on the interesting
case.

In particular, just about *any* dput() can indeed sleep, if you happen
to race with another thread deleting the file in question, and you then
lose the race to the be the last dput() for that file.  But because it's
a very rare race, the debugging code would never trigger it in practice.

Why is this problematic? The new d_rcu_to_refcount() (see commit
15570086: "vfs: reimplement d_rcu_to_refcount() using
lockref_get_or_lock()") does a dput() for the failure case, and it does
it under the RCU lock.  So potentially sleeping really is a bug.

But there's no way I'm going to fix this with the previous complicated
"lockref_get_or_lock()" interface.  And rather than revert to the old
and crufty nested dentry locking code (which did get this right by
delaying the reference count updates until they were verified to be
safe), let's make forward progress.

Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Christopher reported a regression where he was unable to unmount a NFS
filesystem where the root had gone stale. The problem is that
d_revalidate handles the root of the filesystem differently from other
dentries, but d_weak_revalidate does not. We could simply fix this by
making d_weak_revalidate return success on IS_ROOT dentries, but there
are cases where we do want to revalidate the root of the fs.

A umount is really a special case. We generally aren't interested in
anything but the dentry and vfsmount that's attached at that point. If
the inode turns out to be stale we just don't care since the intent is
to stop using it anyway.

Try to handle this situation better by treating umount as a special
case in the lookup code. Have it resolve the parent using normal
means, and then do a lookup of the final dentry without revalidating
it. In most cases, the final lookup will come out of the dcache, but
the case where there's a trailing symlink or !LAST_NORM entry on the
end complicates things a bit.

Cc: Neil Brown <neilb@suse.de>
Reported-by: NChristopher T Vogan <cvogan@us.ibm.com>
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

This moves __d_rcu_to_refcount() from <linux/dcache.h> into fs/namei.c
and re-implements it using the lockref infrastructure instead.  It also
adds a lot of comments about what is actually going on, because turning
a dentry that was looked up using RCU into a long-lived reference
counted entry is one of the more subtle parts of the rcu walk.

We also used to be _particularly_ subtle in unlazy_walk() where we
re-validate both the dentry and its parent using the same sequence
count.  We used to do it by nesting the locks and then verifying the
sequence count just once.

That was silly, because nested locking is expensive, but the sequence
count check is not.  So this just re-validates the dentry and the parent
separately, avoiding the nested locking, and making the lockref lookup
possible.
Acked-by: NWaiman Long <waiman.long@hp.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This just replaces the dentry count/lock combination with the lockref
structure that contains both a count and a spinlock, and does the
mechanical conversion to use the lockref infrastructure.

There are no semantic changes here, it's purely syntactic.  The
reference lockref implementation uses the spinlock exactly the same way
that the old dcache code did, and the bulk of this patch is just
expanding the internal "d_count" use in the dcache code to use
"d_lockref.count" instead.

This is purely preparation for the real change to make the reference
count updates be lockless during the 3.12 merge window.

[ As with the previous commit, this is a rewritten version of a concept
  originally from Waiman, so credit goes to him, blame for any errors
  goes to me.

  Waiman's patch had some semantic differences for taking advantage of
  the lockless update in dget_parent(), while this patch is
  intentionally a pure search-and-replace change with no semantic
  changes.     - Linus ]
Signed-off-by: NWaiman Long <Waiman.Long@hp.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This reverts commit bb2314b4.

It wasn't necessarily wrong per se, but we're still busily discussing
the exact details of this all, so I'm going to revert it for now.

It's true that you can already do flink() through /proc and that flink()
isn't new.  But as Brad Spengler points out, some secure environments do
not mount proc, and flink adds a new interface that can avoid path
lookup of the source for those kinds of environments.

We may re-do this (and even mark it for stable backporting back in 3.11
and possibly earlier) once the whole discussion about the interface is done.

Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Brad Spengler <spender@grsecurity.net>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Every now and then someone proposes a new flink syscall, and this spawns
a long discussion of whether it would be a security problem.  I think
that this is missing the point: flink is *already* allowed without
privilege as long as /proc is mounted -- it's called AT_SYMLINK_FOLLOW.

Now that O_TMPFILE is here, the ability to create a file with O_TMPFILE,
write it, and link it in is very convenient.  The only problem is that
it requires that /proc be mounted so that you can do:

linkat(AT_FDCWD, "/proc/self/fd/<tmpfd>", dfd, path, AT_SYMLINK_NOFOLLOW)

This sucks -- it's much nicer to do:

linkat(tmpfd, "", dfd, path, AT_EMPTY_PATH)

Let's allow it.

If this turns out to be excessively scary, it we could instead require
that the inode in question be I_LINKABLE, but this seems pointless given
the /proc situation
Signed-off-by: NAndy Lutomirski <luto@amacapital.net>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

[suggested by Rasmus Villemoes] make O_DIRECTORY | O_RDWR part of O_TMPFILE;
that will fail on old kernels in a lot more cases than what I came up with.
And make sure O_CREAT doesn't get there...
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Instances either don't look at it at all (the majority of cases) or
only want it to find the superblock (which can be had as dentry->d_sb).
A few cases that want more are actually safe with dentry->d_inode -
the only precaution needed is the check that it hadn't been replaced with
NULL by rmdir() or by overwriting rename(), which case should be simply
treated as cache miss.
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

O_TMPFILE | O_CREAT => linkat() with AT_SYMLINK_FOLLOW and /proc/self/fd/<n>
as oldpath (i.e. flink()) will create a link
O_TMPFILE | O_CREAT | O_EXCL => ENOENT on attempt to link those guys
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

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

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

/proc/self/cwd with O_CREAT should fail with EISDIR.  /proc/self/exe, OTOH,
should fail with ENOTDIR when opened with O_DIRECTORY.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

a couple of places got missed back when Linus has introduced that one...
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Jiri reported a regression in auditing of open(..., O_CREAT) syscalls.
In older kernels, creating a file with open(..., O_CREAT) created
audit_name records that looked like this:

type=PATH msg=audit(1360255720.628:64): item=1 name="/abc/foo" inode=138810 dev=fd:00 mode=0100640 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
type=PATH msg=audit(1360255720.628:64): item=0 name="/abc/" inode=138635 dev=fd:00 mode=040750 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0

...in recent kernels though, they look like this:

type=PATH msg=audit(1360255402.886:12574): item=2 name=(null) inode=264599 dev=fd:00 mode=0100640 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
type=PATH msg=audit(1360255402.886:12574): item=1 name=(null) inode=264598 dev=fd:00 mode=040750 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
type=PATH msg=audit(1360255402.886:12574): item=0 name="/abc/foo" inode=264598 dev=fd:00 mode=040750 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0

Richard bisected to determine that the problems started with commit
bfcec708, but the log messages have changed with some later
audit-related patches.

The problem is that this audit_inode call is passing in the parent of
the dentry being opened, but audit_inode is being called with the parent
flag false. This causes later audit_inode and audit_inode_child calls to
match the wrong entry in the audit_names list.

This patch simply sets the flag to properly indicate that this inode
represents the parent. With this, the audit_names entries are back to
looking like they did before.

Cc: <stable@vger.kernel.org> # v3.7+
Reported-by: NJiri Jaburek <jjaburek@redhat.com>
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Test By: Richard Guy Briggs <rbriggs@redhat.com>
Signed-off-by: NEric Paris <eparis@redhat.com>

It's "normal" - it can happen if the file descriptor you followed was
opened with O_NOFOLLOW.
Reported-by: NDave Jones <davej@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: stable@kernel.org
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

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

The following set of operations on a NFS client and server will cause

    server# mkdir a
    client# cd a
    server# mv a a.bak
    client# sleep 30  # (or whatever the dir attrcache timeout is)
    client# stat .
    stat: cannot stat `.': Stale NFS file handle

Obviously, we should not be getting an ESTALE error back there since the
inode still exists on the server. The problem is that the lookup code
will call d_revalidate on the dentry that "." refers to, because NFS has
FS_REVAL_DOT set.

nfs_lookup_revalidate will see that the parent directory has changed and
will try to reverify the dentry by redoing a LOOKUP. That of course
fails, so the lookup code returns ESTALE.

The problem here is that d_revalidate is really a bad fit for this case.
What we really want to know at this point is whether the inode is still
good or not, but we don't really care what name it goes by or whether
the dcache is still valid.

Add a new d_op->d_weak_revalidate operation and have complete_walk call
that instead of d_revalidate. The intent there is to allow for a
"weaker" d_revalidate that just checks to see whether the inode is still
good. This is also gives us an opportunity to kill off the FS_REVAL_DOT
special casing.

[AV: changed method name, added note in porting, fixed confusion re
having it possibly called from RCU mode (it won't be)]

Cc: NeilBrown <neilb@suse.de>
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

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

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

... always can be found in nameidata now.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

... and clean the main loop a bit
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Based on parts from Anatol's patch (the rest is the next commit).
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

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

...as always, rename is the messiest of the bunch. We have to track
whether to retry or not via a separate flag since the error handling
is already quite complex.
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

...so we can pass in LOOKUP_REVAL. For now, nothing does yet.
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Where we can pass in LOOKUP_DIRECTORY or LOOKUP_REVAL. Any other flags
passed in here are currently ignored.
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

The code that relied on that flag was ripped out of btrfs quite some
time ago, and never added back. Josef indicated that he was going to
take a different approach to the problem in btrfs, and that we
could just eliminate this flag.

Cc: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

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

When path_init is called with a valid dfd, that code checks permissions
on the open directory fd and returns an error if the check fails. This
permission check is redundant, however.

Both callers of path_init immediately call link_path_walk afterward. The
first thing that link_path_walk does for pathnames that do not consist
only of slashes is to check for exec permissions at the starting point of
the path walk.  And this check in path_init() is on the path taken only
when *name != '/' && *name != '\0'.

In most cases, these checks are very quick, but when the dfd is for a
file on a NFS mount with the actimeo=0, each permission check goes
out onto the wire. The result is 2 identical ACCESS calls.

Given that these codepaths are fairly "hot", I think it makes sense to
eliminate the permission check in path_init and simply assume that the
caller will eventually check the permissions before proceeding.
Reported-by: NDave Wysochanski <dwysocha@redhat.com>
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

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

In commit 800179c9 ("This adds symlink and hardlink restrictions to
the Linux VFS"), the new link protections were enabled by default, in
the hope that no actual application would care, despite it being
technically against legacy UNIX (and documented POSIX) behavior.

However, it does turn out to break some applications.  It's rare, and
it's unfortunate, but it's unacceptable to break existing systems, so
we'll have to default to legacy behavior.

In particular, it has broken the way AFD distributes files, see

  http://www.dwd.de/AFD/

along with some legacy scripts.

Distributions can end up setting this at initrd time or in system
scripts: if you have security problems due to link attacks during your
early boot sequence, you have bigger problems than some kernel sysctl
setting. Do:

	echo 1 > /proc/sys/fs/protected_symlinks
	echo 1 > /proc/sys/fs/protected_hardlinks

to re-enable the link protections.

Alternatively, we may at some point introduce a kernel config option
that sets these kinds of "more secure but not traditional" behavioural
options automatically.
Reported-by: NNick Bowler <nbowler@elliptictech.com>
Reported-by: NHolger Kiehl <Holger.Kiehl@dwd.de>
Cc: Kees Cook <keescook@chromium.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org # v3.6
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

In the common case where a name is much smaller than PATH_MAX, an extra
allocation for struct filename is unnecessary. Before allocating a
separate one, try to embed the struct filename inside the buffer first. If
it turns out that that's not long enough, then fall back to allocating a
separate struct filename and redoing the copy.
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>