Commit graph

436 commits

Author SHA1 Message Date
Al Viro
a3fbbde70a VFS: we need to set LOOKUP_JUMPED on mountpoint crossing
Mountpoint crossing is similar to following procfs symlinks - we do
not get ->d_revalidate() called for dentry we have arrived at, with
unpleasant consequences for NFS4.

Simple way to reproduce the problem in mainline:

    cat >/tmp/a.c <<'EOF'
    #include <unistd.h>
    #include <fcntl.h>
    #include <stdio.h>
    main()
    {
            struct flock fl = {.l_type = F_RDLCK, .l_whence = SEEK_SET, .l_len = 1};
            if (fcntl(0, F_SETLK, &fl))
                    perror("setlk");
    }
    EOF
    cc /tmp/a.c -o /tmp/test

then on nfs4:

    mount --bind file1 file2
    /tmp/test < file1		# ok
    /tmp/test < file2		# spews "setlk: No locks available"...

What happens is the missing call of ->d_revalidate() after mountpoint
crossing and that's where NFS4 would issue OPEN request to server.

The fix is simple - treat mountpoint crossing the same way we deal with
following procfs-style symlinks.  I.e.  set LOOKUP_JUMPED...

Cc: stable@kernel.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-11-07 14:58:06 -08:00
Andy Whitcroft
1fa1e7f615 readlinkat: ensure we return ENOENT for the empty pathname for normal lookups
Since the commit below which added O_PATH support to the *at() calls, the
error return for readlink/readlinkat for the empty pathname has switched
from ENOENT to EINVAL:

  commit 65cfc67223
  Author: Al Viro <viro@zeniv.linux.org.uk>
  Date:   Sun Mar 13 15:56:26 2011 -0400

    readlinkat(), fchownat() and fstatat() with empty relative pathnames

This is both unexpected for userspace and makes readlink/readlinkat
inconsistant with all other interfaces; and inconsistant with our stated
return for these pathnames.

As the readlinkat call does not have a flags parameter we cannot use the
AT_EMPTY_PATH approach used in the other calls.  Therefore expose whether
the original path is infact entry via a new user_path_at_empty() path
lookup function.  Use this to determine whether to default to EINVAL or
ENOENT for failures.

Addresses http://bugs.launchpad.net/bugs/817187

[akpm@linux-foundation.org: remove unused getname_flags()]
Signed-off-by: Andy Whitcroft <apw@canonical.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: <stable@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>
2011-11-02 12:53:42 +01:00
J. Bruce Fields
f3c7691e8d leases: fix write-open/read-lease race
In setlease, we use i_writecount to decide whether we can give out a
read lease.

In open, we break leases before incrementing i_writecount.

There is therefore a window between the break lease and the i_writecount
increment when setlease could add a new read lease.

This would leave us with a simultaneous write open and read lease, which
shouldn't happen.

Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
2011-10-28 14:59:00 +02:00
Andreas Gruenbacher
948409c74d vfs: add a comment to inode_permission()
Acked-by: J. Bruce Fields <bfields@redhat.com>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruen@kernel.org>
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
2011-10-28 14:58:55 +02:00
Andreas Gruenbacher
d124b60a83 vfs: pass all mask flags check_acl and posix_acl_permission
Acked-by: J. Bruce Fields <bfields@redhat.com>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruen@kernel.org>
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
2011-10-28 14:58:54 +02:00
Andreas Gruenbacher
8fd90c8d1d vfs: indicate that the permission functions take all the MAY_* flags
Acked-by: J. Bruce Fields <bfields@redhat.com>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruen@kernel.org>
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
2011-10-28 14:58:54 +02:00
Linus Torvalds
b6c8069d35 vfs: remove LOOKUP_NO_AUTOMOUNT flag
That flag no longer makes sense, since we don't look up automount points
as eagerly any more.  Additionally, it turns out that the NO_AUTOMOUNT
handling was buggy to begin with: it would avoid automounting even for
cases where we really *needed* to do the automount handling, and could
return ENOENT for autofs entries that hadn't been instantiated yet.

With our new non-eager automount semantics, one discussion has been
about adding a AT_AUTOMOUNT flag to vfs_fstatat (and thus the
newfstatat() and fstatat64() system calls), but it's probably not worth
it: you can always force at least directory automounting by simply
adding the final '/' to the filename, which works for *all* of the stat
family system calls, old and new.

So AT_NO_AUTOMOUNT (and thus LOOKUP_NO_AUTOMOUNT) really were just a
result of our bad default behavior.

Acked-by: Ian Kent <raven@themaw.net>
Acked-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-09-27 08:12:33 -07:00
Linus Torvalds
d94c177bee vfs pathname lookup: Add LOOKUP_AUTOMOUNT flag
Since we've now turned around and made LOOKUP_FOLLOW *not* force an
automount, we want to add the ability to force an automount event on
lookup even if we don't happen to have one of the other flags that force
it implicitly (LOOKUP_OPEN, LOOKUP_DIRECTORY, LOOKUP_PARENT..)

Most cases will never want to use this, since you'd normally want to
delay automounting as long as possible, which usually implies
LOOKUP_OPEN (when we open a file or directory, we really cannot avoid
the automount any more).

But Trond argued sufficiently forcefully that at a minimum bind mounting
a file and quotactl will want to force the automount lookup.  Some other
cases (like nfs_follow_remote_path()) could use it too, although
LOOKUP_DIRECTORY would work there as well.

This commit just adds the flag and logic, no users yet, though.  It also
doesn't actually touch the LOOKUP_NO_AUTOMOUNT flag that is related, and
was made irrelevant by the same change that made us not follow on
LOOKUP_FOLLOW.

Cc: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: Ian Kent <raven@themaw.net>
Cc: Jeff Layton <jlayton@redhat.com>
Cc: Miklos Szeredi <miklos@szeredi.hu>
Cc: David Howells <dhowells@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Greg KH <gregkh@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-09-26 17:44:55 -07:00
Al Viro
1d2ef59014 restore pinning the victim dentry in vfs_rmdir()/vfs_rename_dir()
We used to get the victim pinned by dentry_unhash() prior to commit
64252c75a2 ("vfs: remove dget() from dentry_unhash()") and ->rmdir()
and ->rename() instances relied on that; most of them don't care, but
ones that used d_delete() themselves do.  As the result, we are getting
rmdir() oopses on NFS now.

Just grab the reference before locking the victim and drop it explicitly
after unlocking, same as vfs_rename_other() does.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Tested-by: Simon Kirby <sim@hostway.ca>
Cc: stable@kernel.org (3.0.x)
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-09-14 11:31:55 -07:00
Miklos Szeredi
0ec26fd069 vfs: automount should ignore LOOKUP_FOLLOW
Prior to 2.6.38 automount would not trigger on either stat(2) or
lstat(2) on the automount point.

After 2.6.38, with the introduction of the ->d_automount()
infrastructure, stat(2) and others would start triggering automount
while lstat(2), etc. still would not.  This is a regression and a
userspace ABI change.

Problem originally reported here:

  http://thread.gmane.org/gmane.linux.kernel.autofs/6098

It appears that there was an attempt at fixing various userspace tools
to not trigger the automount.  But since the stat system call is
rather common it is impossible to "fix" all userspace.

This patch reverts the original behavior, which is to not trigger on
stat(2) and other symlink following syscalls.

[ It's not really clear what the right behavior is.  Apparently Solaris
  does the "automount on stat, leave alone on lstat".  And some programs
  can get unhappy when "stat+open+fstat" ends up giving a different
  result from the fstat than from the initial stat.

  But the change in 2.6.38 resulted in problems for some people, so
  we're going back to old behavior.  Maybe we can re-visit this
  discussion at some future date  - Linus ]

Reported-by: Leonardo Chiquitto <leonardo.lists@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
Acked-by: Ian Kent <raven@themaw.net>
Cc: David Howells <dhowells@redhat.com>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-09-09 15:42:34 -07:00
Linus Torvalds
7813b94a54 vfs: rename 'do_follow_link' to 'should_follow_link'
Al points out that the do_follow_link() helper function really is
misnamed - it's about whether we should try to follow a symlink or not,
not about actually doing the following.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-08-07 13:42:25 -07:00
Ari Savolainen
206b1d09a5 Fix POSIX ACL permission check
After commit 3567866bf2: "RCUify freeing acls, let check_acl() go ahead in
RCU mode if acl is cached" posix_acl_permission is being called with an
unsupported flag and the permission check fails. This patch fixes the issue.

Signed-off-by: Ari Savolainen <ari.m.savolainen@gmail.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-08-07 04:52:23 -04:00
Linus Torvalds
3ddcd0569c vfs: optimize inode cache access patterns
The inode structure layout is largely random, and some of the vfs paths
really do care.  The path lookup in particular is already quite D$
intensive, and profiles show that accessing the 'inode->i_op->xyz'
fields is quite costly.

We already optimized the dcache to not unnecessarily load the d_op
structure for members that are often NULL using the DCACHE_OP_xyz bits
in dentry->d_flags, and this does something very similar for the inode
ops that are used during pathname lookup.

It also re-orders the fields so that the fields accessed by 'stat' are
together at the beginning of the inode structure, and roughly in the
order accessed.

The effect of this seems to be in the 1-2% range for an empty kernel
"make -j" run (which is fairly kernel-intensive, mostly in filename
lookup), so it's visible.  The numbers are fairly noisy, though, and
likely depend a lot on exact microarchitecture.  So there's more tuning
to be done.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-08-06 22:53:23 -07:00
Al Viro
3567866bf2 RCUify freeing acls, let check_acl() go ahead in RCU mode if acl is cached
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-08-03 00:58:42 -04:00
David Howells
5a30d8a2b8 VFS: Fix automount for negative autofs dentries
Autofs may set the DCACHE_NEED_AUTOMOUNT flag on negative dentries.  These
need attention from the automounter daemon regardless of the LOOKUP_FOLLOW flag.

Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Ian Kent <raven@themaw.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-08-01 01:38:01 -04:00
Linus Torvalds
84635d68be vfs: fix check_acl compile error when CONFIG_FS_POSIX_ACL is not set
Commit e77819e57f ("vfs: move ACL cache lookup into generic code")
didn't take the FS_POSIX_ACL config variable into account - when that is
not set, ACL's go away, and the cache helper functions do not exist,
causing compile errors like

  fs/namei.c: In function 'check_acl':
  fs/namei.c:191:10: error: implicit declaration of function 'negative_cached_acl'
  fs/namei.c:196:2: error: implicit declaration of function 'get_cached_acl'
  fs/namei.c:196:6: warning: assignment makes pointer from integer without a cast
  fs/namei.c:212:11: error: implicit declaration of function 'set_cached_acl'

Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de>
Acked-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-07-25 22:47:03 -07:00
Linus Torvalds
14067ff536 vfs: make gcc generate more obvious code for acl permission checking
The "fsuid is the inode owner" case is not necessarily always the likely
case, but it's the case that doesn't do anything odd and that we want in
straight-line code.  Make gcc not generate random "jump around for the
fun of it" code.

This just helps me read profiles.  That thing is one of the hottest
parts of the whole pathname lookup.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-07-25 19:55:52 -07:00
Christoph Hellwig
4e34e719e4 fs: take the ACL checks to common code
Replace the ->check_acl method with a ->get_acl method that simply reads an
ACL from disk after having a cache miss.  This means we can replace the ACL
checking boilerplate code with a single implementation in namei.c.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-25 14:30:23 -04:00
Linus Torvalds
e77819e57f vfs: move ACL cache lookup into generic code
This moves logic for checking the cached ACL values from low-level
filesystems into generic code.  The end result is a streamlined ACL
check that doesn't need to load the inode->i_op->check_acl pointer at
all for the common cached case.

The filesystems also don't need to check for a non-blocking RCU walk
case in their acl_check() functions, because that is all handled at a
VFS layer.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-25 14:23:39 -04:00
Tobias Klauser
8c5dc70aae VFS: Fixup kerneldoc for generic_permission()
The flags parameter went away in
d749519b444db985e40b897f73ce1898b11f997e

Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 20:47:43 -04:00
Al Viro
e3c3d9c838 unexport kern_path_parent()
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:44:16 -04:00
Al Viro
e0a0124936 switch vfs_path_lookup() to struct path
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:44:14 -04:00
Al Viro
ed75e95de5 kill lookup_create()
folded into the only caller (kern_path_create())

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:44:12 -04:00
Al Viro
dae6ad8f37 new helpers: kern_path_create/user_path_create
combination of kern_path_parent() and lookup_create().  Does *not*
expose struct nameidata to caller.  Syscalls converted to that...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:44:05 -04:00
Al Viro
49084c3bb2 kill LOOKUP_CONTINUE
LOOKUP_PARENT is equivalent to it now

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:44:03 -04:00
Al Viro
8a5e929dd2 don't transliterate lower bits of ->intent.open.flags to FMODE_...
->create() instances are much happier that way...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:52 -04:00
Al Viro
554a8b9f54 Don't pass nameidata when calling vfs_create() from mknod()
All instances can cope with that now (and ceph one actually
starts working properly).

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:49 -04:00
Al Viro
d2d9e9fbc2 merge do_revalidate() into its only caller
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:34 -04:00
Al Viro
4ad5abb3d0 no reason to keep exec_permission() separate now
cache footprint alone makes it a bad idea...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:32 -04:00
Al Viro
d594e7ec4d massage generic_permission() to treat directories on a separate path
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:30 -04:00
Al Viro
eecdd358b4 ->permission() sanitizing: don't pass flags to exec_permission()
pass mask instead; kill security_inode_exec_permission() since we can use
security_inode_permission() instead.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:29 -04:00
Al Viro
10556cb21a ->permission() sanitizing: don't pass flags to ->permission()
not used by the instances anymore.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:24 -04:00
Al Viro
2830ba7f34 ->permission() sanitizing: don't pass flags to generic_permission()
redundant; all callers get it duplicated in mask & MAY_NOT_BLOCK and none of
them removes that bit.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:22 -04:00
Al Viro
7e40145eb1 ->permission() sanitizing: don't pass flags to ->check_acl()
not used in the instances anymore.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:21 -04:00
Al Viro
9c2c703929 ->permission() sanitizing: pass MAY_NOT_BLOCK to ->check_acl()
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:19 -04:00
Al Viro
1fc0f78ca9 ->permission() sanitizing: MAY_NOT_BLOCK
Duplicate the flags argument into mask bitmap.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:18 -04:00
Al Viro
178ea73521 kill check_acl callback of generic_permission()
its value depends only on inode and does not change; we might as
well store it in ->i_op->check_acl and be done with that.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:16 -04:00
Al Viro
07b8ce1ee8 lockless get_write_access/deny_write_access
new helpers: atomic_inc_unless_negative()/atomic_dec_unless_positive()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:14 -04:00
Al Viro
f4d6ff89d8 move exec_permission() up to the rest of permission-related functions
... and convert the comment before it into linuxdoc form.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:13 -04:00
Al Viro
3bfa784a65 kill file_permission() completely
convert the last remaining caller to inode_permission()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:11 -04:00
Al Viro
78f32a9b47 switch path_init() to exec_permission()
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:08 -04:00
Al Viro
4cf27141cb make exec_permission(dir) really equivalent to inode_permission(dir, MAY_EXEC)
capability overrides apply only to the default case; if fs has ->permission()
that does _not_ call generic_permission(), we have no business doing them.
Moreover, if it has ->permission() that does call generic_permission(), we
have no need to recheck capabilities.

Besides, the capability overrides should apply only if we got EACCES from
acl_permission_check(); any other value (-EIO, etc.) should be returned
to caller, capabilities or not capabilities.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:05 -04:00
Josef Bacik
44396f4b5c fs: add a DCACHE_NEED_LOOKUP flag for d_flags
Btrfs (and I'd venture most other fs's) stores its indexes in nice disk order
for readdir, but unfortunately in the case of anything that stats the files in
order that readdir spits back (like oh say ls) that means we still have to do
the normal lookup of the file, which means looking up our other index and then
looking up the inode.  What I want is a way to create dummy dentries when we
find them in readdir so that when ls or anything else subsequently does a
stat(), we already have the location information in the dentry and can go
straight to the inode itself.  The lookup stuff just assumes that if it finds a
dentry it is done, it doesn't perform a lookup.  So add a DCACHE_NEED_LOOKUP
flag so that the lookup code knows it still needs to run i_op->lookup() on the
parent to get the inode for the dentry.  I have tested this with btrfs and I
went from something that looks like this

http://people.redhat.com/jwhiter/ls-noreada.png

To this

http://people.redhat.com/jwhiter/ls-good.png

Thats a savings of 1300 seconds, or 22 minutes.  That is a significant savings.
Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-20 01:43:03 -04:00
Linus Torvalds
5943026240 vfs: fix race in rcu lookup of pruned dentry
Don't update *inode in __follow_mount_rcu() until we'd verified that
there is mountpoint there.  Kudos to Hugh Dickins for catching that
one in the first place and eventually figuring out the solution (and
catching a braino in the earlier version of patch).

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-19 21:49:01 -07:00
Al Viro
94c0d4ecbe Fix ->d_lock locking order in unlazy_walk()
Make sure that child is still a child of parent before nested locking
of child->d_lock in unlazy_walk(); otherwise we are risking a violation
of locking order and deadlocks.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-12 21:40:23 -04:00
Al Viro
8e833fd2e1 fix comment in generic_permission()
CAP_DAC_OVERRIDE is enough for MAY_EXEC on directory, even if
no exec bits are set.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-06-20 10:45:56 -04:00
Al Viro
6291176bcd kill obsolete comment for follow_down()
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-06-20 10:45:49 -04:00
Al Viro
8aef188452 VFS: Fix vfsmount overput on simultaneous automount
[Kudos to dhowells for tracking that crap down]

If two processes attempt to cause automounting on the same mountpoint at the
same time, the vfsmount holding the mountpoint will be left with one too few
references on it, causing a BUG when the kernel tries to clean up.

The problem is that lock_mount() drops the caller's reference to the
mountpoint's vfsmount in the case where it finds something already mounted on
the mountpoint as it transits to the mounted filesystem and replaces path->mnt
with the new mountpoint vfsmount.

During a pathwalk, however, we don't take a reference on the vfsmount if it is
the same as the one in the nameidata struct, but do_add_mount() doesn't know
this.

The fix is to make sure we have a ref on the vfsmount of the mountpoint before
calling do_add_mount().  However, if lock_mount() doesn't transit, we're then
left with an extra ref on the mountpoint vfsmount which needs releasing.
We can handle that in follow_managed() by not making assumptions about what
we can and what we cannot get from lookup_mnt() as the current code does.

The callers of follow_managed() expect that reference to path->mnt will be
grabbed iff path->mnt has been changed.  follow_managed() and follow_automount()
keep track of whether such reference has been grabbed and assume that it'll
happen in those and only those cases that'll have us return with changed
path->mnt.  That assumption is almost correct - it breaks in case of
racing automounts and in even harder to hit race between following a mountpoint
and a couple of mount --move.  The thing is, we don't need to make that
assumption at all - after the end of loop in follow_manage() we can check
if path->mnt has ended up unchanged and do mntput() if needed.

The BUG can be reproduced with the following test program:

	#include <stdio.h>
	#include <sys/types.h>
	#include <sys/stat.h>
	#include <unistd.h>
	#include <sys/wait.h>
	int main(int argc, char **argv)
	{
		int pid, ws;
		struct stat buf;
		pid = fork();
		stat(argv[1], &buf);
		if (pid > 0) wait(&ws);
		return 0;
	}

and the following procedure:

 (1) Mount an NFS volume that on the server has something else mounted on a
     subdirectory.  For instance, I can mount / from my server:

	mount warthog:/ /mnt -t nfs4 -r

     On the server /data has another filesystem mounted on it, so NFS will see
     a change in FSID as it walks down the path, and will mark /mnt/data as
     being a mountpoint.  This will cause the automount code to be triggered.

     !!! Do not look inside the mounted fs at this point !!!

 (2) Run the above program on a file within the submount to generate two
     simultaneous automount requests:

	/tmp/forkstat /mnt/data/testfile

 (3) Unmount the automounted submount:

	umount /mnt/data

 (4) Unmount the original mount:

	umount /mnt

     At this point the kernel should throw a BUG with something like the
     following:

	BUG: Dentry ffff880032e3c5c0{i=2,n=} still in use (1) [unmount of nfs4 0:12]

Note that the bug appears on the root dentry of the original mount, not the
mountpoint and not the submount because sys_umount() hasn't got to its final
mntput_no_expire() yet, but this isn't so obvious from the call trace:

 [<ffffffff8117cd82>] shrink_dcache_for_umount+0x69/0x82
 [<ffffffff8116160e>] generic_shutdown_super+0x37/0x15b
 [<ffffffffa00fae56>] ? nfs_super_return_all_delegations+0x2e/0x1b1 [nfs]
 [<ffffffff811617f3>] kill_anon_super+0x1d/0x7e
 [<ffffffffa00d0be1>] nfs4_kill_super+0x60/0xb6 [nfs]
 [<ffffffff81161c17>] deactivate_locked_super+0x34/0x83
 [<ffffffff811629ff>] deactivate_super+0x6f/0x7b
 [<ffffffff81186261>] mntput_no_expire+0x18d/0x199
 [<ffffffff811862a8>] mntput+0x3b/0x44
 [<ffffffff81186d87>] release_mounts+0xa2/0xbf
 [<ffffffff811876af>] sys_umount+0x47a/0x4ba
 [<ffffffff8109e1ca>] ? trace_hardirqs_on_caller+0x1fd/0x22f
 [<ffffffff816ea86b>] system_call_fastpath+0x16/0x1b

as do_umount() is inlined.  However, you can see release_mounts() in there.

Note also that it may be necessary to have multiple CPU cores to be able to
trigger this bug.

Tested-by: Jeff Layton <jlayton@redhat.com>
Tested-by: Ian Kent <raven@themaw.net>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-06-16 11:28:16 -04:00
Török Edwin
50338b889d fix wrong iput on d_inode introduced by e6bc45d65d
Git bisection shows that commit e6bc45d65d causes
BUG_ONs under high I/O load:

kernel BUG at fs/inode.c:1368!
[ 2862.501007] Call Trace:
[ 2862.501007]  [<ffffffff811691d8>] d_kill+0xf8/0x140
[ 2862.501007]  [<ffffffff81169c19>] dput+0xc9/0x190
[ 2862.501007]  [<ffffffff8115577f>] fput+0x15f/0x210
[ 2862.501007]  [<ffffffff81152171>] filp_close+0x61/0x90
[ 2862.501007]  [<ffffffff81152251>] sys_close+0xb1/0x110
[ 2862.501007]  [<ffffffff814c14fb>] system_call_fastpath+0x16/0x1b

A reliable way to reproduce this bug is:
Login to KDE, run 'rsnapshot sync', and apt-get install openjdk-6-jdk,
and apt-get remove openjdk-6-jdk.

The buggy part of the patch is this:
	struct inode *inode = NULL;
.....
-               if (nd.last.name[nd.last.len])
-                       goto slashes;
                inode = dentry->d_inode;
-               if (inode)
-                       ihold(inode);
+               if (nd.last.name[nd.last.len] || !inode)
+                       goto slashes;
+               ihold(inode)
...
	if (inode)
		iput(inode);	/* truncate the inode here */

If nd.last.name[nd.last.len] is nonzero (and thus goto slashes branch is taken),
and dentry->d_inode is non-NULL, then this code now does an additional iput on
the inode, which is wrong.

Fix this by only setting the inode variable if nd.last.name[nd.last.len] is 0.

Reference: https://lkml.org/lkml/2011/6/15/50
Reported-by: Norbert Preining <preining@logic.at>
Reported-by: Török Edwin <edwintorok@gmail.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Török Edwin <edwintorok@gmail.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-06-16 11:27:39 -04:00
Theodore Ts'o
e6bc45d65d vfs: make unlink() and rmdir() return ENOENT in preference to EROFS
If user space attempts to remove a non-existent file or directory, and
the file system is mounted read-only, return ENOENT instead of EROFS.
Either error code is arguably valid/correct, but ENOENT is a more
specific error message.

Reported-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-06-07 08:51:14 -04:00