Remove the __exit mark from cifs_exit_dns_resolver() as it's called by the
module init routine in case of error, and so may have been discarded during
linkage.
Signed-off-by: NDavid Howells <dhowells@redhat.com>
Acked-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

It's possible for get_task_cred() as it currently stands to 'corrupt' a set of
credentials by incrementing their usage count after their replacement by the
task being accessed.

What happens is that get_task_cred() can race with commit_creds():

	TASK_1			TASK_2			RCU_CLEANER
	-->get_task_cred(TASK_2)
	rcu_read_lock()
	__cred = __task_cred(TASK_2)
				-->commit_creds()
				old_cred = TASK_2->real_cred
				TASK_2->real_cred = ...
				put_cred(old_cred)
				  call_rcu(old_cred)
		[__cred->usage == 0]
	get_cred(__cred)
		[__cred->usage == 1]
	rcu_read_unlock()
							-->put_cred_rcu()
							[__cred->usage == 1]
							panic()

However, since a tasks credentials are generally not changed very often, we can
reasonably make use of a loop involving reading the creds pointer and using
atomic_inc_not_zero() to attempt to increment it if it hasn't already hit zero.

If successful, we can safely return the credentials in the knowledge that, even
if the task we're accessing has released them, they haven't gone to the RCU
cleanup code.

We then change task_state() in procfs to use get_task_cred() rather than
calling get_cred() on the result of __task_cred(), as that suffers from the
same problem.

Without this change, a BUG_ON in __put_cred() or in put_cred_rcu() can be
tripped when it is noticed that the usage count is not zero as it ought to be,
for example:

kernel BUG at kernel/cred.c:168!
invalid opcode: 0000 [#1] SMP
last sysfs file: /sys/kernel/mm/ksm/run
CPU 0
Pid: 2436, comm: master Not tainted 2.6.33.3-85.fc13.x86_64 #1 0HR330/OptiPlex
745
RIP: 0010:[<ffffffff81069881>]  [<ffffffff81069881>] __put_cred+0xc/0x45
RSP: 0018:ffff88019e7e9eb8  EFLAGS: 00010202
RAX: 0000000000000001 RBX: ffff880161514480 RCX: 00000000ffffffff
RDX: 00000000ffffffff RSI: ffff880140c690c0 RDI: ffff880140c690c0
RBP: ffff88019e7e9eb8 R08: 00000000000000d0 R09: 0000000000000000
R10: 0000000000000001 R11: 0000000000000040 R12: ffff880140c690c0
R13: ffff88019e77aea0 R14: 00007fff336b0a5c R15: 0000000000000001
FS:  00007f12f50d97c0(0000) GS:ffff880007400000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f8f461bc000 CR3: 00000001b26ce000 CR4: 00000000000006f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process master (pid: 2436, threadinfo ffff88019e7e8000, task ffff88019e77aea0)
Stack:
 ffff88019e7e9ec8 ffffffff810698cd ffff88019e7e9ef8 ffffffff81069b45
<0> ffff880161514180 ffff880161514480 ffff880161514180 0000000000000000
<0> ffff88019e7e9f28 ffffffff8106aace 0000000000000001 0000000000000246
Call Trace:
 [<ffffffff810698cd>] put_cred+0x13/0x15
 [<ffffffff81069b45>] commit_creds+0x16b/0x175
 [<ffffffff8106aace>] set_current_groups+0x47/0x4e
 [<ffffffff8106ac89>] sys_setgroups+0xf6/0x105
 [<ffffffff81009b02>] system_call_fastpath+0x16/0x1b
Code: 48 8d 71 ff e8 7e 4e 15 00 85 c0 78 0b 8b 75 ec 48 89 df e8 ef 4a 15 00
48 83 c4 18 5b c9 c3 55 8b 07 8b 07 48 89 e5 85 c0 74 04 <0f> 0b eb fe 65 48 8b
04 25 00 cc 00 00 48 3b b8 58 04 00 00 75
RIP  [<ffffffff81069881>] __put_cred+0xc/0x45
 RSP <ffff88019e7e9eb8>
---[ end trace df391256a100ebdd ]---
Signed-off-by: NDavid Howells <dhowells@redhat.com>
Acked-by: NJiri Olsa <jolsa@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The function ecryptfs_uid_hash wrongly assumes that the
second parameter to hash_long() is the number of hash
buckets instead of the number of hash bits.
This patch fixes that and renames the variable
ecryptfs_hash_buckets to ecryptfs_hash_bits to make it
clearer.

Fixes: CVE-2010-2492
Signed-off-by: NAndre Osterhues <aosterhues@escrypt.com>
Signed-off-by: NTyler Hicks <tyhicks@linux.vnet.ibm.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

If we don't need a huge amount of memory in ->readdir() then
we can use kmalloc rather than vmalloc to allocate it. This
should cut down on the greater overheads associated with
vmalloc for smaller directories.

We may be able to eliminate vmalloc entirely at some stage,
but this is easy to do right away.

Also using GFP_NOFS to avoid any issues wrt to deleting inodes
while under a glock, and suggestion from Linus to factor out
the alloc/dealloc.

I've given this a test with a variety of different sized
directories and it seems to work ok.

Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: NSteven Whitehouse <swhiteho@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This fixes an issue triggered by running concurrent syncs. One of the syncs
would go through while the other would just hang indefinitely. In any case, we
never actually want to wake a single waiter, so the *_all functions should
be used.
Signed-off-by: NYehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: NSage Weil <sage@newdream.net>

Pass the correct end of the buffer to p9stat_read.
Signed-off-by: NLatchesar Ionkov <lucho@ionkov.net>
Signed-off-by: NEric Van Hensbergen <ericvh@gmail.com>

Supporting symlinks from untagged to tagged directories is reasonable,
and needed to support CONFIG_SYSFS_DEPRECATED.  So don't fail a prior
allowing that case to work.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This happens for network devices when SYSFS_DEPRECATED is enabled.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Recently my tagged sysfs support revealed a flaw in the device core
that a few rare drivers are running into such that we don't always put
network devices in a class subdirectory named net/.

Since we are not creating the class directory the network devices wind
up in a non-tagged directory, but the symlinks to the network devices
from /sys/class/net are in a tagged directory.  All of which works
until we go to remove or rename the symlink.  When we remove or rename
a symlink we look in the namespace of the target of the symlink.
Since the target of the symlink is in a non-tagged sysfs directory we
don't have a namespace to look in, and we fail to remove the symlink.

Detect this problem up front and simply don't create symlinks we won't
be able to remove later.  This prevents symlink leakage and fails in
a much clearer and more understandable way.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Rafael J. Wysocki <rjw@sisk.pl>
Cc: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Kay Sievers <kay.sievers@vrfy.org>
Cc: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Signed-off-by: NRobert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: NSage Weil <sage@newdream.net>

When we embed a dentry lease release notification in a request, invalidate
our lease so we don't think we still have it.  Otherwise we can get all
sorts of incorrect client behavior when multiple clients are interacting
with the same part of the namespace.
Signed-off-by: NSage Weil <sage@newdream.net>

If we fail to allocate a ceph_dentry_info, don't leak the dn reference.
Signed-off-by: NSage Weil <sage@newdream.net>

Free the ceph_pg_mapping structs when they are removed from the pg_temp
rbtree.  Also fix a leak in the __insert_pg_mapping() error path.
Signed-off-by: NSage Weil <sage@newdream.net>

We need to set the d_release dop for snapdir and snapped dentries so that
the ceph_dentry_info struct gets released.  We also use the dcache to
cache readdir results when possible, which only works if we know when
dentries are dropped from the cache.  Since we don't use the dcache for
readdir in the hidden snapdir, avoid that case in ceph_dentry_release.
Signed-off-by: NSage Weil <sage@newdream.net>

We should always go to the MDS for readdir on the hidden snapdir.  The
set of snapshots can change at any time; the client can't trust its cache
for that.
Signed-off-by: NSage Weil <sage@newdream.net>

Fix the security problem in the CIFS filesystem DNS lookup code in which a
malicious redirect could be installed by a random user by simply adding a
result record into one of their keyrings with add_key() and then invoking a
CIFS CFS lookup [CVE-2010-2524].

This is done by creating an internal keyring specifically for the caching of
DNS lookups.  To enforce the use of this keyring, the module init routine
creates a set of override credentials with the keyring installed as the thread
keyring and instructs request_key() to only install lookup result keys in that
keyring.

The override is then applied around the call to request_key().

This has some additional benefits when a kernel service uses this module to
request a key:

 (1) The result keys are owned by root, not the user that caused the lookup.

 (2) The result keys don't pop up in the user's keyrings.

 (3) The result keys don't come out of the quota of the user that caused the
     lookup.

The keyring can be viewed as root by doing cat /proc/keys:

2a0ca6c3 I-----     1 perm 1f030000     0     0 keyring   .dns_resolver: 1/4

It can then be listed with 'keyctl list' by root.

	# keyctl list 0x2a0ca6c3
	1 key in keyring:
	726766307: --alswrv     0     0 dns_resolver: foo.bar.com
Signed-off-by: NDavid Howells <dhowells@redhat.com>
Reviewed-and-Tested-by: NJeff Layton <jlayton@redhat.com>
Acked-by: NSteve French <smfrench@gmail.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Pointed out by Lucas who found the new one in a comment in
setup_percpu.c. And then I fixed the others that I grepped
for.
Reported-by: NLucas <canolucas@gmail.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

https://bugzilla.kernel.org/show_bug.cgi?id=16348

When the filesystem grows to a large number of allocation groups,
the summing of recalimable inodes gets expensive. In many cases,
most AGs won't have any reclaimable inodes and so we are wasting CPU
time aggregating over these AGs. This is particularly important for
the inode shrinker that gets called frequently under memory
pressure.

To avoid the overhead, track AGs with reclaimable inodes in the
per-ag radix tree so that we can find all the AGs with reclaimable
inodes via a simple gang tag lookup. This involves setting the tag
when the first reclaimable inode is tracked in the AG, and removing
the tag when the last reclaimable inode is removed from the tree.
Then the summation process becomes a loop walking the radix tree
summing AGs with the reclaim tag set.

This significantly reduces the overhead of scanning - a 6400 AG
filesystea now only uses about 25% of a cpu in kswapd while slab
reclaim progresses instead of being permanently stuck at 100% CPU
and making little progress. Clean filesystems filesystems will see
no overhead and the overhead only increases linearly with the number
of dirty AGs.
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>

Now the shrinker passes us a context, wire up a shrinker context per
filesystem. This allows us to remove the global mount list and the
locking problems that introduced. It also means that a shrinker call
does not need to traverse clean filesystems before finding a
filesystem with reclaimable inodes.  This significantly reduces
scanning overhead when lots of filesystems are present.
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>

1.  The BTRFS_IOC_CLONE and BTRFS_IOC_CLONE_RANGE ioctls should check
whether the donor file is append-only before writing to it.

2.  The BTRFS_IOC_CLONE_RANGE ioctl appears to have an integer
overflow that allows a user to specify an out-of-bounds range to copy
from the source file (if off + len wraps around).  I haven't been able
to successfully exploit this, but I'd imagine that a clever attacker
could use this to read things he shouldn't.  Even if it's not
exploitable, it couldn't hurt to be safe.
Signed-off-by: NDan Rosenberg <dan.j.rosenberg@gmail.com>
cc: stable@kernel.org
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The CLONE and CLONE_RANGE ioctls round up the range of extents being
cloned to the block size when the range to clone extends to the end of file
(this is always the case with CLONE).  It was then using that offset when
extending the destination file's i_size.  Fix this by not setting i_size
beyond the originally requested ending offset.

This bug was introduced by a22285a6 (2.6.35-rc1).
Signed-off-by: NSage Weil <sage@newdream.net>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

split_leaf was not properly balancing leaves when it was forced to
split a leaf twice.  This commit adds an extra push left and right
before forcing the double split in hopes of getting the slot where
we want to insert at either the start or end of the leaf.

If the extra pushes do work, then we are able to avoid splitting twice
and we keep the tree properly balanced.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Partition boundary calculation fails for DASD FBA disks under the
following conditions:
- disk is formatted with CMS FORMAT with a blocksize of more than
  512 bytes
- all of the disk is reserved to a single CMS file using CMS RESERVE
- the disk is accessed using the DIAG mode of the DASD driver

Under these circumstances, the partition detection code tries to
read the CMS label block containing partition-relevant information
from logical block offset 1, while it is in fact located at physical
block offset 1.

Fix this problem by using the correct CMS label block location
depending on the device type as determined by the DASD SENSE ID
information.
Signed-off-by: NPeter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The current shrinker implementation requires the registered callback
to have global state to work from. This makes it difficult to shrink
caches that are not global (e.g. per-filesystem caches). Pass the shrinker
structure to the callback so that users can embed the shrinker structure
in the context the shrinker needs to operate on and get back to it in the
callback via container_of().
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>

ocfs2_write_zero_page() has a loop that won't ever be skipped, but gcc
doesn't know that.  Set ret=0 just to make gcc happy.
Signed-off-by: NJoel Becker <joel.becker@oracle.com>

Strip the cap and dentry releases from replayed messages.  They can
cause the shared state to get out of sync because they were generated
(with the request message) earlier, and no longer reflect the current
client state.
Signed-off-by: NSage Weil <sage@newdream.net>

Replayed rename operations (after an mds failure/recovery) were broken
because the request paths were regenerated from the dentry names, which
get mangled when d_move() is called.

Instead, resend the previous request message when replaying completed
operations.  Just make sure the REPLAY flag is set and the target ino is
filled in.

This fixes problems with workloads doing renames when the MDS restarts,
where the rename operation appears to succeed, but on mds restart then
fails (leading to client confusion, app breakage, etc.).
Signed-off-by: NSage Weil <sage@newdream.net>

OCFS2 uses t_commit trigger to compute and store checksum of the just
committed blocks. When a buffer has b_frozen_data, checksum is computed
for it instead of b_data but this can result in an old checksum being
written to the filesystem in the following scenario:

1) transaction1 is opened
2) handle1 is opened
3) journal_access(handle1, bh)
    - This sets jh->b_transaction to transaction1
4) modify(bh)
5) journal_dirty(handle1, bh)
6) handle1 is closed
7) start committing transaction1, opening transaction2
8) handle2 is opened
9) journal_access(handle2, bh)
    - This copies off b_frozen_data to make it safe for transaction1 to commit.
      jh->b_next_transaction is set to transaction2.
10) jbd2_journal_write_metadata() checksums b_frozen_data
11) the journal correctly writes b_frozen_data to the disk journal
12) handle2 is closed
    - There was no dirty call for the bh on handle2, so it is never queued for
      any more journal operation
13) Checkpointing finally happens, and it just spools the bh via normal buffer
writeback.  This will write b_data, which was never triggered on and thus
contains a wrong (old) checksum.

This patch fixes the problem by calling the trigger at the moment data is
frozen for journal commit - i.e., either when b_frozen_data is created by
do_get_write_access or just before we write a buffer to the log if
b_frozen_data does not exist. We also rename the trigger to t_frozen as
that better describes when it is called.
Signed-off-by: NJan Kara <jack@suse.cz>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>
Signed-off-by: NJoel Becker <joel.becker@oracle.com>

For migration, we are waiting for DLM_LOCK_RES_MIGRATING flag to be set
before sending DLM_MIG_LOCKRES_MSG message to the target. We are using
dlm_migration_can_proceed() for that purpose.  However, if the node is
down, dlm_migration_can_proceed() will also return "go ahead".  In this
rare case, the DLM_LOCK_RES_MIGRATING flag might not be set yet. Remove
the BUG_ON() that trips over this condition.
Signed-off-by: NWengang Wang <wen.gang.wang@oracle.com>
Signed-off-by: NJoel Becker <joel.becker@oracle.com>

During CoW, the pages after i_size don't contain valid data, so there's
no need to read and duplicate them.
Signed-off-by: NTao Ma <tao.ma@oracle.com>
Signed-off-by: NJoel Becker <joel.becker@oracle.com>

This patch fixes a kernel Oops in the GFS2 rename code.

The problem was in the way the gfs2 directory code was trying
to re-use sentinel directory entries.

In the failing case, gfs2's rename function was renaming a
file to another name that had the same non-trivial length.
The file being renamed happened to be the first directory
entry on the leaf block.

First, the rename code (gfs2_rename in ops_inode.c) found the
original directory entry and decided it could do its job by
simply replacing the directory entry with another.  Therefore
it determined correctly that no block allocations were needed.

Next, the rename code deleted the old directory entry prior to
replacing it with the new name.  Therefore, the soon-to-be
replaced directory entry was temporarily made into a directory
entry "sentinel" or a place holder at the start of a leaf block.

Lastly, it went to re-add the replacement directory entry in
that leaf block.  However, when gfs2_dirent_find_space was
looking for space in the leaf block, it used the wrong value
for the sentinel.  That threw off its calculations so later
it decides it can't really re-use the sentinel and therefore
must allocate a new leaf block.  But because it previously decided
to re-use the directory entry, it didn't waste the time to
grab a new block allocation for the inode.  Therefore, the
inode's i_alloc pointer was still NULL and it crashes trying to
reference it.

In the case of sentinel directory entries, the entire dirent is
reused, not just the "free space" portion of it, and therefore
the function gfs2_dirent_find_space should use the value 0
rather than GFS2_DIRENT_SIZE(0) for the actual dirent size.

Fixing this calculation enables the reproducer programs to work
properly.
Signed-off-by: NBob Peterson <rpeterso@redhat.com>
Signed-off-by: NSteven Whitehouse <swhiteho@redhat.com>

HighMem pages on i686 do not get mapped to the buffer_heads and this was
causing a NULL pointer dereference when we were trying to memset page buffers
to zero.
We now use zero_user() that kmaps the page and directly manipulates page data.
This patch also fixes a boundary condition that was incorrect.
Signed-off-by: NAbhi Das <adas@redhat.com>
Signed-off-by: NSteven Whitehouse <swhiteho@redhat.com>

This patch fixes a problem in an error path when looking
up dinodes.  There are two sister-functions, gfs2_inode_lookup
and gfs2_process_unlinked_inode.  Both functions acquire and
hold the i_iopen glock for the dinode being looked up. The last
thing they try to do is hold the i_gl glock for the dinode.
If that glock fails for some reason, the error path was
incorrectly calling gfs2_glock_put for the i_iopen glock twice.
This resulted in the glock being prematurely freed.  The
"minimum hold time" usually kept the glock in memory, but the
lock interface to dlm (aka lock_dlm) freed its memory for the
glock.  In some circumstances, it would cause dlm's dlm_astd daemon
to try to call the bast function for the freed lock_dlm memory,
which resulted in a NULL pointer dereference.
Signed-off-by: NBob Peterson <rpeterso@redhat.com>
Signed-off-by: NSteven Whitehouse <swhiteho@redhat.com>

This patch fixes bugzilla bug #590878: GFS2: recovery stuck on
transaction lock.  We set the frozen flag on the glock when we receive
a completion that cannot be delivered due to blocked locks. At that
point we check to see whether the first waiting holder has the noexp
flag set. If the noexp lock is queued later, then we need to unfreeze
the glock at that point in time, namely, in the glock work function.

This patch was originally written by Steve Whitehouse, but since
he's on holiday, I'm submitting it.  It's been well tested with a
complex recovery test called revolver.
Signed-off-by: NSteve Whitehouse <swhiteho@redhat.com>
Signed-off-by: NBob Peterson <rpeterso@redhat.com>

This patch replaces a statement that got dropped out by accident.
Without the patch, truncates on stuffed (very small) files cause
those files to have an unpredictable size.
Signed-off-by: NBob Peterson <rpeterso@redhat.com>
Signed-off-by: NSteven Whitehouse <swhiteho@redhat.com>

This function is only called from one place and it's like this:
	dlm_register_domain(conn->cc_name, dlm_key, &fs_version);

The "conn->cc_name" is 64 characters long.  If strlen(conn->cc_name)
were equal to O2NM_MAX_NAME_LEN (64) that would be a bug because
strlen() doesn't count the NULL character.

In fact, if you look how O2NM_MAX_NAME_LEN is used, it mostly describes
64 character buffers.  The only exception is nd_name from struct
o2nm_node.

Anyway I looked into it and in this case the domain string comes from
osb->uuid_str in ocfs2_setup_osb_uuid().  That's 32 characters and NULL
which easily fits into O2NM_MAX_NAME_LEN.  This patch doesn't change how
the code works, but I think it makes the code a little cleaner.
Signed-off-by: NDan Carpenter <error27@gmail.com>
Signed-off-by: NJoel Becker <joel.becker@oracle.com>

The new reservation code in local alloc has add the limitation
that the caller should handle the case that the local alloc
doesn't give use enough contiguous clusters. It make the old
xattr reflink code broken.

So this patch udpate the xattr reflink code so that it can
handle the case that local alloc give us one cluster at a time.
Signed-off-by: NTao Ma <tao.ma@oracle.com>
Signed-off-by: NJoel Becker <joel.becker@oracle.com>

The old ocfs2_xattr_extent_allocation is too optimistic about
the clusters we can get. So actually if the file system is
too fragmented, ocfs2_add_clusters_in_btree will return us
with EGAIN and we need to allocate clusters once again.

So this patch change it to a while loop so that we can allocate
clusters until we reach clusters_to_add.
Signed-off-by: NTao Ma <tao.ma@oracle.com>
Signed-off-by: NJoel Becker <joel.becker@oracle.com>
Cc: stable@kernel.org

In ocfs2_block_group_alloc, we set c_blkno by bg->bg_blkno.
But actually bg->bg_blkno is already changed to little endian
in ocfs2_block_group_fill. So remove the extra cpu_to_le64.
Reported-by: NMarcos Matsunaga <Marcos.Matsunaga@oracle.com>
Signed-off-by: NTao Ma <tao.ma@oracle.com>
Signed-off-by: NJoel Becker <joel.becker@oracle.com>

dlm->recovery_map is defined as
	unsigned long recovery_map[BITS_TO_LONGS(O2NM_MAX_NODES)];

We should treat O2NM_MAX_NODES as the bit map size in bits.
This patches fixes a bit operation that takes O2NM_MAX_NODES + 1 as bitmap size.
Signed-off-by: NWengang Wang <wen.gang.wang@oracle.com>
Signed-off-by: NJoel Becker <joel.becker@oracle.com>