Unionfs: Rename unionfs_data sbcount field to more appropriate open_files

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

Unionfs: Proper comment on rwsem field

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

Unionfs: Documentation updates for branch-management

Describe dynamic branch-management introduced by subsequent patches.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs: Export drop_pagecache_sb symbol

Signed-off-by: Erez Zadok <ezkcs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipekcs.sunysb.edu>

fs: Introduce path{get,put}

Export drop_pagecache_sb symbol (for branch-management).

Signed-off-by: Erez Zadok <ezkcs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipekcs.sunysb.edu>

[PATCH] Unionfs: sioq not __exit

stop_sioq() is called from both __init and __exit functions, so it
shouldn't be marked __exit.

Reported on the kernelnewbies mailing list, but no patch offered there.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

MAINTAINERS: Unionfs URL update

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Fix unlocking in error paths

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Fix dentry leak in copyup_named_dentry

When we chmod a directory on a readonly branch, and have to copy it up, we
forget to dput(). If this was a file, it gets dput indirectly through other
functions we call, but not if it was a directory.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: mntput in __cleanup_dentry

This fixes a mnt refleak which occured during copyup when directory
hierarchy was recreated on a writable branch.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Fix copyup_deleted_file dentry leak

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Miscellaneous coding style fixes

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Check return value of d_path

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Use SEEK_{SET,CUR} instead of hardcoded values

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Remove alloc_filldir_node

alloc_filldir_node was used only once. Additionally, all the arguments
passed to it were ignored wasting stack space for no reason whatsoever.

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Rename unionfs_d_revalidate_wrap

Follow the convention of "foo" calling "__foo".

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Several small cleanups in unionfs_interpose

1) No need to lock the inode - lockdep was complaining about potential
circular dependency

2) No need to use temporary variable for iunique() inode number

3) Removed unneeded comment

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Don't grab dentry private data mutex in unionfs_d_release

Grabbing the UNIONFS_D(dentry)->lock is completely unnecessary and there are
no other references; we are about to free the object anyway. Additionally,
grabbing the mutex produces warning when the slab object is reused - as it
was freed while there still was a reference to it.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Fix a memory leak in unionfs_read_super

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs: Fix a memory leak & null pointer dereference

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

[PATCH] UNION_FS must depend on SLAB

On Sat, Feb 17, 2007 at 09:51:46PM -0800, Andrew Morton wrote:
>...
> Changes since 2.6.20-mm1:
>...
>  git-unionfs.patch
>...
>  git trees
>...

<--  snip  -->

...
  CC      fs/unionfs/copyup.o
/home/bunk/linux/kernel-2.6/linux-2.6.20-mm2/fs/unionfs/copyup.c: In function 'create_parents_named':
/home/bunk/linux/kernel-2.6/linux-2.6.20-mm2/fs/unionfs/copyup.c:620: error: 'malloc_sizes' undeclared (first use in this function)
/home/bunk/linux/kernel-2.6/linux-2.6.20-mm2/fs/unionfs/copyup.c:620: error: (Each undeclared identifier is reported only once
/home/bunk/linux/kernel-2.6/linux-2.6.20-mm2/fs/unionfs/copyup.c:620: error: for each function it appears in.)
make[3]: *** [fs/unionfs/copyup.o] Error 1

<--  snip  -->

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Remove unused structure members & macros

This patch removes:
- hidden_mnt pointer from struct unionfs_data
- mount_flag from struct unionfs_sb_info
- mount_flag related macros

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

eCryptfs: convert lookup_one_len() to lookup_one_len_nd()

Call the new lookup_one_len_nd() rather than lookup_one_len().  This fixes an
oops when stacked on NFS.

Note that there are still some issues with eCryptfs on NFS having to do with
directory deletion (I'm not getting an oops, just an -EBUSY).

Signed-off-by: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

Unionfs: Documentation update

Be little gentler & updated the URLs

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/: Move eCryptfs & Unionfs config options into a sub-menu

Using The Misc filesystems sub-menu for layered/stackable filesystems only
makes it harder for users to find eCryptfs/Unionfs.

Additionally, the menu can be easily turned into a menuconfig, which could
be used to turn on any VFS/VM functionality required by layered filesystems
(there is none at the moment).

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: Michael Halcrow <mhalcrow@us.ibm.com>

fs/unionfs/: Use __roundup_pow_of_two instead of custom rounding code

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Don't duplicate the struct nameidata

The only fields that we have to watch out for are the dentry and vfsmount.
Additionally, this makes Unionfs gentler on the stack as nameidata is rather
large.

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Andrew Morton's comments

- rename {,un}lock_dentry to unionfs_{,un}lock_dentry
- few minor coding style fixes
- removed prototypes from .c files
- replaced dbstart macros etc with static inlines
- replaced UNIONFS_D(d)->sem semaphore with a mutex
- renamed sioq struct workqueue to superio_workqueue
- made unionfs_get_nlinks and alloc_whname not inlined

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: possible cleanups

This patch contains the following possible cleanups:
- every function should #include the headers containing the prototypes
  of it's global functions
- static functions in C files shouldn't be marked "inline", gcc should
  know best when to inline them
- make needlessly global code static
- #if 0 the following unused global function:
  - stale_inode.c: is_stale_inode()

Signed-off-by: Adrian Bunk <bunk@stusta.de>
[removed stale inode related fixes as stale_inode.c is gone]
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

fs/unionfs/: Remove stale_inode.c

The stale inode operations were heavily based on bad inode operations. This
patch removes stale_inode.c and converts all users of stale_inode_ops to
bad_inode_ops as there seems to be no reason to return ESTALE instead of
EIO.

This is the more appropriate than porting the bad_inode.c fix (commit
be6aab0e9fa6d3c6d75aa1e38ac972d8b4ee82b8) to stale_inode.c.

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

Unionfs: Extended Attributes support

Extended attribute support.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Kconfig and Makefile

This patch contains the changes to fs Kconfig file, Makefiles, and Maintainers
file for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Unlink

This patch provides unlink functionality for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Include file

Global include file - can be included from userspace by utilities.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Internal include file

This patch contains an internal Unionfs include file. The include file is
specific to kernel code only, and therefore is separate from
include/linux/unionfs.h.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Helper macros/inlines

This patch contains many macros and inline functions used thoughout Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Handling of stale inodes

Provides nicer handling of stale inodes.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Superblock operations

This patch contains the superblock operations for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Miscellaneous helper functions

This patch contains miscellaneous helper functions used thoughout Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Privileged operations workqueue

Workqueue & helper functions used to perform privileged operations on
behalf of the user process.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Rename

This patch provides rename functionality for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Readdir state

This file contains the routines for maintaining readdir state.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Main module functions

Module init & cleanup code, as well as interposition functions.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Lookup helper functions

This patch provides helper functions for the lookup operations in Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Inode operations

This patch provides the inode operations for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Directory manipulation helper functions

This patch contains directory manipulation helper functions.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Directory file operations

This patch provides directory file operations.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: File operations

This patch provides the file operations for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Dentry operations

This patch contains the dentry operations for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Copyup Functionality

This patch contains the functions used to perform copyup operations in unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Common file operations

This patch contains helper functions used through the rest of the code which
pertains to files.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Unionfs: Branch management functionality

This patch contains the ioctls to increase the union generation and to query
which branch a file exists on.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

lookup_one_len_nd - lookup_one_len with nameidata argument

This patch renames lookup_one_len to lookup_one_len_nd, and adds a nameidata
argument. An inline function, lookup_one_len (which calls lookup_one_len_nd
with nd == NULL) preserves original behavior.

The following Unionfs patches depend on this one.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>

Unionfs: Documentation

This patch contains documentation for Unionfs. You will find several files
outlining basic unification concepts and rename semantics.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>

Linux 2.6.22.19

NETFILTER: nf_conntrack_tcp: conntrack reopening fix

[NETFILTER]: nf_conntrack_tcp: conntrack reopening fix

[Upstream commits b2155e7f + d0c1fd7a]

TCP connection tracking in netfilter did not handle TCP reopening
properly: active close was taken into account for one side only and
not for any side, which is fixed now. The patch includes more comments
to explain the logic how the different cases are handled.
The bug was discovered by Jeff Chua.

Signed-off-by: Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

SCSI: sd: handle bad lba in sense information

patch 366c246de9cec909c5eba4f784c92d1e75b4dc38 in mainline.

Some devices report medium error locations incorrectly.  Add guards to
make sure the reported bad lba is actually in the request that caused
it.  Additionally remove the large case statment for sector sizes and
replace it with the proper u64 divisions.

Tested-by: Mike Snitzer <snitzer@gmail.com>
Cc: Stable Tree <stable@kernel.org>
Cc: Tony Battersby <tonyb@cybernetics.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Be more robust about bad arguments in get_user_pages()

MAINLINE: 900cf086fd2fbad07f72f4575449e0d0958f860f

So I spent a while pounding my head against my monitor trying to figure
out the vmsplice() vulnerability - how could a failure to check for
*read* access turn into a root exploit? It turns out that it's a buffer
overflow problem which is made easy by the way get_user_pages() is
coded.

In particular, "len" is a signed int, and it is only checked at the
*end* of a do {} while() loop.  So, if it is passed in as zero, the loop
will execute once and decrement len to -1.  At that point, the loop will
proceed until the next invalid address is found; in the process, it will
likely overflow the pages array passed in to get_user_pages().

I think that, if get_user_pages() has been asked to grab zero pages,
that's what it should do.  Thus this patch; it is, among other things,
enough to block the (already fixed) root exploit and any others which
might be lurking in similar code.  I also think that the number of pages
should be unsigned, but changing the prototype of this function probably
requires some more careful review.

Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
CC: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

quicklists: Only consider memory that can be used with GFP_KERNEL

patch 96990a4ae979df9e235d01097d6175759331e88c in mainline.

Quicklists calculates the size of the quicklists based on the number of
free pages.  This must be the number of free pages that can be allocated
with GFP_KERNEL.  node_page_state() includes the pages in ZONE_HIGHMEM and
ZONE_MOVABLE which may lead the quicklists to become too large causing OOM.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Tested-by: Dhaval Giani <dhaval@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Oliver Pinter <oliver.pntr@gmail.com>

knfsd: query filesystem for NFSv4 getattr of FATTR4_MAXNAME

mainline: a16e92edcd0a2846455a30823e1bac964e743baa

Without this we always return 2^32-1 as the the maximum namelength.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Andreas Gruenbacher <agruen@suse.de>
CC: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

NFS: Fix an Oops in encode_lookup()

mainline: 54af3bb543c071769141387a42deaaab5074da55

It doesn't look as if the NFS file name limit is being initialised correctly
in the struct nfs_server. Make sure that we limit whatever is being set in
nfs_probe_fsinfo() and nfs_init_server().

Also ensure that readdirplus and nfs4_path_walk respect our file name
limits.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Neil Brown <neilb@suse.de>
CC: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

NFSv2/v3: Fix a memory leak when using -onolock

mainline: 5cef338b30c110daf547fb13d99f0c77f2a79fbc

    Neil Brown said:
    > Hi Trond,
    >
    > We found that a machine which made moderately heavy use of
    > 'automount' was leaking some nfs data structures - particularly the
    > 4K allocated by rpc_alloc_iostats.
    > It turns out that this only happens with filesystems with -onolock
    > set.

    > The problem is that if NFS_MOUNT_NONLM is set, nfs_start_lockd doesn't
    > set server->destroy, so when the filesystem is unmounted, the
    > ->client_acl is not shutdown, and so several resources are still
    > held.  Multiple mount/umount cycles will slowly eat away memory
    > several pages at a time.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Acked-by: Neil Brown <neilb@suse.de>
Signed-off-by: Neil Brown <neilb@suse.de>
CC: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

NFS: Fix nfs_reval_fsid()

mainline: a0356862bcbeb20acf64bc1a82d28a4c5bb957a7

We don't need to revalidate the fsid on the root directory. It suffices to
revalidate it on the current directory.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Acked-by: Neil Brown <neilb@suse.de>
CC: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

knfsd: fix spurious EINVAL errors on first access of new filesystem

mainline: ac8587dcb58e40dd336d99d60f852041e06cc3dd

The v2/v3 acl code in nfsd is translating any return from fh_verify() to
nfserr_inval.  This is particularly unfortunate in the case of an
nfserr_dropit return, which is an internal error meant to indicate to
callers that this request has been deferred and should just be dropped
pending the results of an upcall to mountd.

Thanks to Roland <devzero@web.de> for bug report and data collection.

Cc: Roland <devzero@web.de>
Acked-by: Andreas Gruenbacher <agruen@suse.de>
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Reviewed-By: NeilBrown <neilb@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
CC: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

via-velocity: don't oops on MTU change (resend)

mainline: 48f6b053613b62fed7a2fe3255e5568260a8d615

The VIA veloicty driver needs the following to allow changing MTU when down.
The buffer size needs to be computed when device is brought up, not when
device is initialized.  This also fixes a bug where the buffer size was
computed differently on change_mtu versus initial setting.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Acked-by: Jeff Mahoney <jeffm@suse.com>
CC: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

VIA_VELOCITY: Don't oops on MTU change.

mainline: bd7b3f34198071d8bec05180530c362f1800ba46

Simple mtu change when device is down.
Fix http://bugzilla.kernel.org/show_bug.cgi?id=9382.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Jeff Mahoney <jeffm@suse.com>
CC: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

sony-laptop: call sonypi_compat_init earlier

mainline: 015a916fbbf105bb15f4bbfd80c3b9b2f2e0d7db

sonypi_compat uses a kfifo that needs to be present before _SRS is
called to be able to cope with the IRQs triggered when setting
resources.

Signed-off-by: Mattia Dongili <malattia@linux.it>
Signed-off-by: Len Brown <len.brown@intel.com>
Acked-by: Jeff Mahoney <jeffm@suse.com>
CC: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

pci: fix unterminated pci_device_id lists

mainline: 248bdd5efca5a113cbf443a993c69e53d370236b

Fix a couple drivers that do not correctly terminate their pci_device_id
lists.  This results in garbage being spewed into modules.pcimap when the
module happens to not have 28 NULL bytes following the table, and/or the
last PCI ID is actually truncated from the table when calculating the
modules.alias PCI aliases, cause those unfortunate device IDs to not
auto-load.

Signed-off-by: Kees Cook <kees@ubuntu.com>
Acked-by: Corey Minyard <minyard@acm.org>
Cc: David Woodhouse <dwmw2@infradead.org>
Acked-by: Jeff Garzik <jeff@garzik.org>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Jeff Mahoney <jeffm@suse.com>
CC: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Intel_agp: really fix 945/965GME

mainline: dde4787642ee3cb85aef80bdade04b6f8ddc3df8

Fix some missing places to check with device id info, which
should probe the device gart correctly.

Signed-off-by: Wang Zhenyu <zhenyu.z.wang@intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Acked-by: Takashi Iwai <tiwai@suse.de>
CC: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

i386: fixup TRACE_IRQ breakage (CVE-2007-3731)

mainline: a10d9a71bafd3a283da240d2868e71346d2aef6f

The TRACE_IRQS_ON function in iret_exc: calls a C function without
ensuring that the segments are set properly. Move the trace function and
the enabling of interrupt into the C stub.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Jeff Mahoney <jeffm@suse.com>
CC: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Handle bogus %cs selector in single-step instruction decoding (CVE-2007-3731)

Handle bogus %cs selector in single-step instruction decoding

mainline: 29eb51101c02df517ca64ec472d7501127ad1da8

The code for LDT segment selectors was not robust in the face of a bogus
selector set in %cs via ptrace before the single-step was done.

Signed-off-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Jeff Mahoney <jeffm@suse.com>
CC: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

NFS: Fix a potential file corruption issue when writing

patch 5d47a35600270e7115061cb1320ee60ae9bcb6b8 in mainline.

If the inode is flagged as having an invalid mapping, then we can't rely on
the PageUptodate() flag. Ensure that we don't use the "anti-fragmentation"
write optimisation in nfs_updatepage(), since that will cause NFS to write
out areas of the page that are no longer guaranteed to be up to date.

A potential corruption could occur in the following scenario:

client 1 client 2
=============== ===============
fd=open("f",O_CREAT|O_WRONLY,0644);
write(fd,"fubar\n",6); // cache last page
close(fd);
fd=open("f",O_WRONLY|O_APPEND);
write(fd,"foo\n",4);
close(fd);

fd=open("f",O_WRONLY|O_APPEND);
write(fd,"bar\n",4);
close(fd);
-----
The bug may lead to the file "f" reading 'fubar\n\0\0\0\nbar\n' because
client 2 does not update the cached page after re-opening the file for
write. Instead it keeps it marked as PageUptodate() until someone calls
invalidate_inode_pages2() (typically by calling read()).

The bug was introduced by commit 44b11874ff583b6e766a05856b04f3c492c32b84
"NFS: Separate metadata and page cache revalidation mechanisms"

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

quicklists: do not release off node pages early

dhaval@linux.vnet.ibm.com, clameter@sgi.com
Message-ID: <200712222203.lBMM3Nsk021922@imap1.linux-foundation.org>

From: Christoph Lameter <clameter@sgi.com>

patch ed367fc3a7349b17354c7acef551533337764859 in mainline.

quicklists must keep even off node pages on the quicklists until the TLB
flush has been completed.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Cc: Dhaval Giani <dhaval@linux.vnet.ibm.com>
Cc: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

PCI: Fix fakephp deadlock

This patch works around a problem in the fakephp driver when a process
writing "0" to a "power" sysfs file to fake removal of a PCI device ends
up deadlocking itself in the sysfs code.

The patch is functionally identical to the one in Linus' tree post 2.6.24:
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=5c796ae7a7ebe56967ed9b9963d7c16d733635ff

I have tested it on a 2.6.22 kernel.

Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

sata_promise: ASIC PRD table bug workaround

patch b9ccd4a90bbb964506f01b4bdcff4f50f8d5d334 in mainline.

Second-generation Promise SATA controllers have an ASIC bug
which can trigger if the last PRD entry is larger than 164 bytes,
resulting in intermittent errors and possible data corruption.

Work around this by replacing calls to ata_qc_prep() with a
private version that fills the PRD, checks the size of the
last entry, and if necessary splits it to avoid the bug.
Also reduce sg_tablesize by 1 to accommodate the new entry.

Tested on the second-generation SATA300 TX4 and SATA300 TX2plus,
and the first-generation PDC20378.

Thanks to Alexander Sabourenkov for verifying the bug by
studying the vendor driver, and for writing the initial patch
upon which this one is based.

Signed-off-by: Mikael Pettersson <mikpe@it.uu.se>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

sata_promise: FastTrack TX4200 is a second-generation chip

patch 7f9992a23190418592f0810900e4f91546ec41da in mainline.

This patch corrects sata_promise to classify FastTrack TX4200
(DID 3515/3519) as a second-generation chip. Promise's partial-
source FT TX4200 driver confirms this classification.

Treating it as a first-generation chip causes several problems:
1. Detection failures. This is a recent regression triggered by
   the hotplug-enabling changes in 2.6.23-rc1.
2. Various "failed to resume link for reset" warnings.

This patch fixes <http://bugzilla.kernel.org/show_bug.cgi?id=8936>.

Thanks to Stephen Ziemba for reporting the bug and for testing the fix.

Signed-off-by: Mikael Pettersson <mikpe@it.uu.se>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

cciss: fix memory leak

mainline: f2912a1223c0917a7b4e054f18086209137891ea

There's a memory leak in the cciss driver.

in alloc_cciss_hba() we may leak sizeof(ctlr_info_t) bytes if a
call to alloc_disk(1 << NWD_SHIFT) fails.
This patch should fix the issue.

Spotted by the Coverity checker.

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Acked-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oliver Pinter <oliver.pntr@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Linux 2.6.22.18

splice: fix user pointer access in get_iovec_page_array() (CVE-2008-0600)

patch 712a30e63c8066ed84385b12edbfb804f49cbc44 in mainline.

Commit 8811930dc74a503415b35c4a79d14fb0b408a361 ("splice: missing user
pointer access verification") added the proper access_ok() calls to
copy_from_user_mmap_sem() which ensures we can copy the struct iovecs
from userspace to the kernel.

But we also must check whether we can access the actual memory region
pointed to by the struct iovec to fix the access checks properly.

Signed-off-by: Bastian Blank <waldi@debian.org>
Acked-by: Oliver Pinter <oliver.pntr@gmail.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Linux 2.6.22.17

vm audit: add VM_DONTEXPAND to mmap for drivers that need it (CVE-2008-0007)

Drivers that register a ->fault handler, but do not range-check the
offset argument, must set VM_DONTEXPAND in the vm_flags in order to
prevent an expanding mremap from overflowing the resource.

I've audited the tree and attempted to fix these problems (usually by
adding VM_DONTEXPAND where it is not obvious).

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ACPI: apply quirk_ich6_lpc_acpi to more ICH8 and ICH9

patch d1ec7298fcefd7e4d1ca612da402ce9e5d5e2c13 in mainline.

It is important that these resources be reserved
to avoid conflicts with well known ACPI registers.

Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

POWERPC: Fix invalid semicolon after if statement

Patch 2b02d13996fe28478e45605de9bd8bdca25718de in mainline

[POWERPC] Fix invalid semicolon after if statement

A similar fix to netfilter from Eric Dumazet inspired me to
look around a bit by using some grep/sed stuff as looking for
this kind of bugs seemed easy to automate.  This is one of them
I found where it looks like this semicolon is not valid.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

chelsio: Fix skb->dev setting

patch 7de6af0f23b25df8da9719ecae1916b669d0b03d in mainline.

eth_type_trans() now sets skb->dev.
Access skb->def after it gets set.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

cxgb: fix stats

patch e0348b9ae5374f9a24424ae680bcd80724415f60 in mainline.

Fix MAC stats accounting.
Fix get_stats.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

cxgb: fix T2 GSO

patch 7832ee034b6ef78aab020c9ec1348544cd65ccbd in mainline.

The patch ensures that a GSO skb has enough headroom
to push an encapsulating cpl_tx_pkt_lso header.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

vfs: coredumping fix (CVE-2007-6206)

vfs: coredumping fix

patch c46f739dd39db3b07ab5deb4e3ec81e1c04a91af in mainline

fix: http://bugzilla.kernel.org/show_bug.cgi?id=3043

only allow coredumping to the same uid that the coredumping
task runs under.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Alan Cox <alan@redhat.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Acked-by: Al Viro <viro@ftp.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: maximilian attems <max@stro.at>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ACPICA: fix acpi-cpufreq boot crash due to _PSD return-by-reference

patch 152c300d007c70c4a1847dad39ecdaba22e7d457 in mainline.

Changed resolution of named references in packages

Fixed a problem with the Package operator where all named
references were created as object references and left otherwise
unresolved. According to the ACPI specification, a Package can
only contain Data Objects or references to control methods. The
implication is that named references to Data Objects (Integer,
Buffer, String, Package, BufferField, Field) should be resolved
immediately upon package creation. This is the approach taken
with this change. References to all other named objects (Methods,
Devices, Scopes, etc.) are all now properly created as reference objects.

http://bugzilla.kernel.org/show_bug.cgi?id=5328
http://bugzilla.kernel.org/show_bug.cgi?id=9429

Signed-off-by: Bob Moore <robert.moore@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

CASSINI: Set skb->truesize properly on receive packets.

[ Upstream commit: d011a231675b240157a3c335dd53e9b849d7d30d ]

skb->truesize was not being incremented at all to
reflect the page based data added to RX SKBs.

Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

CASSINI: Revert 'dont touch page_count'.

[ Upstream commit: 9de4dfb4c7176e5bb232a21cdd8df78da2b15cac ]

This reverts changeset fa4f0774d7c6cccb4d1fda76b91dd8eddcb2dd6a
([CASSINI]: dont touch page_count) because it breaks the driver.

The local page counting added by this changeset did not account
for the asynchronous page count changes done by kfree_skb()
and friends.

The change adds extra atomics and on top of it all appears to be
totally unnecessary as well.

Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

CASSINI: Fix endianness bug.

[ Upstream commit: e5e025401f6e926c1d9dc3f3f2813cf98a2d8708 ]

Here's proposed fix for RX checksum handling in cassini; it affects
little-endian working with half-duplex gigabit, but obviously needs
testing on big-endian too.

The problem is, we need to convert checksum to fixed-endian *before*
correcting for (unstripped) FCS.  On big-endian it won't matter
(conversion is no-op), on little-endian it will, but only if FCS is
not stripped by hardware; i.e. in half-duplex gigabit mode when
->crc_size is set.

cassini.c part is that fix, cassini.h one consists of trivial
endianness annotations.  With that applied the sucker is endian-clean,
according to sparse.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ATM: Check IP header validity in mpc_send_packet

[ATM]: Check IP header validity in mpc_send_packet

[ Upstream commit: 1c9b7aa1eb40ab708ef3242f74b9a61487623168 ]

Al went through the ip_fast_csum callers and found this piece of code
that did not validate the IP header.  While root crashing the machine
by sending bogus packets through raw or AF_PACKET sockets isn't that
serious, it is still nice to react gracefully.

This patch ensures that the skb has enough data for an IP header and
that the header length field is valid.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ATM: [nicstar] delay irq setup until card is configured

[ATM]: [nicstar] delay irq setup until card is configured

[ Upstream commit: 52961955aa180959158faeb9fd6b4f8a591450f5 ]

Signed-off-by: Chas Williams <chas@cmf.nrl.navy.mil>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

CONNECTOR: Don't touch queue dev after decrement of ref count.

[CONNECTOR]: Don't touch queue dev after decrement of ref count.

[ Upstream commit: cf585ae8ae9ac7287a6d078425ea32f22bf7f1f7 ]

cn_queue_free_callback() will touch 'dev'(i.e. cbq->pdev), so it
should be called before atomic_dec(&dev->refcnt).

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Fix sparc64 cpu cross call hangs.

[SPARC64]: Fix endless loop in cheetah_xcall_deliver().

[ Upsteam commit: 0de56d1ab83323d604d95ca193dcbd28388dbabb ]

We need to mask out the proper bits when testing the dispatch status
register else we can see unrelated NACK bits from previous cross call
sends.

Signed-off-by: David S. Miller <davem@davemloft.net>

INET: Fix netdev renaming and inet address labels

[INET]: Fix netdev renaming and inet address labels

[ Upstream commit: 44344b2a85f03326c7047a8c861b0c625c674839 ]

When re-naming an interface, the previous secondary address
labels get lost e.g.

  $> brctl addbr foo
  $> ip addr add 192.168.0.1 dev foo
  $> ip addr add 192.168.0.2 dev foo label foo:00
  $> ip addr show dev foo | grep inet
    inet 192.168.0.1/32 scope global foo
    inet 192.168.0.2/32 scope global foo:00
  $> ip link set foo name bar
  $> ip addr show dev bar | grep inet
    inet 192.168.0.1/32 scope global bar
    inet 192.168.0.2/32 scope global bar:2

Turns out to be a simple thinko in inetdev_changename() - clearly we
want to look at the address label, rather than the device name, for
a suffix to retain.

Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

IPSEC: Avoid undefined shift operation when testing algorithm ID

[IPSEC]: Avoid undefined shift operation when testing algorithm ID

[ Upstream commit: f398035f2dec0a6150833b0bc105057953594edb ]

The aalgos/ealgos fields are only 32 bits wide.  However, af_key tries
to test them with the expression 1 << id where id can be as large as
253.  This produces different behaviour on different architectures.

The following patch explicitly checks whether ID is greater than 31
and fails the check if that's the case.

We cannot easily extend the mask to be longer than 32 bits due to
exposure to user-space.  Besides, this whole interface is obsolete
anyway in favour of the xfrm_user interface which doesn't use this
bit mask in templates (well not within the kernel anyway).

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

IPSEC: Fix potential dst leak in xfrm_lookup

[IPSEC]: Fix potential dst leak in xfrm_lookup

[ Upstream commit: 75b8c133267053c9986a7c8db5131f0e7349e806 ]

If we get an error during the actual policy lookup we don't free the
original dst while the caller expects us to always free the original
dst in case of error.

This patch fixes that.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

IPV4: ip_gre: set mac_header correctly in receive path

[IPV4] ip_gre: set mac_header correctly in receive path

[ Upstream commit: 1d0691674764098304ae4c63c715f5883b4d3784 ]

mac_header update in ipgre_recv() was incorrectly changed to
skb_reset_mac_header() when it was introduced.

Signed-off-by: Timo Teras <timo.teras@iki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

IPV4 ROUTE: ip_rt_dump() is unecessary slow

[IPV4] ROUTE: ip_rt_dump() is unecessary slow

[ Upstream commit: d8c9283089287341c85a0a69de32c2287a990e71 ]

I noticed "ip route list cache x.y.z.t" can be *very* slow.

While strace-ing -T it I also noticed that first part of route cache
is fetched quite fast :

recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202
GXm\0\0\2  \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3772 <0.000047>
recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\234\0\0\0\30\0\2\0\254i\
202GXm\0\0\2  \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3736 <0.000042>
recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\204\0\0\0\30\0\2\0\254i\
202GXm\0\0\2  \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3740 <0.000055>
recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\234\0\0\0\30\0\2\0\254i\
202GXm\0\0\2  \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3712 <0.000043>
recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\204\0\0\0\30\0\2\0\254i\
202GXm\0\0\2  \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3732 <0.000053>
recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202
GXm\0\0\2  \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3708 <0.000052>
recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202
GXm\0\0\2  \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3680 <0.000041>

while the part at the end of the table is more expensive:

recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\204\0\0\0\30\0\2\0\254i\202GXm\0\0\2  \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3656 <0.003857>
recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\204\0\0\0\30\0\2\0\254i\202GXm\0\0\2  \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3772 <0.003891>
recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202GXm\0\0\2  \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3712 <0.003765>
recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202GXm\0\0\2  \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3700 <0.003879>
recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202GXm\0\0\2  \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3676 <0.003797>
recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202GXm\0\0\2  \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3724 <0.003856>
recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\234\0\0\0\30\0\2\0\254i\202GXm\0\0\2  \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3736 <0.003848>

The following patch corrects this performance/latency problem,
removing quadratic behavior.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>