Signed-off-by: NPhil Edworthy <phil.edworthy@renesas.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The mdio-mux driver scans all child mdio nodes, without regard to whether
the node is actually used.  Some device trees include all possible
mdio-mux nodes and rely on the boot loader to disable those that are not
present, based on some run-time configuration.  Those nodes need to be
skipped.
Signed-off-by: NTimur Tabi <timur@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Macro for_each_child_of_node() makes it easy to iterate over all of the
children for a given device tree node, including those nodes that are
marked as unavailable (i.e. status = "disabled").

Introduce for_each_available_child_of_node(), which is like
for_each_child_of_node(), but it automatically skips unavailable nodes.
This also requires the introduction of helper function
of_get_next_available_child(), which returns the next available child
node.
Signed-off-by: NTimur Tabi <timur@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

A race exists where creating cgroups and also updating the priomap
may result in losing a priomap update. This is because priomap
writers are not protected by rtnl_lock.

Move priority writer into rtnl_lock()/rtnl_unlock().

CC: Neil Horman <nhorman@tuxdriver.com>
Reported-by: NAl Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: NJohn Fastabend <john.r.fastabend@intel.com>
Acked-by: NNeil Horman <nhorman@tuxdriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

A socket fd passed in a SCM_RIGHTS datagram was not getting
updated with the new tasks cgrp prioidx. This leaves IO on
the socket tagged with the old tasks priority.

To fix this add a check in the scm recvmsg path to update the
sock cgrp prioidx with the new tasks value.

Thanks to Al Viro for catching this.

CC: Neil Horman <nhorman@tuxdriver.com>
Reported-by: NAl Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: NJohn Fastabend <john.r.fastabend@intel.com>
Acked-by: NNeil Horman <nhorman@tuxdriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add lock to prevent a race with a file closing and also remove
useless and ugly sscanf code. The extra code was never needed
and the case it supposedly protected against is in fact handled
correctly by sock_from_file as pointed out by Al Viro.

CC: Neil Horman <nhorman@tuxdriver.com>
Reported-by: NAl Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: NJohn Fastabend <john.r.fastabend@intel.com>
Acked-by: NNeil Horman <nhorman@tuxdriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We drop packet unconditionally when we fail to mirror it. This is not intended
in some cases. Consdier for kvm guest, we may mirror the traffic of the bridge
to a tap device used by a VM. When kernel fails to mirror the packet in
conditions such as when qemu crashes or stop polling the tap, it's hard for the
management software to detect such condition and clean the the mirroring
before. This would lead all packets to the bridge to be dropped and break the
netowrk of other virtual machines.

To solve the issue, the patch does not drop packets when kernel fails to mirror
it, and only drop the redirected packets.
Signed-off-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NJamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The implementation of dev_ifconf() for the compat ioctl interface uses
an intermediate ifc structure allocated in userland for the duration of
the syscall. Though, it fails to initialize the padding bytes inserted
for alignment and that for leaks four bytes of kernel stack. Add an
explicit memset(0) before filling the structure to avoid the info leak.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If at least one of CONFIG_IP_VS_PROTO_TCP or CONFIG_IP_VS_PROTO_UDP is
not set, __ip_vs_get_timeouts() does not fully initialize the structure
that gets copied to userland and that for leaks up to 12 bytes of kernel
stack. Add an explicit memset(0) before passing the structure to
__ip_vs_get_timeouts() to avoid the info leak.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Cc: Wensong Zhang <wensong@linux-vs.org>
Cc: Simon Horman <horms@verge.net.au>
Cc: Julian Anastasov <ja@ssi.bg>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The CCID3 code fails to initialize the trailing padding bytes of struct
tfrc_tx_info added for alignment on 64 bit architectures. It that for
potentially leaks four bytes kernel stack via the getsockopt() syscall.
Add an explicit memset(0) before filling the structure to avoid the
info leak.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Cc: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

ccid_hc_rx_getsockopt() and ccid_hc_tx_getsockopt() might be called with
a NULL ccid pointer leading to a NULL pointer dereference. This could
lead to a privilege escalation if the attacker is able to map page 0 and
prepare it with a fake ccid_ops pointer.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Cc: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Cc: stable@vger.kernel.org
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The LLC code wrongly returns 0, i.e. "success", when the socket is
zapped. Together with the uninitialized uaddrlen pointer argument from
sys_getsockname this leads to an arbitrary memory leak of up to 128
bytes kernel stack via the getsockname() syscall.

Return an error instead when the socket is zapped to prevent the info
leak. Also remove the unnecessary memset(0). We don't directly write to
the memory pointed by uaddr but memcpy() a local structure at the end of
the function that is properly initialized.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The L2TP code for IPv6 fails to initialize the l2tp_unused member of
struct sockaddr_l2tpip6 and that for leaks two bytes kernel stack via
the getsockname() syscall. Initialize l2tp_unused with 0 to avoid the
info leak.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Cc: James Chapman <jchapman@katalix.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The L2CAP code fails to initialize the l2_bdaddr_type member of struct
sockaddr_l2 and the padding byte added for alignment. It that for leaks
two bytes kernel stack via the getsockname() syscall. Add an explicit
memset(0) before filling the structure to avoid the info leak.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Johan Hedberg <johan.hedberg@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The RFCOMM code fails to initialize the trailing padding byte of struct
sockaddr_rc added for alignment. It that for leaks one byte kernel stack
via the getsockname() syscall. Add an explicit memset(0) before filling
the structure to avoid the info leak.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Johan Hedberg <johan.hedberg@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The RFCOMM code fails to initialize the two padding bytes of struct
rfcomm_dev_list_req inserted for alignment before copying it to
userland. Additionally there are two padding bytes in each instance of
struct rfcomm_dev_info. The ioctl() that for disclosures two bytes plus
dev_num times two bytes uninitialized kernel heap memory.

Allocate the memory using kzalloc() to fix this issue.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Johan Hedberg <johan.hedberg@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The RFCOMM code fails to initialize the key_size member of struct
bt_security before copying it to userland -- that for leaking one
byte kernel stack. Initialize key_size with 0 to avoid the info
leak.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Johan Hedberg <johan.hedberg@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The HCI code fails to initialize the hci_channel member of struct
sockaddr_hci and that for leaks two bytes kernel stack via the
getsockname() syscall. Initialize hci_channel with 0 to avoid the
info leak.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Johan Hedberg <johan.hedberg@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The HCI code fails to initialize the two padding bytes of struct
hci_ufilter before copying it to userland -- that for leaking two
bytes kernel stack. Add an explicit memset(0) before filling the
structure to avoid the info leak.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Johan Hedberg <johan.hedberg@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The ATM code fails to initialize the two padding bytes of struct
sockaddr_atmpvc inserted for alignment. Add an explicit memset(0)
before filling the structure to avoid the info leak.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The ATM code fails to initialize the two padding bytes of struct
sockaddr_atmpvc inserted for alignment. Add an explicit memset(0)
before filling the structure to avoid the info leak.
Signed-off-by: NMathias Krause <minipli@googlemail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

dev_open() can complete before register_netdev() returns.
Fix vmxnet3_probe_device() to support this.
Signed-off-by: NSteve Hodgson <steve@purestorage.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Alternative solution for problem found by Linux Driver Verification
project (linuxtesting.org).

As it noted in the comment before the br_handle_frame_finish
function, this function should be called under rcu_read_lock.

The problem callgraph:
br_dev_xmit -> br_nf_pre_routing_finish_bridge_slow ->
 -> br_handle_frame_finish -> br_port_get_rcu -> rcu_dereference

And in this case there is no read-lock section.
Reported-by: NDenis Efremov <yefremov.denis@gmail.com>
Signed-off-by: NStephen Hemminger <shemminger@vyatta.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

John W. Linville says:

====================
Alexey Khoroshilov provides a potential memory leak in rndis_wlan.

Bob Copeland gives us an ath5k fix for a lockdep problem.

Dan Carpenter fixes a signedness mismatch in at76c50x.

Felix Fietkau corrects a regression caused by an earlier commit that can
lead to an IRQ storm.

Lorenzo Bianconi offers a fix for a bad variable initialization in ath9k
that can cause it to improperly mark decrypted frames.

Rajkumar Manoharan fixes ath9k to prevent the btcoex time from running
when the hardware is asleep.

The remainder are Bluetooth fixes, about which Gustavo says:

	"Here goes some fixes for 3.6-rc1, there are a few fix to
	thte inquiry code by Ram Malovany, support for 2 new devices,
	and few others fixes for NULL dereference, possible deadlock
	and a memory leak."
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Cong Wang reports that lockdep detected suspicious RCU usage while
enabling IPV6 forwarding:

 [ 1123.310275] ===============================
 [ 1123.442202] [ INFO: suspicious RCU usage. ]
 [ 1123.558207] 3.6.0-rc1+ #109 Not tainted
 [ 1123.665204] -------------------------------
 [ 1123.768254] include/linux/rcupdate.h:430 Illegal context switch in RCU read-side critical section!
 [ 1123.992320]
 [ 1123.992320] other info that might help us debug this:
 [ 1123.992320]
 [ 1124.307382]
 [ 1124.307382] rcu_scheduler_active = 1, debug_locks = 0
 [ 1124.522220] 2 locks held by sysctl/5710:
 [ 1124.648364]  #0:  (rtnl_mutex){+.+.+.}, at: [<ffffffff81768498>] rtnl_trylock+0x15/0x17
 [ 1124.882211]  #1:  (rcu_read_lock){.+.+.+}, at: [<ffffffff81871df8>] rcu_lock_acquire+0x0/0x29
 [ 1125.085209]
 [ 1125.085209] stack backtrace:
 [ 1125.332213] Pid: 5710, comm: sysctl Not tainted 3.6.0-rc1+ #109
 [ 1125.441291] Call Trace:
 [ 1125.545281]  [<ffffffff8109d915>] lockdep_rcu_suspicious+0x109/0x112
 [ 1125.667212]  [<ffffffff8107c240>] rcu_preempt_sleep_check+0x45/0x47
 [ 1125.781838]  [<ffffffff8107c260>] __might_sleep+0x1e/0x19b
[...]
 [ 1127.445223]  [<ffffffff81757ac5>] call_netdevice_notifiers+0x4a/0x4f
[...]
 [ 1127.772188]  [<ffffffff8175e125>] dev_disable_lro+0x32/0x6b
 [ 1127.885174]  [<ffffffff81872d26>] dev_forward_change+0x30/0xcb
 [ 1128.013214]  [<ffffffff818738c4>] addrconf_forward_change+0x85/0xc5
[...]

addrconf_forward_change() uses RCU iteration over the netdev list,
which is unnecessary since it already holds the RTNL lock.  We also
cannot reasonably require netdevice notifier functions not to sleep.
Reported-by: NCong Wang <amwang@redhat.com>
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Convert a 0 error return code to a negative one, as returned elsewhere in the
function.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@@
identifier ret;
expression e,e1,e2,e3,e4,x;
@@

(
if (\(ret != 0\|ret < 0\) || ...) { ... return ...; }
|
ret = 0
)
... when != ret = e1
*x = \(kmalloc\|kzalloc\|kcalloc\|devm_kzalloc\|ioremap\|ioremap_nocache\|devm_ioremap\|devm_ioremap_nocache\)(...);
... when != x = e2
    when != ret = e3
*if (x == NULL || ...)
{
  ... when != ret = e4
*  return ret;
}
// </smpl>
Signed-off-by: NJulia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Convert a 0 error return code to a negative one, as returned elsewhere in the
function.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@@
identifier ret;
expression e,e1,e2,e3,e4,x;
@@

(
if (\(ret != 0\|ret < 0\) || ...) { ... return ...; }
|
ret = 0
)
... when != ret = e1
*x = \(kmalloc\|kzalloc\|kcalloc\|devm_kzalloc\|ioremap\|ioremap_nocache\|devm_ioremap\|devm_ioremap_nocache\)(...);
... when != x = e2
    when != ret = e3
*if (x == NULL || ...)
{
  ... when != ret = e4
*  return ret;
}
// </smpl>
Signed-off-by: NJulia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If the NULL test is necessary, the initialization involving a dereference of
the tested value should be moved after the NULL test.

The sematic patch that fixes this problem is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@
type T;
expression E;
identifier i,fld;
statement S;
@@

- T i = E->fld;
+ T i;
  ... when != E
      when != i
  if (E == NULL) S
+ i = E->fld;
// </smpl>
Signed-off-by: NJulia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When registering the handlers, any state they rely on must be
completely initialised first.  When unregistering, we must wait until
they are definitely no longer running.  llc_rcv() must also avoid
reading the handler pointers again after checking for NULL.
Signed-off-by: NBen Hutchings <ben@decadent.org.uk>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Otherwise the station packet handler will remain registered even though
the module is unloaded.
Signed-off-by: NBen Hutchings <ben@decadent.org.uk>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

llc_station_init() creates and processes an event skb with no effect
other than to change the state from DOWN to UP.  Allocation failure is
reported, but then ignored by its caller, llc2_init().  Remove this
possibility by simply initialising the state as UP.
Signed-off-by: NBen Hutchings <ben@decadent.org.uk>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fix error handling in case making of dir dev_snmp6 failes
Signed-off-by: NIgor Maravic <igorm@etf.rs>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit caacf05e causes big drop of UDP loop back performance.
The cause of the regression is that we do not cache the local output
routes. Each time we send a datagram from unconnected UDP socket,
the kernel allocates a dst_entry and adds it to the rt_uncached_list.
It creates lock contention on the rt_uncached_lock.
Reported-by: NAlex Shi <alex.shi@intel.com>
Signed-off-by: NYan, Zheng <zheng.z.yan@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NBjørn Mork <bjorn@mork.no>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Take advantage of the matching macros to make the device id
list easier to read and maintain.
Signed-off-by: NBjørn Mork <bjorn@mork.no>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add 6 new devices and one modified device, based on
information from laptop vendor Windows drivers.

Sony provides a driver with two new devices using
a Gobi 2k+ layout (1199:68a5 and 1199:68a9).  The
Sony driver also adds a non-standard QMI/net
interface to the already supported 1199:9011
Gobi device. We do not know whether this is an
alternate interface number or an additional
interface which might be present, but that doesn't
really matter.

Lenovo provides a driver supporting 4 new devices:
 - MC7770 (1199:901b) with standard Gobi 2k+ layout
 - MC7700 (0f3d:68a2) with layout similar to MC7710
 - MC7750 (114f:68a2) with layout similar to MC7710
 - EM7700 (1199:901c) with layout similar to MC7710

Note regaring the three devices similar to MC7710:

The Windows drivers only support interface #8 on these
devices.  The MC7710 can support QMI/net functions on
interface #19 and #20 as well, and this driver is
verified to work on interface #19 (a firmware bug is
suspected to prevent #20 from working).

We do not enable these additional interfaces until they
either show up in a Windows driver or are verified to
work in some other way.  Therefore limiting the new
devices to interface #8 for now.
Signed-off-by: NBjørn Mork <bjorn@mork.no>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This driver support many composite USB devices where the
interface class/subclass/protocol provides no information
about the interface function. Interfaces with different
functions may all use ff/ff/ff, like this example of
a device with three serial interfaces and three QMI/wwan
interfaces:

T:  Bus=02 Lev=01 Prnt=01 Port=03 Cnt=01 Dev#=116 Spd=480  MxCh= 0
D:  Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 #Cfgs=  1
P:  Vendor=1199 ProdID=68a2 Rev= 0.06
S:  Manufacturer=Sierra Wireless, Incorporated
S:  Product=MC7710
S:  SerialNumber=3581780xxxxxx
C:* #Ifs= 6 Cfg#= 1 Atr=e0 MxPwr=  0mA
I:* If#= 0 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=qcserial
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=4ms
I:* If#= 2 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=qcserial
E:  Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=4ms
I:* If#= 3 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=qcserial
E:  Ad=83(I) Atr=03(Int.) MxPS=  64 Ivl=2ms
E:  Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=4ms
I:* If#= 8 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=qmi_wwan
E:  Ad=85(I) Atr=03(Int.) MxPS=  64 Ivl=2ms
E:  Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=4ms
I:* If#=19 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=qmi_wwan
E:  Ad=87(I) Atr=03(Int.) MxPS=  64 Ivl=2ms
E:  Ad=88(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=05(O) Atr=02(Bulk) MxPS= 512 Ivl=4ms
I:* If#=20 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none)
E:  Ad=89(I) Atr=03(Int.) MxPS=  64 Ivl=2ms
E:  Ad=8a(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=06(O) Atr=02(Bulk) MxPS= 512 Ivl=4ms

Instead of class/subclass/protocol the vendor use fixed
interface numbers for each function, and the Windows
drivers use these numbers to match driver and function.

The driver has had its own interface number whitelisting
code to simulate this functionality.  Replace this with
generic interface number matching now that the USB subsystem
support is there. This
 - removes the need for a driver_info structure per
   interface number,
 - avoids running the probe function for unsupported
   interfaces, and
 - simplifies the code.
Signed-off-by: NBjørn Mork <bjorn@mork.no>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

There are at least 4 implementations of netcat with the BSD-based
being the only one that has to be used without the -p switch to
specify the listening port.

Jan Engelhardt suggested to add an example for socat(1).
Signed-off-by: NDirk Gouders <gouders@et.bocholt.fh-gelsenkirchen.de>
Signed-off-by: NCong Wang <amwang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

napi->poll() needs IRQ enabled, so we have to re-enable IRQ before
calling it.

Cc: David Miller <davem@davemloft.net>
Signed-off-by: NCong Wang <amwang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Without this patch, I can't get netconsole logs remotely over
vlan. The reason is probably we don't handle vlan tags in either
netpoll tx or rx path.

I am not sure if I use these vlan functions correctly, at
least this patch works.

Cc: Benjamin LaHaise <bcrl@kvack.org>
Cc: Patrick McHardy <kaber@trash.net>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: NCong Wang <amwang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>