If an attempt to wake up users of broadcast link is made when there is
no enough place in send queue than it may hang up inside the
tipc_sk_rcv() function since the loop breaks only after the wake up
queue becomes empty. This can lead to complete CPU stall with the
following message generated by RCU:

INFO: rcu_sched self-detected stall on CPU { 0}  (t=2101 jiffies
					g=54225 c=54224 q=11465)
Task dump for CPU 0:
tpch            R  running task        0 39949  39948 0x0000000a
 ffffffff818536c0 ffff88181fa037a0 ffffffff8106a4be 0000000000000000
 ffffffff818536c0 ffff88181fa037c0 ffffffff8106d8a8 ffff88181fa03800
 0000000000000001 ffff88181fa037f0 ffffffff81094a50 ffff88181fa15680
Call Trace:
 <IRQ>  [<ffffffff8106a4be>] sched_show_task+0xae/0x120
 [<ffffffff8106d8a8>] dump_cpu_task+0x38/0x40
 [<ffffffff81094a50>] rcu_dump_cpu_stacks+0x90/0xd0
 [<ffffffff81097c3b>] rcu_check_callbacks+0x3eb/0x6e0
 [<ffffffff8106e53f>] ? account_system_time+0x7f/0x170
 [<ffffffff81099e64>] update_process_times+0x34/0x60
 [<ffffffff810a84d1>] tick_sched_handle.isra.18+0x31/0x40
 [<ffffffff810a851c>] tick_sched_timer+0x3c/0x70
 [<ffffffff8109a43d>] __run_hrtimer.isra.34+0x3d/0xc0
 [<ffffffff8109aa95>] hrtimer_interrupt+0xc5/0x1e0
 [<ffffffff81030d52>] ? native_smp_send_reschedule+0x42/0x60
 [<ffffffff81032f04>] local_apic_timer_interrupt+0x34/0x60
 [<ffffffff810335bc>] smp_apic_timer_interrupt+0x3c/0x60
 [<ffffffff8165a3fb>] apic_timer_interrupt+0x6b/0x70
 [<ffffffff81659129>] ? _raw_spin_unlock_irqrestore+0x9/0x10
 [<ffffffff8107eb9f>] __wake_up_sync_key+0x4f/0x60
 [<ffffffffa313ddd1>] tipc_write_space+0x31/0x40 [tipc]
 [<ffffffffa313dadf>] filter_rcv+0x31f/0x520 [tipc]
 [<ffffffffa313d699>] ? tipc_sk_lookup+0xc9/0x110 [tipc]
 [<ffffffff81659259>] ? _raw_spin_lock_bh+0x19/0x30
 [<ffffffffa314122c>] tipc_sk_rcv+0x2dc/0x3e0 [tipc]
 [<ffffffffa312e7ff>] tipc_bclink_wakeup_users+0x2f/0x40 [tipc]
 [<ffffffffa313ce26>] tipc_node_unlock+0x186/0x190 [tipc]
 [<ffffffff81597c1c>] ? kfree_skb+0x2c/0x40
 [<ffffffffa313475c>] tipc_rcv+0x2ac/0x8c0 [tipc]
 [<ffffffffa312ff58>] tipc_l2_rcv_msg+0x38/0x50 [tipc]
 [<ffffffff815a76d3>] __netif_receive_skb_core+0x5a3/0x950
 [<ffffffff815a98d3>] __netif_receive_skb+0x13/0x60
 [<ffffffff815a993e>] netif_receive_skb_internal+0x1e/0x90
 [<ffffffff815aa138>] napi_gro_receive+0x78/0xa0
 [<ffffffffa07f93f4>] tg3_poll_work+0xc54/0xf40 [tg3]
 [<ffffffff81597c8c>] ? consume_skb+0x2c/0x40
 [<ffffffffa07f9721>] tg3_poll_msix+0x41/0x160 [tg3]
 [<ffffffff815ab0f2>] net_rx_action+0xe2/0x290
 [<ffffffff8104b92a>] __do_softirq+0xda/0x1f0
 [<ffffffff8104bc26>] irq_exit+0x76/0xa0
 [<ffffffff81004355>] do_IRQ+0x55/0xf0
 [<ffffffff8165a12b>] common_interrupt+0x6b/0x6b
 <EOI>

The issue occurs only when tipc_sk_rcv() is used to wake up postponed
senders:

	tipc_bclink_wakeup_users()
		// wakeupq - is a queue which consists of special
		// 		 messages with SOCK_WAKEUP type.
		tipc_sk_rcv(wakeupq)
			...
			while (skb_queue_len(inputq)) {
				filter_rcv(skb)
					// Here the type of message is checked
					// and if it is SOCK_WAKEUP then
					// it tries to wake up a sender.
					tipc_write_space(sk)
						wake_up_interruptible_sync_poll()
			}

After the sender thread is woke up it can gather control and perform
an attempt to send a message. But if there is no enough place in send
queue it will call link_schedule_user() function which puts a message
of type SOCK_WAKEUP to the wakeup queue and put the sender to sleep.
Thus the size of the queue actually is not changed and the while()
loop never exits.

The approach I proposed is to wake up only senders for which there is
enough place in send queue so the described issue can't occur.
Moreover the same approach is already used to wake up senders on
unicast links.

I have got into the issue on our product code but to reproduce the
issue I changed a benchmark test application (from
tipcutils/demos/benchmark) to perform the following scenario:
	1. Run 64 instances of test application (nodes). It can be done
	   on the one physical machine.
	2. Each application connects to all other using TIPC sockets in
	   RDM mode.
	3. When setup is done all nodes start simultaneously send
	   broadcast messages.
	4. Everything hangs up.

The issue is reproducible only when a congestion on broadcast link
occurs. For example, when there are only 8 nodes it works fine since
congestion doesn't occur. Send queue limit is 40 in my case (I use a
critical importance level) and when 64 nodes send a message at the
same moment a congestion occurs every time.
Signed-off-by: Dmitry S Kolmakov <kolmakov.dmitriy@huawei.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Barry Song <Baohua.Song@csr.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Remove the unnecessary switchdev.h include from br_netlink.c.
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Acked-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>

Since __vlan_del can return an error code, change its inner function
__vlan_vid_del to return an eventual error from switchdev_port_obj_del.
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Acked-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>

The comparison check between cur_hw_state and hw_state is currently
invalid because cur_hw_state is right shifted by G_MISTP_SHIFT, while
hw_state is not, so we end-up comparing bits 2:0 with bits 7:5, which is
going to cause an additional aging to occur. Fix this by not shifting
cur_hw_state while reading it, but instead, mask the value with the
appropriately shitfted bitmask.

The other problem with the fast-ageing process is that we did not set
the EN_AGE_DYNAMIC bit to request the ageing to occur for dynamically
learned MAC addresses. Finally, write back 0 to the FAST_AGE_CTRL
register to avoid leaving spurious bits sets from one operation to the
other.

Fixes: 12f460f2 ("net: dsa: bcm_sf2: add HW bridging support")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The function device_get_mac_address is trying different property names
in order to get the mac address. To check the return value, the variable
addr (which contain the buffer pass by the caller) will be re-used. This
means that if the previous property is not found, the next property will
be read using a NULL buffer.

Therefore it's only possible to retrieve the mac if node contains a
property "mac-address". Fix it by using a temporary buffer for the
return value.

This has been introduced by commit 4c96b7dc
"Add a matching set of device_ functions for determining mac/phy"
Signed-off-by: Julien Grall <julien.grall@citrix.com>
Cc: Jeremy Linton <jeremy.linton@arm.com>
Cc: David S. Miller <davem@davemloft.net>
Reviewed-by: Jeremy Linton <jeremy.linton@arm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The race may happen when a device (e.g. YOTA 4G LTE Modem) is
unplugged while the system is downloading a large file from the Net.

Hardware breakpoints and Kprobes with delays were used to confirm that
the race does actually happen.

The race is on skb_queue ('next' pointer) between usbnet_stop()
and rx_complete(), which, in turn, calls usbnet_bh().

Here is a part of the call stack with the code where the changes to the
queue happen. The line numbers are for the kernel 4.1.0:

*0 __skb_unlink (skbuff.h:1517)
    prev->next = next;
*1 defer_bh (usbnet.c:430)
    spin_lock_irqsave(&list->lock, flags);
    old_state = entry->state;
    entry->state = state;
    __skb_unlink(skb, list);
    spin_unlock(&list->lock);
    spin_lock(&dev->done.lock);
    __skb_queue_tail(&dev->done, skb);
    if (dev->done.qlen == 1)
        tasklet_schedule(&dev->bh);
    spin_unlock_irqrestore(&dev->done.lock, flags);
*2 rx_complete (usbnet.c:640)
    state = defer_bh(dev, skb, &dev->rxq, state);

At the same time, the following code repeatedly checks if the queue is
empty and reads these values concurrently with the above changes:

*0  usbnet_terminate_urbs (usbnet.c:765)
    /* maybe wait for deletions to finish. */
    while (!skb_queue_empty(&dev->rxq)
        && !skb_queue_empty(&dev->txq)
        && !skb_queue_empty(&dev->done)) {
            schedule_timeout(msecs_to_jiffies(UNLINK_TIMEOUT_MS));
            set_current_state(TASK_UNINTERRUPTIBLE);
            netif_dbg(dev, ifdown, dev->net,
                  "waited for %d urb completions\n", temp);
    }
*1  usbnet_stop (usbnet.c:806)
    if (!(info->flags & FLAG_AVOID_UNLINK_URBS))
        usbnet_terminate_urbs(dev);

As a result, it is possible, for example, that the skb is removed from
dev->rxq by __skb_unlink() before the check
"!skb_queue_empty(&dev->rxq)" in usbnet_terminate_urbs() is made. It is
also possible in this case that the skb is added to dev->done queue
after "!skb_queue_empty(&dev->done)" is checked. So
usbnet_terminate_urbs() may stop waiting and return while dev->done
queue still has an item.

Locking in defer_bh() and usbnet_terminate_urbs() was revisited to avoid
this race.
Signed-off-by: Eugene Shatokhin <eugene.shatokhin@rosalab.ru>
Reviewed-by: Bjørn Mork <bjorn@mork.no>
Acked-by: Oliver Neukum <oneukum@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c: In function ‘init_one’:
drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c:4579:8: warning: ‘chip’ may be used uninitialized in this function [-Wmaybe-uninitialized]
   chip |= CHELSIO_CHIP_CODE(CHELSIO_T4, pl_rev);
        ^
drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c:4571:11: note: ‘chip’ was declared here
  int ver, chip;
           ^

Fixes: d86bd29e ("cxgb4/cxgb4vf: read the correct bits of PL Who Am I register")
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Cc: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

I've noticed  that fixed_phy_register() ignores its 'irq' parameter instead of
passing it to fixed_phy_add(). Luckily, fixed_phy_register()  seems to  always
be  called with PHY_POLL  for 'irq'... :-)

Fixes: a7595121 ("net: phy: extend fixed driver with fixed_phy_register()")
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

There's no particular desire to have conntrack action support in Open
vSwitch as an independently configurable bit, rather just to ensure
there is not a hard dependency. This exposed option doesn't accurately
reflect the conntrack dependency when enabled, so simplify this by
removing the option. Compile the support if NF_CONNTRACK is enabled.

Fixes: 7f8a436e ("openvswitch: Add conntrack action")
Signed-off-by: Joe Stringer <joestringer@nicira.com>
Acked-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The Marvell 88E6171 switch is in the 88E6351 family, which supports
802.1Q, thus add support from the generic mv88e6xxx functions.
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge tag 'mac80211-for-davem-2015-09-04' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211

Johannes Berg says:

====================
For the first round of fixes, we have this:
 * fix for the sizeof() pointer type issue
 * a fix for regulatory getting into a restore loop
 * a fix for rfkill global 'all' state, it needs to be stored
   everywhere to apply correctly to new rfkill instances
 * properly refuse CQM RSSI when it cannot actually be used
 * protect HT TDLS traffic properly in non-HT networks
 * don't incorrectly advertise 80 MHz support when not allowed
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

If NO_DMA=y:

    ERROR: "dma_alloc_coherent" [drivers/net/ethernet/synopsys/dwc_eth_qos.ko] undefined!
    ERROR: "dma_free_coherent" [drivers/net/ethernet/synopsys/dwc_eth_qos.ko] undefined!
    ERROR: "dma_unmap_single" [drivers/net/ethernet/synopsys/dwc_eth_qos.ko] undefined!
    ERROR: "dma_map_page" [drivers/net/ethernet/synopsys/dwc_eth_qos.ko] undefined!
    ERROR: "dma_mapping_error" [drivers/net/ethernet/synopsys/dwc_eth_qos.ko] undefined!
    ERROR: "dma_map_single" [drivers/net/ethernet/synopsys/dwc_eth_qos.ko] undefined!
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Acked-by: Lars Persson <larper@axis.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/vxlan.c: In function ‘vxlan_udp_encap_recv’:
drivers/net/vxlan.c:1226: warning: ‘info’ may be used uninitialized in this function

While this warning is a false positive, it can be killed easily by
getting rid of the pointer intermediary and referring directly to the
ip_tunnel_info structure.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: Jiri Benc <jbenc@redhat.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/usb/lan78xx.c: In function ‘lan78xx_link_reset’:
net/usb/lan78xx.c:1107: warning: comparison is always false due to limited range of data type
net/usb/lan78xx.c:1111: warning: comparison is always false due to limited range of data type

Assigning return values that can be negative error codes to "u16"
variables makes them positive, ignoring the errors.  Hence use "int"
instead.

Drop the "unlikely"s (unlikely considered harmful) and propagate the
actual error values instead of overriding them to -EIO while we're at
it.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Only return a conn if the rds_conn_net(conn) matches the struct
net passed to rds_conn_lookup().

Fixes: 467fa153 ("RDS-TCP: Support multiple RDS-TCP listen endpoints,
       one per netns.")
Signed-off-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

If fec MDIO write method succeeds its return value comes from
call to pm_runtime_get_sync().
But pm_runtime_get_sync() can also return 1.

In case of Micrel KSZ9031 PHY this value will then
be returned along the call chain of phy_write() ->
ksz9031_extended_write() -> ksz9031_center_flp_timing() ->
ksz9031_config_init() -> phy_init_hw() -> phy_attach_direct() ->
phy_connect_direct().

Then phy_connect() will cast it into a pointer using ERR_PTR(),
which then fec_enet_mii_probe() will try to dereference
resulting in an oops.

Fix it by normalizing return value of pm_runtime_get_sync()
to be zero if positive in MDIO write method.

Fixes: 8fff755e ("net: fec: Ensure clocks are enabled while using mdio bus")
Signed-off-by: Maciej Szmigiero <mail@maciej.szmigiero.name>
Acked-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

switchdev_port_fdb_dump is used as .ndo_fdb_dump. Its return value is
idx, so we cannot return errval.

Fixes: 45d4122c ("switchdev: add support for fdb add/del/dump via switchdev_port_obj ops.")
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Acked-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Acked-by: Scott Feldman<sfeldma@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The be_cmd_rx_filter() routine sends a non-embedded cmd to the FW and used
a pre-allocated dma memory to hold the cmd payload. This worked fine when
this cmd was synchronous. This cmd was changed to asynchronous mode by the
commit 8af65c2f("make the RX_FILTER command asynchronous"). So now when
there are two quick invocations of this cmd, the 2nd request may end up
overwriting the first request, causing FW cmd corruption.

This patch reverts the offending commit and hence fixes the regression.

Fixes: 8af65c2f("be2net: make the RX_FILTER command asynchronous")
Signed-off-by: Sathya Perla <sathya.perla@avagotech.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Conflicts:
	include/net/netfilter/nf_conntrack.h

The conflict was an overlap between changing the type of the zone
argument to nf_ct_tmpl_alloc() whilst exporting nf_ct_tmpl_free.

Pablo Neira Ayuso says:

====================
Netfilter fixes for net

The following patchset contains Netfilter fixes for net, they are:

1) Oneliner to restore maps in nf_tables since we support addressing registers
   at 32 bits level.

2) Restore previous default behaviour in bridge netfilter when CONFIG_IPV6=n,
   oneliner from Bernhard Thaler.

3) Out of bound access in ipset hash:net* set types, reported by Dave Jones'
   KASan utility, patch from Jozsef Kadlecsik.

4) Fix ipset compilation with gcc 4.4.7 related to C99 initialization of
   unnamed unions, patch from Elad Raz.

5) Add a workaround to address inconsistent endianess in the res_id field of
   nfnetlink batch messages, reported by Florian Westphal.

6) Fix error paths of CT/synproxy since the conntrack template was moved to use
   kmalloc, patch from Daniel Borkmann.

All of them look good to me to reach 4.2, I can route this to -stable myself
too, just let me know what you prefer.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

dev_get_by_name() will increment the usage count if the matching device
is found. But we were not decrementing the count if we have got the
device and the device is non-active.
Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

In the IPv6 multicast routing code the mrt_lock was not being released
correctly in the MFC iterator, as a result adding or deleting a MIF would
cause a hang because the mrt_lock could not be acquired.

This fix is a copy of the code for the IPv4 case and ensures that the lock
is released correctly.
Signed-off-by: Richard Laing <richard.laing@alliedtelesis.co.nz>
Acked-by: Cong Wang <cwang@twopensource.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When beacon filtering is enabled the mac80211 software implementation
for RSSI CQM cannot work as beacons will not be available. Rather than
accepting such a configuration without proper effect, reject it.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

Currently if 80MHz channels are not allowed for use, the VHT IE is not
included in the probe request for an AP. This is not good enough if the
AP is configured with the wrong regulatory and supports VHT even where
prohibited or in TDLS scenarios.
Mark the ifmgd with the DISABLE_VHT flag for the misbehaving-AP case, and
unset VHT support from the peer-station entry for the TDLS case.
Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

restore_regulatory_settings() should restore alpha2
as computed in restore_alpha2(), not raw user_alpha2 to
behave as described in the comment just above that code.

This fixes endless loop of calling CRDA for "00" and "97"
countries after resume from suspend on my laptop.

Looks like others had the same problem, too:
http://ath9k-devel.ath9k.narkive.com/knY5W6St/ath9k-and-crda-messages-in-logs
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/899335
https://forum.porteus.org/viewtopic.php?t=4975&p=36436
https://forums.opensuse.org/showthread.php/483356-Authentication-Regulatory-Domain-issues-ath5k-12-2Signed-off-by: Maciej Szmigiero <mail@maciej.szmigiero.name>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

When switching the state of all RFKill switches of type all we need to
replicate the RFKILL_TYPE_ALL global state to all the other types global
state, so it is used to initialize persistent RFKill switches on
register.
Signed-off-by: João Paulo Rechi Vita <jprvita@endlessm.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

HT TDLS traffic should be protected in a non-HT BSS to avoid
collisions. Therefore, when TDLS peers join/leave, check if
protection is (now) needed and set the ht_operation_mode of
the virtual interface according to the HT capabilities of the
TDLS peer(s).

This works because a non-HT BSS connection never sets (or
otherwise uses) the ht_operation_mode; it just means that
drivers must be aware that this field applies to all HT
traffic for this virtual interface, not just the traffic
within the BSS. Document that.
Signed-off-by: Avri Altman <avri.altman@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

The rate_control_cap_mask() function takes a parameter mcs_mask, which
GCC will take to be u8 * even though it was declared with a fixed size.
This causes the following warning:

	net/mac80211/rate.c: In function 'rate_control_cap_mask':
	net/mac80211/rate.c:719:25: warning: 'sizeof' on array function parameter 'mcs_mask' will return size of 'u8 * {aka unsigned char *}' [-Wsizeof-array-argument]
	   for (i = 0; i < sizeof(mcs_mask); i++)
	                         ^
	net/mac80211/rate.c:684:10: note: declared here
	       u8 mcs_mask[IEEE80211_HT_MCS_MASK_LEN],
	          ^

This can be easily fixed by using the IEEE80211_HT_MCS_MASK_LEN directly
within the loop condition.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

Marcelo Ricardo Leitner says:

====================
couple of sctp fixes for 0ca50d12

These are two fixes for sctp after my patch on 0ca50d12 ("sctp: fix
src address selection if using secondary addresses")

The first, fix a dst leak on those it decided to skip.

The second, adds the fallback on src selection that Vlad had asked
about. Unfortunatelly a lot of ipvs setups relies on the old behavior
and I don't see a better fix for it.

Please consider both to -stable tree.
====================
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit 0ca50d12 added a restriction that the address must belong to
the output interface, so that sctp will use the right interface even
when using secondary addresses.

But it breaks IPVS setups, on which people is used to attach VIP
addresses to loopback interface on real servers. It's preferred to
attach to the interface actually in use, but it's a very common setup
and that used to work.

This patch then saves the first routing good result, even if it would be
going out through an interface that doesn't have that address. If no
better hit found, it's then used. This effectively restores the original
behavior if no better interface could be found.

Fixes: 0ca50d12 ("sctp: fix src address selection if using secondary addresses")
Signed-off-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit 0ca50d12 failed to release the reference to dst entries that
it decided to skip.

Fixes: 0ca50d12 ("sctp: fix src address selection if using secondary addresses")
Signed-off-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tse_poll() calls __napi_complete() with irq enabled.  This leads napi
poll_list corruption and may stop all napi drivers working.
Use napi_complete() instead of __napi_complete().
Signed-off-by: Atsushi Nemoto <nemoto@toshiba-tops.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>

Pull networking updates from David Miller:
 "Another merge window, another set of networking changes.  I've heard
  rumblings that the lightweight tunnels infrastructure has been voted
  networking change of the year.  But what do I know?

   1) Add conntrack support to openvswitch, from Joe Stringer.

   2) Initial support for VRF (Virtual Routing and Forwarding), which
      allows the segmentation of routing paths without using multiple
      devices.  There are some semantic kinks to work out still, but
      this is a reasonably strong foundation.  From David Ahern.

   3) Remove spinlock fro act_bpf fast path, from Alexei Starovoitov.

   4) Ignore route nexthops with a link down state in ipv6, just like
      ipv4.  From Andy Gospodarek.

   5) Remove spinlock from fast path of act_gact and act_mirred, from
      Eric Dumazet.

   6) Document the DSA layer, from Florian Fainelli.

   7) Add netconsole support to bcmgenet, systemport, and DSA.  Also
      from Florian Fainelli.

   8) Add Mellanox Switch Driver and core infrastructure, from Jiri
      Pirko.

   9) Add support for "light weight tunnels", which allow for
      encapsulation and decapsulation without bearing the overhead of a
      full blown netdevice.  From Thomas Graf, Jiri Benc, and a cast of
      others.

  10) Add Identifier Locator Addressing support for ipv6, from Tom
      Herbert.

  11) Support fragmented SKBs in iwlwifi, from Johannes Berg.

  12) Allow perf PMUs to be accessed from eBPF programs, from Kaixu Xia.

  13) Add BQL support to 3c59x driver, from Loganaden Velvindron.

  14) Stop using a zero TX queue length to mean that a device shouldn't
      have a qdisc attached, use an explicit flag instead.  From Phil
      Sutter.

  15) Use generic geneve netdevice infrastructure in openvswitch, from
      Pravin B Shelar.

  16) Add infrastructure to avoid re-forwarding a packet in software
      that was already forwarded by a hardware switch.  From Scott
      Feldman.

  17) Allow AF_PACKET fanout function to be implemented in a bpf
      program, from Willem de Bruijn"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1458 commits)
  netfilter: nf_conntrack: make nf_ct_zone_dflt built-in
  netfilter: nf_dup{4, 6}: fix build error when nf_conntrack disabled
  net: fec: clear receive interrupts before processing a packet
  ipv6: fix exthdrs offload registration in out_rt path
  xen-netback: add support for multicast control
  bgmac: Update fixed_phy_register()
  sock, diag: fix panic in sock_diag_put_filterinfo
  flow_dissector: Use 'const' where possible.
  flow_dissector: Fix function argument ordering dependency
  ixgbe: Resolve "initialized field overwritten" warnings
  ixgbe: Remove bimodal SR-IOV disabling
  ixgbe: Add support for reporting 2.5G link speed
  ixgbe: fix bounds checking in ixgbe_setup_tc for 82598
  ixgbe: support for ethtool set_rxfh
  ixgbe: Avoid needless PHY access on copper phys
  ixgbe: cleanup to use cached mask value
  ixgbe: Remove second instance of lan_id variable
  ixgbe: use kzalloc for allocating one thing
  flow: Move __get_hash_from_flowi{4,6} into flow_dissector.c
  ixgbe: Remove unused PCI bus types
  ...

Pull device mapper update from Mike Snitzer:

 - a couple small cleanups in dm-cache, dm-verity, persistent-data's
   dm-btree, and DM core.

 - a 4.1-stable fix for dm-cache that fixes the leaking of deferred bio
   prison cells

 - a 4.2-stable fix that adds feature reporting for the dm-stats
   features added in 4.2

 - improve DM-snapshot to not invalidate the on-disk snapshot if
   snapshot device write overflow occurs; but a write overflow triggered
   through the origin device will still invalidate the snapshot.

 - optimize DM-thinp's async discard submission a bit now that late bio
   splitting has been included in block core.

 - switch DM-cache's SMQ policy lock from using a mutex to a spinlock;
   improves performance on very low latency devices (eg. NVMe SSD).

 - document DM RAID 4/5/6's discard support

[ I did not pull the slab changes, which weren't appropriate for this
  tree, and weren't obviously the right thing to do anyway.  At the very
  least they need some discussion and explanation before getting merged.

  Because not pulling the actual tagged commit but doing a partial pull
  instead, this merge commit thus also obviously is missing the git
  signature from the original tag ]

* tag 'dm-4.3-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm:
  dm cache: fix use after freeing migrations
  dm cache: small cleanups related to deferred prison cell cleanup
  dm cache: fix leaking of deferred bio prison cells
  dm raid: document RAID 4/5/6 discard support
  dm stats: report precise_timestamps and histogram in @stats_list output
  dm thin: optimize async discard submission
  dm snapshot: don't invalidate on-disk image on snapshot write overflow
  dm: remove unlikely() before IS_ERR()
  dm: do not override error code returned from dm_get_device()
  dm: test return value for DM_MAPIO_SUBMITTED
  dm verity: remove unused mempool
  dm cache: move wake_waker() from free_migrations() to where it is needed
  dm btree remove: remove unused function get_nr_entries()
  dm btree: remove unused "dm_block_t root" parameter in btree_split_sibling()
  dm cache policy smq: change the mutex to a spinlock

Fengguang reported, that some randconfig generated the following linker
issue with nf_ct_zone_dflt object involved:

  [...]
  CC      init/version.o
  LD      init/built-in.o
  net/built-in.o: In function `ipv4_conntrack_defrag':
  nf_defrag_ipv4.c:(.text+0x93e95): undefined reference to `nf_ct_zone_dflt'
  net/built-in.o: In function `ipv6_defrag':
  nf_defrag_ipv6_hooks.c:(.text+0xe3ffe): undefined reference to `nf_ct_zone_dflt'
  make: *** [vmlinux] Error 1

Given that configurations exist where we have a built-in part, which is
accessing nf_ct_zone_dflt such as the two handlers nf_ct_defrag_user()
and nf_ct6_defrag_user(), and a part that configures nf_conntrack as a
module, we must move nf_ct_zone_dflt into a fixed, guaranteed built-in
area when netfilter is configured in general.

Therefore, split the more generic parts into a common header under
include/linux/netfilter/ and move nf_ct_zone_dflt into the built-in
section that already holds parts related to CONFIG_NF_CONNTRACK in the
netfilter core. This fixes the issue on my side.

Fixes: 308ac914 ("netfilter: nf_conntrack: push zone object into functions")
Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

While testing various Kconfig options on another issue, I found that
the following one triggers as well on allmodconfig and nf_conntrack
disabled:

  net/ipv4/netfilter/nf_dup_ipv4.c: In function ‘nf_dup_ipv4’:
  net/ipv4/netfilter/nf_dup_ipv4.c:72:20: error: ‘nf_skb_duplicated’ undeclared (first use in this function)
    if (this_cpu_read(nf_skb_duplicated))
  [...]
  net/ipv6/netfilter/nf_dup_ipv6.c: In function ‘nf_dup_ipv6’:
  net/ipv6/netfilter/nf_dup_ipv6.c:66:20: error: ‘nf_skb_duplicated’ undeclared (first use in this function)
    if (this_cpu_read(nf_skb_duplicated))

Fix it by including directly the header where it is defined.

Fixes: bbde9fc1 ("netfilter: factor out packet duplication for IPv4/IPv6")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

The patch just to re-submit the patch "db3421c1" because the
patch "4d494cdc" remove the change.

Clear any pending receive interrupt before we process a pending packet.
This helps to avoid any spurious interrupts being raised after we have
fully cleaned the receive ring, while still allowing an interrupt to be
raised if we receive another packet.

The position of this is critical: we must do this prior to reading the
next packet status to avoid potentially dropping an interrupt when a
packet is still pending.
Acked-by: Fugang Duan <B38611@freescale.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

We previously register IPPROTO_ROUTING offload under inet6_add_offload(),
but in error path, we try to unregister it with inet_del_offload(). This
doesn't seem correct, it should actually be inet6_del_offload(), also
ipv6_exthdrs_offload_exit() from that commit seems rather incorrect (it
also uses rthdr_offload twice), but it got removed entirely later on.

Fixes: 3336288a ("ipv6: Switch to using new offload infrastructure.")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

Pull SG updates from Jens Axboe:
 "This contains a set of scatter-gather related changes/fixes for 4.3:

   - Add support for limited chaining of sg tables even for
     architectures that do not set ARCH_HAS_SG_CHAIN.  From Christoph.

   - Add sg chain support to target_rd.  From Christoph.

   - Fixup open coded sg->page_link in crypto/omap-sham.  From
     Christoph.

   - Fixup open coded crypto ->page_link manipulation.  From Dan.

   - Also from Dan, automated fixup of manual sg_unmark_end()
     manipulations.

   - Also from Dan, automated fixup of open coded sg_phys()
     implementations.

   - From Robert Jarzmik, addition of an sg table splitting helper that
     drivers can use"

* 'for-4.3/sg' of git://git.kernel.dk/linux-block:
  lib: scatterlist: add sg splitting function
  scatterlist: use sg_phys()
  crypto/omap-sham: remove an open coded access to ->page_link
  scatterlist: remove open coded sg_unmark_end instances
  crypto: replace scatterwalk_sg_chain with sg_chain
  target/rd: always chain S/G list
  scatterlist: allow limited chaining without ARCH_HAS_SG_CHAIN

Pull block driver updates from Jens Axboe:
 "On top of the 4.3 core block IO changes, here are the driver related
  changes for 4.3.  Basically just NVMe and nbd this time around:

   - NVMe:
      - PRACT PI improvement from Alok Pandey.
      - Cleanups and improvements on submission queue doorbell and
        writing, using CMB if available.  From Jon Derrick.
      - From Keith, support for setting queue maximum segments, and
        reset support.
      - Also from Jon, fixup of u64 division issue on 32-bit archs and
        wiring up of the reset support through and ioctl.
      - Two small cleanups from Matias and Sunad

  - Various code cleanups and fixes from Markus Pargmann"

* 'for-4.3/drivers' of git://git.kernel.dk/linux-block:
  NVMe: Using PRACT bit to generate and verify PI by controller
  NVMe:Remove unreachable code in nvme_abort_req
  NVMe: Add nvme subsystem reset IOCTL
  NVMe: Add nvme subsystem reset support
  NVMe: removed unused nn var from nvme_dev_add
  NVMe: Set queue max segments
  nbd: flags is a u32 variable
  nbd: Rename functions for clearness of recv/send path
  nbd: Change 'disconnect' to be boolean
  nbd: Add debugfs entries
  nbd: Remove variable 'pid'
  nbd: Move clear queue debug message
  nbd: Remove 'harderror' and propagate error properly
  nbd: restructure sock_shutdown
  nbd: sock_shutdown, remove conditional lock
  nbd: Fix timeout detection
  nvme: Fixes u64 division which breaks i386 builds
  NVMe: Use CMB for the IO SQes if available
  NVMe: Unify SQ entry writing and doorbell ringing