Merge remote-tracking branch 'l2-mtd/master'

Merge remote-tracking branch 'rdma/for-next'

Initial roundup of 4.3 merge window candidates

Merge remote-tracking branch 'bluetooth/master'

Merge remote-tracking branch 'net-next/master'

Merge remote-tracking branch 'slave-dma/next'

Merge remote-tracking branch 'dlm/next'

Merge remote-tracking branch 'thermal/next'

Merge remote-tracking branch 'idle/next'

sock: fix kernel doc error

The symbol '__sk_reclaim' is not present in the current tree. Apparently
'__sk_reclaim' was meant to be '__sk_mem_reclaim', so fix it with the
right symbol name for the kernel doc.

Signed-off-by: Jean Sacren <sakiwit@gmail.com>
Cc: Hideo Aoki <haoki@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge remote-tracking branch 'pm/linux-next'

Merge remote-tracking branch 'libata/for-next'

Merge remote-tracking branch 'kbuild/for-next'

Merge remote-tracking branch 'v4l-dvb/master'

Merge remote-tracking branch 'hwmon-staging/hwmon-next'

Merge branch 'jdelvare-hwmon/master'

Merge remote-tracking branch 'i2c/i2c/for-next'

Merge remote-tracking branch 'hid/for-next'

Merge remote-tracking branch 'pci/next'

net/mlx4_core: Fix unintialized variable used in error path

The uninitialized value name in mlx4_en_activate_cq was used in order
to print an error message. Fixing it by replacing it with cq->vector.

Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Carol L Soto <clsoto@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/mlx4_core: Capping number of requested MSIXs to MAX_MSIX

We currently manage IRQs in pool_bm which is a bit field
of MAX_MSIX bits. Thus, allocating more than MAX_MSIX
interrupts can't be managed in pool_bm.
Fixing this by capping number of requested MSIXs to
MAX_MSIX.

Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Carol L Soto <clsoto@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bridge: fdb: rearrange net_bridge_fdb_entry

While looking into fixing the local entries scalability issue I noticed
that the structure is badly arranged because vlan_id would fall in a
second cache line while keeping rcu which is used only when deleting
in the first, so re-arrange the structure and push rcu to the end so we
can get 16 bytes which can be used for other fields (by pushing rcu
fully in the second 64 byte chunk). With this change all the core
necessary information when doing fdb lookups will be available in a
single cache line.

pahole before (note vlan_id):
struct net_bridge_fdb_entry {
struct hlist_node          hlist;                /*     0    16 */
struct net_bridge_port *   dst;                  /*    16     8 */
struct callback_head       rcu;                  /*    24    16 */
long unsigned int          updated;              /*    40     8 */
long unsigned int          used;                 /*    48     8 */
mac_addr                   addr;                 /*    56     6 */
unsigned char              is_local:1;           /*    62: 7  1 */
unsigned char              is_static:1;          /*    62: 6  1 */
unsigned char              added_by_user:1;      /*    62: 5  1 */
unsigned char              added_by_external_learn:1; /*    62: 4  1 */

/* XXX 4 bits hole, try to pack */
/* XXX 1 byte hole, try to pack */

/* --- cacheline 1 boundary (64 bytes) --- */
__u16                      vlan_id;              /*    64     2 */

/* size: 72, cachelines: 2, members: 11 */
/* sum members: 65, holes: 1, sum holes: 1 */
/* bit holes: 1, sum bit holes: 4 bits */
/* padding: 6 */
/* last cacheline: 8 bytes */
}

pahole after (note vlan_id):
struct net_bridge_fdb_entry {
struct hlist_node          hlist;                /*     0    16 */
struct net_bridge_port *   dst;                  /*    16     8 */
long unsigned int          updated;              /*    24     8 */
long unsigned int          used;                 /*    32     8 */
mac_addr                   addr;                 /*    40     6 */
__u16                      vlan_id;              /*    46     2 */
unsigned char              is_local:1;           /*    48: 7  1 */
unsigned char              is_static:1;          /*    48: 6  1 */
unsigned char              added_by_user:1;      /*    48: 5  1 */
unsigned char              added_by_external_learn:1; /*    48: 4  1 */

/* XXX 4 bits hole, try to pack */
/* XXX 7 bytes hole, try to pack */

struct callback_head       rcu;                  /*    56    16 */
/* --- cacheline 1 boundary (64 bytes) was 8 bytes ago --- */

/* size: 72, cachelines: 2, members: 11 */
/* sum members: 65, holes: 1, sum holes: 7 */
/* bit holes: 1, sum bit holes: 4 bits */
/* last cacheline: 8 bytes */
}

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'ovs-v6-build-err'

Joe Stringer says:

====================
OPENVSWITCH && !NETFILTER build fix.

Fix issues reported by kbuild test robot:

All error/warnings (new ones prefixed by >>):

   net/openvswitch/actions.c: In function 'ovs_fragment':
>> net/openvswitch/actions.c:705:16: error: implicit declaration of
function 'nf_get_ipv6_ops' [-Werror=implicit-function-declaration]
      const struct nf_ipv6_ops *v6ops = nf_get_ipv6_ops();
                   ^
>> net/openvswitch/actions.c:705:37: warning: initialization makes
pointer from integer without a cast
      const struct nf_ipv6_ops *v6ops = nf_get_ipv6_ops();
                                        ^
>> net/openvswitch/actions.c:707:19: error: storage size of 'ovs_rt'
isn't known
      struct rt6_info ovs_rt;
                      ^
>> net/openvswitch/actions.c:724:8: error: dereferencing pointer to
incomplete type
      v6ops->fragment(skb->sk, skb, ovs_vport_output);
           ^
>> net/openvswitch/actions.c:707:19: warning: unused variable 'ovs_rt'
[-Wunused-variable]
      struct rt6_info ovs_rt;
                      ^
   cc1: some warnings being treated as errors
====================

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

openvswitch: Include ip6_fib.h.

kbuild test robot reports that certain configurations will not
automatically pick up on the "struct rt6_info" definition, so explicitly
include the header for this structure.

Fixes: 7f8a436 "openvswitch: Add conntrack action"
Signed-off-by: Joe Stringer <joestringer@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

netfilter: Define v6ops in !CONFIG_NETFILTER case.

When CONFIG_OPENVSWITCH is set, and CONFIG_NETFILTER is not set, the
openvswitch IPv6 fragmentation handling cannot refer to ipv6_ops because
it isn't defined. Add a dummy version to avoid #ifdefs in source files.

Fixes: 7f8a436 "openvswitch: Add conntrack action"
Signed-off-by: Joe Stringer <joestringer@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'mlxsw-small-updates'

Jiri Pirko says:

====================
mlxsw: small driver update

Ido Schimmel (2):
  mlxsw: Remove duplicate included header
  mlxsw: Make mailboxes 4KB aligned

Jiri Pirko (1):
  mlxsw: adjust transmit fail log message level in __mlxsw_emad_transmit
====================

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

mlxsw: Make mailboxes 4KB aligned

The HW-SW contract requires mailboxes passed to the firmware to be 4KB
aligned. Previously, these mailboxes were mapped using streaming DMA
routines, which do not guarantee the bus addresses to be 4KB aligned.
Under certain conditions this constraint was indeed violated and errors
were observed.

By using consistent DMA mapping routines together with a mailbox size of
4KB we are guaranteed not to violate the constraint.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlxsw: adjust transmit fail log message level in __mlxsw_emad_transmit

When transmit fails, it is an error, not a warning.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Elad Raz <eladr@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlxsw: Remove duplicate included header

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Elad Raz <eladr@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge remote-tracking branch 'vfs/for-next'

Merge remote-tracking branch 'xfs/for-next'

Merge branch 'rocker-master-change'

Jiri Pirko says:

====================
rocker: make master change handling nicer

Jiri Pirko (6):
  net: introduce change upper device notifier change info
  net: add netif_is_bridge_master helper
  net: add netif_is_ovs_master helper with IFF_OPENVSWITCH private flag
  net: kill long time unused bonding private flags
  rocker: use new helper to figure out master kind
  rocker: use change upper info
====================

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

rocker: use change upper info

Since now information about changed upper is passed along, benefit from
that and use this info directly.

This also fixes possible issues that could happen when non-master device
is added (current code does not distinguish between master and non-master
upper device).

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Acked-by: Scott Feldman <sfeldma@gmail.com
Signed-off-by: David S. Miller <davem@davemloft.net>

rocker: use new helper to figure out master kind

Looking at rtnl kind string is kind of ugly. So use new helpers to do
this in nicer way.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Acked-by: Scott Feldman <sfeldma@gmail.com
Signed-off-by: David S. Miller <davem@davemloft.net>

net: kill long time unused bonding private flags

We don't use them for years, just kill them now.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: add netif_is_ovs_master helper with IFF_OPENVSWITCH private flag

Add this helper so code can easily figure out if netdev is openswitch.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: add netif_is_bridge_master helper

Add this helper so code can easily figure out if netdev is a bridge.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: introduce change upper device notifier change info

Add info that is passed along with NETDEV_CHANGEUPPER event.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge remote-tracking branch 'ubifs/linux-next'

Merge remote-tracking branch 'v9fs/for-next'

Merge remote-tracking branch 'orangefs/for-next'

Merge remote-tracking branch 'nfsd/nfsd-next'

Merge remote-tracking branch 'nfs/linux-next'

Merge remote-tracking branch 'gfs2/for-next'

Merge remote-tracking branch 'fscache/fscache'

Merge remote-tracking branch 'f2fs/dev'

Merge remote-tracking branch 'ext4/dev'

Merge remote-tracking branch 'ext3/for_next'

Merge remote-tracking branch 'ecryptfs/next'

Merge remote-tracking branch 'cifs/for-next'

Merge remote-tracking branch 'ceph/master'

Merge remote-tracking branch 'btrfs/next'

Merge remote-tracking branch 'xtensa/for_next'

Merge remote-tracking branch 'uml/linux-next'

virtio-net: avoid unnecessary sg initialzation

Usually an skb does not have up to MAX_SKB_FRAGS frags. So no need to
initialize the unuse part of sg. This patch initialize the sg based on
the real number it will used:

- during xmit, it could be inferred from nr_frags and can_push.
- for small receive buffer, it will also be 2.

Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge remote-tracking branch 'tile/master'

Merge branch 'geneve-consolidation'

Pravin B Shelar says:

====================
Geneve: Add support for tunnel metadata mode

Following patches adds support for Geneve tunnel metadata
mode. OVS can make use of Geneve net-device with tunnel
metadata API from kernel.

This also allows us to consolidate Geneve implementation
from two kernel modules geneve_core and geneve to single
geneve module. geneve_core module was targeted to share
Geneve encap and decap code between Geneve netdevice and
OVS Geneve tunnel implementation, Since OVS no longer
needs these API, Geneve code can be consolidated into
single geneve module.

v3-v4:
- Drop NETIF_F_NETNS_LOCAL feature.
- Fix geneve device newlink check

v2-v3:
- make tunnel medata device and regular device mutually exclusive.
- Fix Kconfig dependency for Geneve.
- Fix dst-port netlink encoding.
- drop changelink patch.

v1-v2:
- Replaced per hash table tunnel pointer (metadata enabled) with flag.
- Added support for changelink.
- Improve geneve device route lookup with more parameters.
====================

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

geneve: Move device hash table to geneve socket.

This change simplifies Geneve Tunnel hash table management.

Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Reviewed-by: Jesse Gross <jesse@nicira.com>
Reviewed-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

geneve: Consolidate Geneve functionality in single module.

geneve_core module handles send and receive functionality.
This way OVS could use the Geneve API. Now with use of
tunnel meatadata mode OVS can directly use Geneve netdevice.
So there is no need for separate module for Geneve. Following
patch consolidates Geneve protocol processing in single module.

Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Reviewed-by: Jesse Gross <jesse@nicira.com>
Acked-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

openvswitch: Use Geneve device.

With help of tunnel metadata mode OVS can directly use
Geneve devices to implement Geneve tunnels.
This patch removes all of the OVS specific Geneve code
and make OVS use a Geneve net_device. Basic geneve vport
is still there to handle compatibility with current
userspace application.

Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Reviewed-by: Jesse Gross <jesse@nicira.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

geneve: Add support to collect tunnel metadata.

Following patch create new tunnel flag which enable
tunnel metadata collection on given device. These devices
can be used by tunnel metadata based routing or by OVS.
Geneve Consolidation patch get rid of collect_md_tun to
simplify tunnel lookup further.

Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Reviewed-by: Jesse Gross <jesse@nicira.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

geneve: Make dst-port configurable.

Add netlink interface to configure Geneve UDP port number.
So that user can configure it for a Gevene device.

Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Reviewed-by: Jesse Gross <jesse@nicira.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Acked-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tunnel: introduce udp_tun_rx_dst()

Introduce function udp_tun_rx_dst() to initialize tunnel dst on
receive path.

Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Reviewed-by: Jesse Gross <jesse@nicira.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

geneve: Use skb mark and protocol to lookup route.

On packet transmit path geneve need to lookup route. Following
patch improves route lookup using more parameters.

Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Reviewed-by: Jesse Gross <jesse@nicira.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Acked-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

geneve: Initialize ethernet address in device setup.

Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Reviewed-by: Jesse Gross <jesse@nicira.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Acked-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge remote-tracking branch 's390/features'

Merge remote-tracking branch 'powerpc/next'

Merge remote-tracking branch 'parisc-hd/for-next'

Merge remote-tracking branch 'nios2/for-next'

Merge remote-tracking branch 'mips/mips-for-linux-next'

bridge: Add netlink support for vlan_protocol attribute

This enables bridge vlan_protocol to be configured through netlink.

When CONFIG_BRIDGE_VLAN_FILTERING is disabled, kernel behaves the
same way as this feature is not implemented.

Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: Add new device ids under the Qlogic vendor

This adds support for 3 new PCI device combinations -
1077:16a1, 1077:16a4 and 1077:16ad.

Signed-off-by: Yuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge remote-tracking branch 'microblaze/next'

Merge remote-tracking branch 'metag/for-next'

Merge remote-tracking branch 'm68knommu/for-next'

Merge remote-tracking branch 'm68k/for-next'

Merge remote-tracking branch 'ia64/next'

Merge remote-tracking branch 'hexagon/linux-next'

Merge remote-tracking branch 'h8300/h8300-next'

Merge remote-tracking branch 'cris/for-next'

Merge remote-tracking branch 'c6x/for-linux-next'

Merge remote-tracking branch 'arm64/for-next/core'

Merge remote-tracking branch 'tegra/for-next'

Merge remote-tracking branch 'sunxi/sunxi/for-next'

Merge remote-tracking branch 'samsung-krzk/for-next'

Merge remote-tracking branch 'samsung/for-next'

Merge remote-tracking branch 'rockchip/for-next'

Merge remote-tracking branch 'renesas/next'

Merge remote-tracking branch 'qcom/qcom/for-next'

Merge remote-tracking branch 'omap-pending/for-next'

Merge remote-tracking branch 'omap/for-next'

Merge remote-tracking branch 'mvebu/for-next'

Merge remote-tracking branch 'keystone/next'

Merge remote-tracking branch 'imx-mxs/for-next'

Merge remote-tracking branch 'cortex-m/for-next'

Merge remote-tracking branch 'at91/at91-next'

Merge remote-tracking branch 'arm-soc/for-next'

Merge remote-tracking branch 'arm/for-next'

Merge remote-tracking branch 'arc/for-next'

Merge remote-tracking branch 'rr-fixes/fixes'

Merge remote-tracking branch 'input-current/for-linus'