Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-next into for-davem

Conflicts:
drivers/net/ethernet/broadcom/Kconfig

mwifiex: fix build error when CONFIG_PM is not set

config: make ARCH=m68k allmodconfig

All error/warnings:

   drivers/net/wireless/mwifiex/cfg80211.c: In function
       'mwifiex_fill_coalesce_rule_info':
>> drivers/net/wireless/mwifiex/cfg80211.c:2493:3: error: implicit
       declaration of function 'mwifiex_is_pattern_supported'
       [-Werror=implicit-function-declaration]
   drivers/net/wireless/mwifiex/cfg80211.c: At top level:
   drivers/net/wireless/mwifiex/cfg80211.c:2537:12: warning:
       'mwifiex_cfg80211_set_coalesce' defined but not used
       [-Wunused-function]
   cc1: some warnings being treated as errors

Reported-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

af_unix: fix bug on large send()

commit e370a723632 ("af_unix: improve STREAM behavior with fragmented
memory") added a bug on large send() because the
skb_copy_datagram_from_iovec() call always start from the beginning
of iovec.

We must instead use the @sent variable to properly skip the
already processed part.

Reported-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: Add MOXA ART SoCs ethernet driver

The MOXA UC-711X hardware(s) has an ethernet controller that seem
to be developed internally. The IC used is "RTL8201CP".

Since there is no public documentation, this driver is mostly the
one published by MOXA that has been heavily cleaned up / ported
from linux 2.6.9.

Signed-off-by: Jonas Jensen <jonas.jensen@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: attempt high order allocations in sock_alloc_send_pskb()

Adding paged frags skbs to af_unix sockets introduced a performance
regression on large sends because of additional page allocations, even
if each skb could carry at least 100% more payload than before.

We can instruct sock_alloc_send_pskb() to attempt high order
allocations.

Most of the time, it does a single page allocation instead of 8.

I added an additional parameter to sock_alloc_send_pskb() to
let other users to opt-in for this new feature on followup patches.

Tested:

Before patch :

$ netperf -t STREAM_STREAM
STREAM STREAM TEST
Recv   Send    Send
Socket Socket  Message  Elapsed
Size   Size    Size     Time     Throughput
bytes  bytes   bytes    secs.    10^6bits/sec

 2304  212992  212992    10.00    46861.15

After patch :

$ netperf -t STREAM_STREAM
STREAM STREAM TEST
Recv   Send    Send
Socket Socket  Message  Elapsed
Size   Size    Size     Time     Throughput
bytes  bytes   bytes    secs.    10^6bits/sec

 2304  212992  212992    10.00    57981.11

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

af_unix: improve STREAM behavior with fragmented memory

unix_stream_sendmsg() currently uses order-2 allocations,
and we had numerous reports this can fail.

The __GFP_REPEAT flag present in sock_alloc_send_pskb() is
not helping.

This patch extends the work done in commit eb6a24816b247c
("af_unix: reduce high order page allocations) for
datagram sockets.

This opens the possibility of zero copy IO (splice() and
friends)

The trick is to not use skb_pull() anymore in recvmsg() path,
and instead add a @consumed field in UNIXCB() to track amount
of already read payload in the skb.

There is a performance regression for large sends
because of extra page allocations that will be addressed
in a follow-up patch, allowing sock_alloc_send_pskb()
to attempt high order page allocations.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tcp: add server ip to encrypt cookie in fast open

Encrypt the cookie with both server and client IPv4 addresses,
such that multi-homed server will grant different cookies
based on both the source and destination IPs. No client change
is needed since cookie is opaque to the client.

Signed-off-by: Yuchung Cheng <ycheng@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

pptp: fix sparse pointer warning

callid_sock array is referenced via rcu_dereference and
sparse rcu checks complains about address space mismatch.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: clean up unnecessary MSI/MSI-X capability find

PCI core will initialize device MSI/MSI-X capability in
pci_msi_init_pci_dev(). So device driver should use
pci_dev->msi_cap/msix_cap to determine whether the device
support MSI/MSI-X instead of using
pci_find_capability(pci_dev, PCI_CAP_ID_MSI/MSIX).
Access to PCIe device config space again will consume more time.

Signed-off-by: Yijing Wang <wangyijing@huawei.com>
Cc: Nithin Nayak Sujir <nsujir@broadcom.com>
Cc: Michael Chan <mchan@broadcom.com>
Cc: netdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>

netxen: clean up unnecessary MSI/MSI-X capability find

PCI core will initialize device MSI/MSI-X capability in
pci_msi_init_pci_dev(). So device driver should use
pci_dev->msi_cap/msix_cap to determine whether the device
support MSI/MSI-X instead of using
pci_find_capability(pci_dev, PCI_CAP_ID_MSI/MSIX).
Access to PCIe device config space again will consume more time.

Signed-off-by: Yijing Wang <wangyijing@huawei.com>
Cc: Manish Chopra <manish.chopra@qlogic.com>
Cc: Sony Chacko <sony.chacko@qlogic.com>
Cc: Rajesh Borundia <rajesh.borundia@qlogic.com>
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>

myri10ge: clean up unnecessary MSI/MSI-X capability find

PCI core will initialize device MSI/MSI-X capability in
pci_msi_init_pci_dev(). So device driver should use
pci_dev->msi_cap/msix_cap to determine whether the device
support MSI/MSI-X instead of using
pci_find_capability(pci_dev, PCI_CAP_ID_MSI/MSIX).
Access to PCIe device config space again will consume more time.

Signed-off-by: Yijing Wang <wangyijing@huawei.com>
Cc: Andrew Gallatin <gallatin@myri.com>
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: clean up unnecessary MSI/MSI-X capability find

PCI core will initialize device MSI/MSI-X capability in
pci_msi_init_pci_dev(). So device driver should use
pci_dev->msi_cap/msix_cap to determine whether the device
support MSI/MSI-X instead of using
pci_find_capability(pci_dev, PCI_CAP_ID_MSI/MSIX).
Access to PCIe device config space again will consume more time.

Signed-off-by: Yijing Wang <wangyijing@huawei.com>
Cc: Eilon Greenstein <eilong@broadcom.com>
Cc: netdev@vger.kernel.org
Acked-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2: clean up unnecessary MSI/MSI-X capability find

PCI core will initialize device MSI/MSI-X capability in
pci_msi_init_pci_dev(). So device driver should use
pci_dev->msi_cap/msix_cap to determine whether the device
support MSI/MSI-X instead of using
pci_find_capability(pci_dev, PCI_CAP_ID_MSI/MSIX).
Access to PCIe device config space again will consume more time.

Signed-off-by: Yijing Wang <wangyijing@huawei.com>
Cc: Michael Chan <mchan@broadcom.com>
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>

net: via-rhine: Fix incorrect placement of __initdata

__initdata should be placed between the variable name and equal
sign for the variable to be placed in the intended section.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: wan: sbni: Fix incorrect placement of __initdata

__initdata should be placed between the variable name and equal
sign for the variable to be placed in the intended section.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Revert "net: sctp: convert sctp_checksum_disable module param into sctp sysctl"

This reverts commit cda5f98e36576596b9230483ec52bff3cc97eb21.

As per Vlad's request.

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

ath9k: Run the LNA combining algorithm properly

The LNA combining algorithm has to be run for cards
that support the required diversity features, make
sure that that correct conditions are met before
enabing this algorithm.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: Fix BTCOEX usage for RX diversity

BTCOEX has to be *disabled* for WLAN RX diversity to
work on combo cards.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: use software queueing for multicast traffic

Create a per-vif dummy node entry for keeping the multicast software
queues. This helps in setups with a lot of mulitcast traffic that could
otherwise potentially drown out unicast traffic to stations.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: improve tx scheduling fairness

Instead of trying to schedule the same TID multiple times in a loop,
iterate over other TIDs/stations first.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: use software queues for un-aggregated data packets

This is a first step for improving fairness between legacy and 802.11n
traffic, and it should also improve reliability of resets and channel
changes by keeping the hardware queue depth very short.

When an aggregation session is torn down, all packets in the retry queue
will be removed from the BAW and freed.

For all subframes that have not been transmitted yet, the A-MPDU flag
will be cleared, and a sequence number allocated. This ensures that the
next A-MPDU session will get the correct initial sequence number.
This happens both on aggregation session start and stop.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: always clear ps filter bit on new assoc

Otherwise in some cases, EAPOL frames might be filtered during the
initial handshake, causing delays and assoc failures.

Cc: stable@vger.kernel.org
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: fix clearing expired A-MPDU subframes in tx completion

When the tid aggregation state has been marked as inactive, free
completed tx packets immediately. When a new aggregation session has not
been initialized yet, the BAW checks do not recognize it as expired.

Might fix potential stalls in setting up a new aggregation session.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: prepare queueing code for handling unaggregated traffic

- Allow ath_tx_get_tid_subframe to return non-AMPDU subframes.
- Reset the tid paused state on aggregation stop
- Initialize software queues even when HT is not supported

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: fix block ack window tracking check

When a packet has been tracked as part of the BlockAck window and added
to the hardware queue, it can end up back in the TID queue again with
fi->retries still set to 0 (e.g. if the frame was filtered). Keep an
extra bit for the BAW tracking status to fix this corner case.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: simplify ath_tx_form_aggr

The check for ATH_AMPDU_SUBFRAME_DEFAULT is unnecessary, since it's set
to half the maximum BlockAck Window size, which is already the maximum
value that h_baw could possibly have. Also remove unnecessary variables.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: add CAB queue info to debugfs

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: add function for getting the tx tid for a packet

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: split tid retry packets into a separate queue

Improves packet retry order and helps with further tx queueing
improvements.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: add utility functions for accessing tid queues

Useful for further fixes / cleanups

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

brcmfmac: use CFG80211_TESTMODE_CMD

This is essentially the same, but written shorter.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Acked-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

net: wireless: rt2x00: Staticize rt2x00queue_pause_queue_nocheck()

rt2x00queue_pause_queue_nocheck()is used only in this file.
Fix the following sparse warning:

drivers/net/wireless/rt2x00/rt2x00queue.c:939:6: warning: symbol 'rt2x00queue_pause_queue_nocheck' was not declared. Should it be
static?

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: Fix build failure

Make sure that CONFIG_ATH9K_BTCOEX_SUPPORT is used for
the WLAN/BT RX diversity hooks.

Reported by the kernel build testing backend.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

mwifiex: add packet coalesce support

Coalesce filters are configured in firmware based on settings
received from cfg80211.

Packet type which is required by firmware is determined based on
provided patterns in a rule:

Unicast: if pattern '01' with offset 0 is found
Multicast: if pattern '33:33' or '01:00:5e' with offset 0 is found
Broadcast: if pattern 'ff:ff:ff:ff' with offset 0 is found

Some example coalesce configuration files:

1) Coalesce Rx data packets from 192.168.0.88
mac address of our device is 00:50:43:21:53:7A
Source IP address offset comes out as 52 after following
calculations:
    32 bytes of HW 802.11 header + 8 bytes LLC +
    12 bytes in IPV4 header till source IP address
Destination mac is at offset 6 in HW header.

delay=100
condition=1
patterns=01,6+00:50:43:22,10+53:7A,52+c0:a8:00:58

2) Coalesce all broadcast and multicast packets(Multiple packet
types are not allowed in a single rule. Hence created separate
rules)

delay=400
condition=1
patterns=33:33
delay=400
condition=1
patterns=ff:ff:ff:ff

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

mwifiex: increase max supported pattern offset

The offset number is increased to accomodate requests from
user to match more fields in a Rx packet.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

mwifiex: modify mwifiex_is_pattern_supported() routine

It is modified so that it can be reused for coalesce feature.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

mwifiex: rename mef macros

Their names were generic. We need to define similar macros
for coalesce feature. Hence they are renamed here.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

Merge branch 'for-john' of git://git./linux/kernel/git/jberg/mac80211-next

Merge branch 'for-john' of git://git./linux/kernel/git/iwlwifi/iwlwifi-next

Merge branch 'master' of git://git./linux/kernel/git/linville/wireless

net: sctp: trivial: update bug report in header comment

With the restructuring of the lksctp.org site, we only allow bug
reports through the SCTP mailing list linux-sctp@vger.kernel.org,
not via SF, as SF is only used for web hosting and nothing more.
While at it, also remove the obvious statement that bugs will be
fixed and incooperated into the kernel.

Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: sctp: convert sctp_checksum_disable module param into sctp sysctl

Get rid of the last module parameter for SCTP and make this
configurable via sysctl for SCTP like all the rest of SCTP's
configuration knobs.

Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: igmp: Allow user-space configuration of igmp unsolicited report interval

Adds the new procfs knobs:

    /proc/sys/net/ipv4/conf/*/igmpv2_unsolicited_report_interval
    /proc/sys/net/ipv4/conf/*/igmpv3_unsolicited_report_interval

Which will allow userspace configuration of the IGMP unsolicited report
interval (see below) in milliseconds.  The defaults are 10000ms for IGMPv2
and 1000ms for IGMPv3 in accordance with RFC2236 and RFC3376.

Background:

If an IGMP join packet is lost you will not receive data sent to the
multicast group so if no data arrives from that multicast group in a
period of time after the IGMP join a second IGMP join will be sent.  The
delay between joins is the "IGMP Unsolicited Report Interval".

Prior to this patch this value was hard coded in the kernel to 10s for
IGMPv2 and 1s for IGMPv3.  10s is unsuitable for some use-cases, such as
IPTV as it can cause channel change to be slow in the presence of packet
loss.

This patch allows the value to be overridden from userspace for both
IGMPv2 and IGMPv3 such that it can be tuned accoding to the network.

Tested with Wireshark and a simple program to join a (non-existent)
multicast group.  The distribution of timings for the second join differ
based upon setting the procfs knobs.

igmpvX_unsolicited_report_interval is intended to follow the pattern
established by force_igmp_version, and while a procfs entry has been added
a corresponding sysctl knob has not as it is my understanding that sysctl
is deprecated[1].

[1]: http://lwn.net/Articles/247243/

Signed-off-by: William Manley <william.manley@youview.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Benjamin LaHaise <bcrl@kvack.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: igmp: Don't flush routing cache when force_igmp_version is modified

The procfs knob /proc/sys/net/ipv4/conf/*/force_igmp_version allows the
IGMP protocol version to use to be explicitly set.  As a side effect this
caused the routing cache to be flushed as it was declared as a
DEVINET_SYSCTL_FLUSHING_ENTRY.  Flushing is unnecessary and this patch
makes it so flushing does not occur.

Requested by Hannes Frederic Sowa as he was reviewing other patches
adding procfs entries.

Suggested-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: William Manley <william.manley@youview.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Benjamin LaHaise <bcrl@kvack.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: igmp: Reduce Unsolicited report interval to 1s when using IGMPv3

If an IGMP join packet is lost you will not receive data sent to the
multicast group so if no data arrives from that multicast group in a
period of time after the IGMP join a second IGMP join will be sent.  The
delay between joins is the "IGMP Unsolicited Report Interval".

Previously this value was hard coded to be chosen randomly between 0-10s.
This can be too long for some use-cases, such as IPTV as it can cause
channel change to be slow in the presence of packet loss.

The value 10s has come from IGMPv2 RFC2236, which was reduced to 1s in
IGMPv3 RFC3376.  This patch makes the kernel use the 1s value from the
later RFC if we are operating in IGMPv3 mode.  IGMPv2 behaviour is
unaffected.

Tested with Wireshark and a simple program to join a (non-existent)
multicast group.  The distribution of timings for the second join differ
based upon setting /proc/sys/net/ipv4/conf/eth0/force_igmp_version.

Signed-off-by: William Manley <william.manley@youview.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Benjamin LaHaise <bcrl@kvack.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

bonding: unwind on bond_add_vlan failure

In case of bond_add_vlan() failure currently we'll have the vlan's
refcnt bumped up in all slaves, but it will never go down because it
failed to get added to the bond, so properly unwind the added vlan if
bond_add_vlan fails.

Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
Acked-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bonding: change the bond's vlan syncing functions with the standard ones

Now we have vlan_vids_add/del_by_dev() which serve the same purpose as
bond's bond_add/del_vlans_on_slave() with the good side effect of
reverting the changes if one of the additions fails.
There's only 1 change in the behaviour of enslave: if adding of the
vlans to the slave fails, we'll fail the enslaving because otherwise we
might delete some vlan that wasn't added by the bonding.
The only way this may happen is with ENOMEM currently, so we're in trouble
anyway.

Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
Acked-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: add SNMP counters tracking incoming ECN bits

With GRO/LRO processing, there is a problem because Ip[6]InReceives SNMP
counters do not count the number of frames, but number of aggregated
segments.

Its probably too late to change this now.

This patch adds four new counters, tracking number of frames, regardless
of LRO/GRO, and on a per ECN status basis, for IPv4 and IPv6.

Ip[6]NoECTPkts : Number of packets received with NOECT
Ip[6]ECT1Pkts  : Number of packets received with ECT(1)
Ip[6]ECT0Pkts  : Number of packets received with ECT(0)
Ip[6]CEPkts    : Number of packets received with Congestion Experienced

lph37:~# nstat | egrep "Pkts|InReceive"
IpInReceives                    1634137            0.0
Ip6InReceives                   3714107            0.0
Ip6InNoECTPkts                  19205              0.0
Ip6InECT0Pkts                   52651828           0.0
IpExtInNoECTPkts                33630              0.0
IpExtInECT0Pkts                 15581379           0.0
IpExtInCEPkts                   6                  0.0

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

packet: Revert recent header parsing changes.

This reverts commits:

0f75b09c798ed00c30d7d5551b896be883bc2aeb
cbd89acb9eb257ed3b2be867142583fdcf7fdc5b
c483e02614551e44ced3fe6eedda8e36d3277ccc

Amongst other things, it's modifies the SKB header
to pull the ethernet headers off via eth_type_trans()
on the output path which is bogus.

It's causing serious regressions for people.

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

net: phy: mdio: add missing __iomem annotation

Added missing __iomem annotation in order to fix the following
sparse warnings:

drivers/net/phy/mdio-mux-mmioreg.c:51:27: warning: incorrect type in initializer (different address spaces)
drivers/net/phy/mdio-mux-mmioreg.c:51:27:    expected void *p
drivers/net/phy/mdio-mux-mmioreg.c:51:27:    got void [noderef] <asn:2>*
drivers/net/phy/mdio-mux-mmioreg.c:57:21: warning: incorrect type in argument 1 (different address spaces)
drivers/net/phy/mdio-mux-mmioreg.c:57:21:    expected void const volatile [noderef] <asn:2>*addr
drivers/net/phy/mdio-mux-mmioreg.c:57:21:    got void *p
drivers/net/phy/mdio-mux-mmioreg.c:60:25: warning: incorrect type in argument 2 (different address spaces)
drivers/net/phy/mdio-mux-mmioreg.c:60:25:    expected void volatile [noderef] <asn:2>*addr
drivers/net/phy/mdio-mux-mmioreg.c:60:25:    got void *p
drivers/net/phy/mdio-mux-mmioreg.c:64:25: warning: incorrect type in argument 1 (different address spaces)
drivers/net/phy/mdio-mux-mmioreg.c:64:25:    expected void volatile [noderef] <asn:2>*addr
drivers/net/phy/mdio-mux-mmioreg.c:64:25:    got void *p

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: micrel: Staticize ksz8873mll_read_status()

ksz8873mll_read_status() is used only in this file.
Fix the following sparse warning:

drivers/net/phy/micrel.c:147:5: warning: symbol 'ksz8873mll_read_status' was not declared. Should it be static?

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: use skb_copy_datagram_from_iovec() in zerocopy_sg_from_iovec()

Use skb_copy_datagram_from_iovec() to avoid code duplication and make it easy to
be read. Also we can do the skipping inside the zero-copy loop.

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

net: use release_pages() in zerocopy_sg_from_iovec()

To reduce the duplicated codes.

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

net: remove the useless comment in zerocopy_sg_from_iovec()

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

net: use skb_fill_page_desc() in zerocopy_sg_from_iovec()

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

net: move zerocopy_sg_from_iovec() to net/core/datagram.c

To let it be reused and reduce code duplication. Also document this function.

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

net: move iov_pages() to net/core/iovec.c

To let it be reused and reduce code duplication.

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

ip_tunnel: embed hash list head

The IP tunnel hash heads can be embedded in the per-net structure
since it is a fixed size. Reduce the size so that the total structure
fits in a page size. The original size was overly large, even NETDEV_HASHBITS
is only 8 bits!

Also, add some white space for readability.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Pravin B Shelar <pshelar@nicira.com>.
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2: Update version to 2.2.4

and update copyright year.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2: Add pci shutdown handler.

WoL and power state changes will now be done in the shutdown handler.
open/close/ethtool will no longer change the power state.  NVRAM
operations can now be permitted whether the device is up or down.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2: Use SIMPLE_DEV_PM_OPS.

This simplifies the suspend/resume code.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2: Refactor WoL setup into a separate function.

Separate MAC and PHY WoL setup code into a separate function.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2: Use kernel APIs for WoL and power state changes.

Simple API changes with no functional changes.

Signed-off-by: Michael Chan <mchan@broadocm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2: Handle error condition in ->slot_reset()

by closing the device if necessary.  Otherwise, since NAPI state is
already disabled, a subsequent close will hang the system.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: update driver version

Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: Initialize "status" in be_cmd_get_die_temperature()

Uninitialized value was being returned in the non-failure case.
Signed-off-by: Vasundhara Volam <vasundhara.volam@emulex.com>
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: fixup log msgs for async events

Log the event type for unknown async events

Signed-off-by: Vasundhara Volam <vasundhara.volam@emulex.com>
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: Fix displaying supported speeds for BE2

The BE2 FW GET_PHY_DETAILS cmd does not return fixed speeds supported.

Signed-off-by: Vasundhara Volam <vasundhara.volam@emulex.com>
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: don't limit max MAC and VLAN counts

For SH-R and Lancer-R, use the FW supported values for Max unicast MACs,
Max VLANs and Max multicast MACs.

Signed-off-by: Vasundhara Volam <vasundhara.volam@emulex.com>
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: Do not call get_die_temperature cmd for VF

This is a chip wide value and the PFs already report it.

Signed-off-by: Vasundhara Volam <vasundhara.volam@emulex.com>
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: Adding more speeds reported by get_settings

The new speeds are supported by variants of the Skyhawk-R chip.

Signed-off-by: Vasundhara Volam <vasundhara.volam@emulex.com>
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

iwlwifi: mvm: use designated initialization for some arrays

rs_ht_to_legacy and ant_toggle_lookup are arrays that
represent some state-machine. initialize them explicitly
with designated initialization to make them more clear
and avoid errors.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

iwlwifi: mvm: add some missing cleanups in iwl_mvm_mac_add_interface

iwl_mvm_mac_add_interface() didn't clean up beacon filtering
configuration and ctxt allocation in some error cases.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Reviewed-by: Alexander Bondar <alexander.bondar@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

iwlwifi: remove transport suspend/resume indirection

There's no reason for the transport to call itself through
indirect function pointers, inline the (little) code there
is and remove the indirection completely.

Reviewed-by: Gregory Greenman <gregory.greenman@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

iwlwifi: pcie: some little cleanups

do some little cleanups in tx.c - eliminate duplicate checks,
use locally cached fields and predefined macros.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

iwlwifi: fix some documentation typos

Fix some typos.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

iwlwifi: mvm: small cleanups in quota management code

Use a C99 initializer to clear the command and move the lockdep
assertion before the restart check. Since this causes problems
with the BUILD_BUG_ON() with some compilers, change that a bit.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>

iwlwifi: mvm: fix signal reporting for < 3 antennas

When fewer than three antennas are connected (as is
always the case for the current devices), the signal
strength reporting was wrong; fix it.

Signed-off-by: Avri Altman <avri.altman@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

iwlwifi: mvm: remove traffic load monitoring in rs

The traffic load monitoring isn't used anymore to decide whether
a Tx aggregation on a specific TID should be started.
No point in collecting these statistics. Remove the relevant code.

Signed-off-by: Eyal Shapira <eyal@wizery.com>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

iwlwifi: pcie: fix resume when no opmode is present

If no opmode is present during suspend/resume (i.e. if
the iwldvm or iwlmvm isn't loaded) the driver crashes
during resume, trying to call the rfkill notification.
Avoid that, and also don't enable the rfkill interrupt
in this case (to avoid crashing trying to handle the
interrupt later.)

Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

iwlwifi: mvm: Add RX statistics debugfs entry

Add a debugfs entry for the RX statistics received from
the firmware.

Signed-off-by: Matti Gottlieb <matti.gottlieb@intel.com>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

gianfar: Cleanup TxFCB insertion on xmit

Cleanup gfar_start_xmit()'s fast path by factoring out "redundant"
FCB insertion code (repeated gfar_add_fcb() calls and related)
and by reducing the number of if() clauses (i.e. if(fcb) checks).
Improve maintainability (e.g. there's less code and easier to read)
also by introducing do_csum and do_vlan to mark the other 2 Tx TOE
functionalities, following the same model as do_tstamp.
fcb_len may also be 0 now, to mark that Tx FCB insertion conditions
(do_csum, do_vlan, do_tstamp) have not been met.

Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

gianfar: Fix Tx csum generation errata handling

Both [eTSEC76] and [eTSEC12] errata relate to Tx checksum generation
(for some MPC83xx and MCP8548 older revisions). They require the same
workaround: manual checksum computation and insertion, and disabling
the H/W Tx csum acceleration feature (per frame) through Tx FCB
(Frame Control Block) csum offload settings.

The workaround for [eTSEC76] needs to be fixed because it currently
fails to disable H/W Tx csum insertion via FCB. This patch fixes it
and provides a common workaround implementation for both Tx csum errata.

Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bonding: remove locking from bond_set_rx_mode()

We're already protected by RTNL lock, so nothing can happen to bond/its
slaves, and thus the locking is useless here (both bond->lock and
bond->curr_active_slave).

Also, add ASSERT_RTNL() both to bond_set_rx_mode() and bond_hw_addr_swap()
to catch possible uses of it without RTNL locking.

This patch also saves us from a lockdep false-positive in
bond_set_rx_mode() vs bond_hw_addr_swap().

CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
CC: Nikolay Aleksandrov <nikolay@redhat.com>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers: net: cpsw: Add support for new CPSW IP version

The new IP version has a minor changes and the offsets are same as the
previous version, so adding new IP version support in the driver.

Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Reviewed-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

sctp: Pack dst_cookie into 1st cacheline hole for 64bit host

As dst_cookie is used in fast path sctp_transport_dst_check.

Before:
struct sctp_transport {
struct list_head           transports;           /*     0    16 */
atomic_t                   refcnt;               /*    16     4 */
__u32                      dead:1;               /*    20:31  4 */
__u32                      rto_pending:1;        /*    20:30  4 */
__u32                      hb_sent:1;            /*    20:29  4 */
__u32                      pmtu_pending:1;       /*    20:28  4 */

/* XXX 28 bits hole, try to pack */

__u32                      sack_generation;      /*    24     4 */

/* XXX 4 bytes hole, try to pack */

struct flowi               fl;                   /*    32    64 */
/* --- cacheline 1 boundary (64 bytes) was 32 bytes ago --- */
union sctp_addr            ipaddr;               /*    96    28 */

After:
struct sctp_transport {
struct list_head           transports;           /*     0    16 */
atomic_t                   refcnt;               /*    16     4 */
__u32                      dead:1;               /*    20:31  4 */
__u32                      rto_pending:1;        /*    20:30  4 */
__u32                      hb_sent:1;            /*    20:29  4 */
__u32                      pmtu_pending:1;       /*    20:28  4 */

/* XXX 28 bits hole, try to pack */

__u32                      sack_generation;      /*    24     4 */
u32                        dst_cookie;           /*    28     4 */
struct flowi               fl;                   /*    32    64 */
/* --- cacheline 1 boundary (64 bytes) was 32 bytes ago --- */
union sctp_addr            ipaddr;               /*    96    28 */

Signed-off-by: Fan Du <fan.du@windriver.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bonding: add bond_time_in_interval() and use it for time comparison

Currently we use a lot of time comparison math for arp_interval
comparisons, which are sometimes quite hard to read and understand.

All the time comparisons have one pattern:
(time - arp_interval_jiffies) <= jiffies <= (time + mod *
arp_interval_jiffies + arp_interval_jiffies/2)

Introduce a new helper - bond_time_in_interval(), which will do the math in
one place and, thus, will clean up the logical code. This helper introduces
a bit of overhead (by always calculating the jiffies from arp_interval),
however it's really not visible, considering that functions using it
usually run once in arp_interval milliseconds.

There are several lines slightly over 80 chars, however breaking them would
result in more hard-to-read code than several character after the 80 mark.

CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bonding: call slave_last_rx() only once per slave

Simple cleanup to not call slave_last_rx() on every time function. It won't
give any measurable boost - but looks cleaner and easier to understand.

There are no time-consuming functions in between these calls, so it's safe
to call it in the beginning only once.

CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ath9k: Add more PCI IDs for WB225 cards

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: Fix antenna control init for AR9485

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: Remove ath_ant_comb_update()

During a HW reset, the diversity config is programmed
in the set_board_values() eeprom callback, there is no
need to do it again by calling ath_ant_comb_update().

Fixed antenna support is not fully handled for 1-stream
cards, it can be done later.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

hostap: do not return positive number on failure path in prism2_open()

prism2_open() as an .ndo_open handler should not return positive numbers
in case of failure, but it does return 1 in a couple of places.

Found by Linux Driver Verification project (linuxtesting.org).

Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: Support ANT diversity for WB225

WB225 based cards like CUS198 and CUS230 support
both fast antenna diversity and LNA combining. Add support
for this and also program the SWCOM register with the
correct "ant_ctrl_comm2g_switch_enable" value.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: Set SWCOM value for CUS198

CUS198/CUS230 cards require a custom value to be
programmed into the SWCOM register. Assign this during
init time.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: Remove "shared_chain_lnadiv"

This variable is redundant since we can use
common->bt_ant_diversity to determine if diversity
has to be enabled/disabled.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: Program HW for WB195 diversity

The MC_GAIN_CTL/CCK_DETECT registers have to be programmed
with the correct configuration values if WLAN/BT RX diversity
is enabled. Add this and also take care of the BTCOEX mode
when fast diversity is enabled/disabled.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: Enable WLAN/BT Ant Diversity for WB225/WB195

A custom solution for Asus is WB195 based and supports
WLAN/BT Rx diversity. Identify this card and set the
capability.

CUS198/CUS230, which are based on WB225 also support
WLAN/BT Rx diversity.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: Rename ath9k_hw_antctrl_shared_chain_lnadiv

Use "ath9k_hw_set_bt_ant_diversity" instead.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: Add a HW capability for WLAN/BT RX diversity

Make use of this capability to restrict the usage of the
debugfs file and modparam using which this feature can
be enabled.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: Cleanup WLAN/BT RX diversity

For single-chain WLAN+BT cards, the BT antenna can be used for
WLAN RX when the BT interface is disabled. Rename the modparam
"antenna_diversity" to "bt_ant_diversity" to clarify this.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>