Merge branch 'for-davem' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless-next

John W. Linville says:

====================
Please pull this batch of updates intended for the 3.14 stream...

For the mac80211 bits, Johannes says:

"I have various improvements/cleanups/fixes all over, but the shortlog
shows that Luis's regulatory work and mesh work from the cozybit folks
are the biggest ones, along with the CSA fixes."

Along with that, we have big batches of updates to brcmfmac, rtlwifi,
and ath9k.  There are updates to wcn36xx, rt2x00, and a handful of
others as well.
====================

Signed-off-by: David S. Miller <[email protected]>

tcp: auto corking

With the introduction of TCP Small Queues, TSO auto sizing, and TCP
pacing, we can implement Automatic Corking in the kernel, to help
applications doing small write()/sendmsg() to TCP sockets.

Idea is to change tcp_push() to check if the current skb payload is
under skb optimal size (a multiple of MSS bytes)

If under 'size_goal', and at least one packet is still in Qdisc or
NIC TX queues, set the TCP Small Queue Throttled bit, so that the push
will be delayed up to TX completion time.

This delay might allow the application to coalesce more bytes
in the skb in following write()/sendmsg()/sendfile() system calls.

The exact duration of the delay is depending on the dynamics
of the system, and might be zero if no packet for this flow
is actually held in Qdisc or NIC TX ring.

Using FQ/pacing is a way to increase the probability of
autocorking being triggered.

Add a new sysctl (/proc/sys/net/ipv4/tcp_autocorking) to control
this feature and default it to 1 (enabled)

Add a new SNMP counter : nstat -a | grep TcpExtTCPAutoCorking
This counter is incremented every time we detected skb was under used
and its flush was deferred.

Tested:

Interesting effects when using line buffered commands under ssh.

Excellent performance results in term of cpu usage and total throughput.

lpq83:~# echo 1 >/proc/sys/net/ipv4/tcp_autocorking
lpq83:~# perf stat ./super_netperf 4 -t TCP_STREAM -H lpq84 -- -m 128
9410.39

 Performance counter stats for './super_netperf 4 -t TCP_STREAM -H lpq84 -- -m 128':

      35209.439626 task-clock                #    2.901 CPUs utilized
             2,294 context-switches          #    0.065 K/sec
               101 CPU-migrations            #    0.003 K/sec
             4,079 page-faults               #    0.116 K/sec
    97,923,241,298 cycles                    #    2.781 GHz                     [83.31%]
    51,832,908,236 stalled-cycles-frontend   #   52.93% frontend cycles idle    [83.30%]
    25,697,986,603 stalled-cycles-backend    #   26.24% backend  cycles idle    [66.70%]
   102,225,978,536 instructions              #    1.04  insns per cycle
                                             #    0.51  stalled cycles per insn [83.38%]
    18,657,696,819 branches                  #  529.906 M/sec                   [83.29%]
        91,679,646 branch-misses             #    0.49% of all branches         [83.40%]

      12.136204899 seconds time elapsed

lpq83:~# echo 0 >/proc/sys/net/ipv4/tcp_autocorking
lpq83:~# perf stat ./super_netperf 4 -t TCP_STREAM -H lpq84 -- -m 128
6624.89

 Performance counter stats for './super_netperf 4 -t TCP_STREAM -H lpq84 -- -m 128':
      40045.864494 task-clock                #    3.301 CPUs utilized
               171 context-switches          #    0.004 K/sec
                53 CPU-migrations            #    0.001 K/sec
             4,080 page-faults               #    0.102 K/sec
   111,340,458,645 cycles                    #    2.780 GHz                     [83.34%]
    61,778,039,277 stalled-cycles-frontend   #   55.49% frontend cycles idle    [83.31%]
    29,295,522,759 stalled-cycles-backend    #   26.31% backend  cycles idle    [66.67%]
   108,654,349,355 instructions              #    0.98  insns per cycle
                                             #    0.57  stalled cycles per insn [83.34%]
    19,552,170,748 branches                  #  488.244 M/sec                   [83.34%]
       157,875,417 branch-misses             #    0.81% of all branches         [83.34%]

      12.130267788 seconds time elapsed

Signed-off-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

3c59x/net: Use dev_is_pci() instead of hardcoding

Use PCI standard macro dev_is_pci() instead of hardcoding.

Signed-off-by: Yijing Wang <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net/fddi: Replace local macro with PCI standard macro

Replace local macro DFX_BUS_PCI() with PCI standard macro
dev_is_pci().

Acked-by: Maciej W. Rozycki <[email protected]>
Signed-off-by: Yijing Wang <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

tcp: optimize some skb_shinfo(skb) uses

Compiler doesn't know skb_shinfo(skb) pointer is usually constant.

By using a temporary variable, we help generating smaller code.

For example, tcp_init_nondata_skb() is inlined after this patch.

Signed-off-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

gro: small napi_get_frags() optim

Remove one useless conditional branch :
napi->skb is NULL, so nothing bad can happen.

Signed-off-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ipv6: consistent use of IP6_INC_STATS_BH() in ip6_forward()

ip6_forward() runs from softirq context, we can use the SNMP macros
assuming this.

Use same indentation for all IP6_INC_STATS_BH() calls.

Signed-off-by: Eric Dumazet <[email protected]>
Acked-by: Hannes Frederic Sowa <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

packet: use macro GET_PBDQC_FROM_RB to simplify the codes

Signed-off-by: Duan Jiong <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

tun: spelling fixes

Fix spelling errors in tun driver.

Signed-off-by: Stephen Hemminger <[email protected]>
Acked-by: Jason Wang <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net/*: Fix FSF address in file headers

Several files refer to an old address for the Free Software Foundation
in the file header comment.  Resolve by replacing the address with
the URL <http://www.gnu.org/licenses/> so that we do not have to keep
updating the header comments anytime the address changes.

CC: John Fastabend <[email protected]>
CC: Alex Duyck <[email protected]>
CC: Marcel Holtmann <[email protected]>
CC: Gustavo Padovan <[email protected]>
CC: Johan Hedberg <[email protected]>
CC: Jamal Hadi Salim <[email protected]>
Signed-off-by: Jeff Kirsher <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net/irda: Fix FSF address in file headers

Several files refer to an old address for the Free Software Foundation
in the file header comment.  Resolve by replacing the address with
the URL <http://www.gnu.org/licenses/> so that we do not have to keep
updating the header comments anytime the address changes.

CC: Samuel Ortiz <[email protected]>
Signed-off-by: Jeff Kirsher <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

netfilter: Fix FSF address in file headers

Several files refer to an old address for the Free Software Foundation
in the file header comment.  Resolve by replacing the address with
the URL <http://www.gnu.org/licenses/> so that we do not have to keep
updating the header comments anytime the address changes.

CC: [email protected]
CC: Pablo Neira Ayuso <[email protected]>
CC: Patrick McHardy <[email protected]>
CC: Jozsef Kadlecsik <[email protected]>
Signed-off-by: Jeff Kirsher <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

netlabel: Fix FSF address in file headers

Several files refer to an old address for the Free Software Foundation
in the file header comment.  Resolve by replacing the address with
the URL <http://www.gnu.org/licenses/> so that we do not have to keep
updating the header comments anytime the address changes.

CC: Paul Moore <[email protected]>
Signed-off-by: Jeff Kirsher <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

include/net/: Fix FSF address in file headers

Several files refer to an old address for the Free Software Foundation
in the file header comment.  Resolve by replacing the address with
the URL <http://www.gnu.org/licenses/> so that we do not have to keep
updating the header comments anytime the address changes.

Signed-off-by: Jeff Kirsher <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ipv4/ipv6: Fix FSF address in file headers

Several files refer to an old address for the Free Software Foundation
in the file header comment.  Resolve by replacing the address with
the URL <http://www.gnu.org/licenses/> so that we do not have to keep
updating the header comments anytime the address changes.

CC: Alexey Kuznetsov <[email protected]>
CC: James Morris <[email protected]>
CC: Hideaki YOSHIFUJI <[email protected]>
CC: Patrick McHardy <[email protected]>
Signed-off-by: Jeff Kirsher <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

sctp: Fix FSF address in file headers

Several files refer to an old address for the Free Software Foundation
in the file header comment.  Resolve by replacing the address with
the URL <http://www.gnu.org/licenses/> so that we do not have to keep
updating the header comments anytime the address changes.

CC: Vlad Yasevich <[email protected]>
CC: Neil Horman <[email protected]>
Signed-off-by: Jeff Kirsher <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

drivers/net/*: Fix FSF address in file headers

Several files refer to an old address for the Free Software Foundation
in the file header comment.  Resolve by replacing the address with
the URL <http://www.gnu.org/licenses/> so that we do not have to keep
updating the header comments anytime the address changes.

CC: Jay Vosburgh <[email protected]>
CC: Veaceslav Falico <[email protected]>
CC: Andy Gospodarek <[email protected]>
CC: Haiyang Zhang <[email protected]>
CC: "K. Y. Srinivasan" <[email protected]>
CC: Paul Mackerras <[email protected]>
CC: Ian Campbell <[email protected]>
CC: Wei Liu <[email protected]>
CC: Rusty Russell <[email protected]>
CC: "Michael S. Tsirkin" <[email protected]>
Signed-off-by: Jeff Kirsher <[email protected]>
Acked-by: Wei Liu <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

usb: Fix FSF address in file headers

Several files refer to an old address for the Free Software Foundation
in the file header comment.  Resolve by replacing the address with
the URL <http://www.gnu.org/licenses/> so that we do not have to keep
updating the header comments anytime the address changes.

CC: Oliver Neukum <[email protected]>
CC: Steve Glendinning <[email protected]>
CC: Oliver Neukum <[email protected]>
Signed-off-by: Jeff Kirsher <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

irda: Fix FSF address in file headers

Several files refer to an old address for the Free Software Foundation
in the file header comment.  Resolve by replacing the address with
the URL <http://www.gnu.org/licenses/> so that we do not have to keep
updating the header comments anytime the address changes.

CC: Samuel Ortiz <[email protected]>
Signed-off-by: Jeff Kirsher <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ethernet: Fix FSF address in file headers

Several files refer to an old address for the Free Software Foundation
in the file header comment.  Resolve by replacing the address with
the URL <http://www.gnu.org/licenses/> so that we do not have to keep
updating the header comments anytime the address changes.

CC: Santosh Raspatur <[email protected]>
CC: Dimitris Michailidis <[email protected]>
CC: Michael Chan <[email protected]>
CC: Santiago Leon <[email protected]>
CC: Sebastian Hesselbarth <[email protected]>
CC: Olof Johansson <[email protected]>
CC: Manish Chopra <[email protected]>
CC: Sony Chacko <[email protected]>
CC: Rajesh Borundia <[email protected]>
CC: Nicolas Pitre <[email protected]>
CC: Steve Glendinning <[email protected]>
Signed-off-by: Jeff Kirsher <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

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

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

Conflicts:
drivers/net/wireless/brcm80211/Kconfig
net/mac80211/util.c

macvlan: Support creating macvtaps from macvlans

When running in a network namespace whose only link to the outside
world is a macvlan device, not being able to create a macvtap off of
it is a real pain.

So modify macvtap creation to automatically forward a creation of a
macvtap on a macvlan to become a creation of a macvtap on the
underlying network device, just like is currently done with
macvlan-on-macvlan devices.

v2: Use netif_is_macvlan and macvlan_dev_real_dev helpers to make it
    more clear what we're doing.

Signed-off-by: Kevin Wallace <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Merge branch 'siocghwtstamp' of git://git.kernel.org/pub/scm/linux/kernel/git/bwh/sfc-next

Ben Hutchings says:

====================
SIOCGHWTSTAMP ioctl

1. Add the SIOCGHWTSTAMP ioctl and update the timestamping
documentation.
2. Implement SIOCGHWTSTAMP in most drivers that support SIOCSHWTSTAMP.
3. Add a test program to exercise SIOC{G,S}HWTSTAMP.
====================

Signed-off-by: David S. Miller <[email protected]>

Merge branch 'for-davem' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless

John W. Linville says:

====================
Please pull this batch of fixes intende for the 3.13 stream!

For the mac80211 bits, Johannes says:

"For now I have various fixes all over, mostly for issues introduced in
relatively recent patches. There's no real pattern to it. Some of the
issues like go back longer, but still seemed 3.13 material."

And...

"These are just two patches disabling the broken CSA code. Once this
goes into your tree I'll merge it into mac80211-next and revert there
(since we fixed the bugs there)."

For the iwlwifi bits, Emmanuel says:

"I have here a few fixes for BT Coex. One of them is a NULL pointer
dereference. Another one avoids to enable a feature that can make the
firmware unhappy since the firmware isn't ready for it yet. WE also
avoid a WARNING that can be triggered upon association in not-so-bad
cases even if the association succeeded. We add support for new NICs
(not yet on the market) and bump the API so that 3.13 will be able to
work with the new firmware that will be out soon hopefully.
I also have a boundary check from Johannes."

In addition to those...

- Arend van Spriel fixes a brcmfmac problem that could use an
uninitialized variable in an error path.

- Borislav Petkov fixes a Kconfig-based build breakage problem for
brcmsmac.

- Michal Nazarewicz fixes a couple of NULL pointer dereference problems
in ath9k and wcn36xx.

- Sujith Manoharan fixes a couple of ath9k problems related to
incorrect interpretation of EEPROM configuration data.

- Ujjal Roy fixes a memory leak in mwifiex.
====================

Signed-off-by: David S. Miller <[email protected]>

ath: fix dynamic user regulatory domain support

As it stands dynamic user regulatory domain support is
only possible for a few programmed regulatory domains as
a few countries do not allow for this.

The existing code however only would take advantage of
the feature if a custom world regulatory domain is used
though as that's when we clear beconing flags. We need
to lift this restriction as otherwise this feature is
pointless.

Signed-off-by: Luis R. Rodriguez <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

ath: dynamic user allow check helper to the top

This will be used later.

Signed-off-by: Luis R. Rodriguez <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

ath: move dynamic_country_user_possible() to the top

This will be used later.

Signed-off-by: Luis R. Rodriguez <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

ath: fix usage of freq_reg_info()

freq_reg_info() expects KHz and not MHz, fix this. In
this case we'll now be getting the no-ir flags cleared
on channels for any channel when the country IE trusts
that channel.

@@
struct ieee80211_channel *ch;
struct wiphy *wiphy;
const struct ieee80211_reg_rule *rule;
@@

-rule = freq_reg_info(wiphy, ch->center_freq);
+rule = freq_reg_info(wiphy, MHZ_TO_KHZ(ch->center_freq));

Generated-by: Coccinelle SmPL
Cc: Julia Lawall <[email protected]>
Cc: Peter Senna Tschudin <[email protected]>
Reported-by: Mihir Shete <[email protected]>
Signed-off-by: Luis R. Rodriguez <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

ath: move the channel for ath_reg_apply_beaconing_flags() into helper

While at it convert this into a switch statement, this
makes it easier and manage.

Cc: [email protected]
Cc: [email protected]
Signed-off-by: Luis R. Rodriguez <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

ath: simplify ath_reg_apply_beaconing_flags()

Simplify ath_reg_apply_beaconing_flags() by making use of
thew new no-ir helper.

Cc: [email protected]
Cc: [email protected]
Signed-off-by: Luis R. Rodriguez <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

ath: rename ath_reg_apply_active_scan_flags() to ath_reg_apply_ir_flags()

This also applies the no-ibss flag to the channels or clears it.
The idea here is to clarify no initiated radiation should be
allowed on these channels.

Cc: [email protected]
Cc: [email protected]
Signed-off-by: Luis R. Rodriguez <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

ath: clean up ath_reg_apply_active_scan_flags()

The routine ath_reg_apply_active_scan_flags() can be a bit
hard to read, this cleans it up by adding helpers for the
two cases of clearing IR flags or adding them. This approach
also makes no assumptions on the index of channels 12 and 13
so it should be portable accross different drivers.

Cc: [email protected]
Cc: [email protected]
Signed-off-by: Luis R. Rodriguez <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

ath: fix logic on ath_reg_apply_active_scan_flags()

The existing logic removes the passive scan flag from
channels 12 and 13 when a regulatory hint coming from
something other than a country IE has been passed. This
is incorrect, the original intention was to ensure we
always have passive scan enabled for these two channels
for a specific set of custom world regulatory domains.

Cc: [email protected]
Cc: [email protected]
Signed-off-by: Luis R. Rodriguez <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

wcn36xx: fix typo error

Signed-off-by: Chun-Yeow Yeoh <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: rtl8188ee: Fix typo in code

The static analyser "cppcheck" shows the following typo:
drivers/net/wireless/rtlwifi/rtl8188ee/dm.c:1081]: (style) Same expression on both sides of '!='.

Signed-off-by: Larry Finger <[email protected]>
Reported-by: David Binderman <[email protected]>
Cc: David Binderman <[email protected]>
Cc: Stable <[email protected]> [3.10+]
Signed-off-by: John W. Linville <[email protected]>

bcma: pci: remove unnecessary pci_set_drvdata()

The driver core clears the driver data to NULL after device_release
or on probe failure. Thus, it is not needed to manually clear the
device driver data to NULL.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

wireless: libertas: remove unnecessary spi_set_drvdata()

The driver core clears the driver data to NULL after device_release
or on probe failure. Thus, it is not needed to manually clear the
device driver data to NULL.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

wireless: rt2x00: remove unnecessary pci_set_drvdata()

The driver core clears the driver data to NULL after device_release
or on probe failure. Thus, it is not needed to manually clear the
device driver data to NULL.

Signed-off-by: Jingoo Han <[email protected]>
Acked-by: Gertjan van Wingerde <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

wireless: prism54: remove unnecessary pci_set_drvdata()

The driver core clears the driver data to NULL after device_release
or on probe failure. Thus, it is not needed to manually clear the
device driver data to NULL.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

wireless: ipw2100: remove unnecessary pci_set_drvdata()

The driver core clears the driver data to NULL after device_release
or on probe failure. Thus, it is not needed to manually clear the
device driver data to NULL.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

brcmsmac: remove dependency on CRC8 and CRC_CCITT

There is no code in brcmsmac that uses a function from CRC8 or
CRC_CCITT any more. Building brcmsmac with these two disabled works
without any problems. This was probably only used by the bus code which
was replaced by bcma some time ago.

Signed-off-by: Hauke Mehrtens <[email protected]>
Acked-by: Arend van Spriel <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

ath9k: fix retry chain initialization in tx99 code

Initialize first chain attempt counter to 1 in ath9k_build_tx99_skb().
Otherwise multi-retry chain is initialized to {idx,count} = {-1, 0} in
rate_control_fill_sta_table() and tx99 transmission rate is not configured in
rate_control_apply_mask() since first chain idx is set to -1

Signed-off-by: Lorenzo Bianconi <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

wcn36xx: set self STA default HT parameters

These default HT parameters are required for self STA entry.
In example, set the HT capable of self STA entry for bss
configuration in mesh allows the MCS rate to be used. Otherwise,
only legacy rate will be used.

Signed-off-by: Chun-Yeow Yeoh <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

libertas sdio: claim device before calling sdio_disable_func()

There is a failure path in if_sdio_power_on(), where
sdio_disable_func() is called without claiming the device.

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

Signed-off-by: Alexey Khoroshilov <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: rtl8192cu: Update table.c

The latest vendor driver has some new values.

Signed-off-by: Larry Finger <[email protected]>
Signed-off-by: Mark Cave-Ayland <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: rtl8192cu: Bring rtl92cu_phy_bb_config() up to date

The latest vendor driver contains some changes not in the kernel version.

Signed-off-by: Mark Cave-Ayland <[email protected]>
Signed-off-by: Larry Finger <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: Bring _rtl92c_phy_iq_calibrate() in line with the vendor driver

Bring _rtl92c_phy_iq_calibrate() in line with the vendor driver function _PHY_IQCalibrate().

Also fix incorrect initialisation for rtl8192cu.

Signed-of-by: Mark Cave-Ayland <[email protected]>
Signed-off-by: Larry Finger <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: rtl8192c: Update dynamic gain calculations

The vendor driver contained a number of improvements in the gain settings
for the rtl8192c{e,u} devices. This patch implements them in the kernel
driver.

Signed-off-by: Larry Finger <[email protected]>
Cc: Stable <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: Add missing code to PWDB statics routine

Signed-off-by: Larry Finger <[email protected]>
Cc: Stable <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: rtl8192cu: Fix some code in RF handling

Signed-off-by: Larry Finger <[email protected]>
Cc: Stable <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: rtl8192cu: Update the power index registers

This patch uses the newly introduced power index register routines.

Signed-off-by: Larry Finger <[email protected]>
Cc: Stable <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: rtl8192c: Add routines to save/restore power index registers

Signed-off-by: Larry Finger <[email protected]>
Cc: Stable <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: Update beacon statistics for USB driver

The USB drivers were not updating the beacon statistics, which led to
false beacon loss indications.

Signed-off-by: Larry Finger <[email protected]>
Cc: Stable <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: Increase the RX queue length for USB drivers

The current number of RX buffers queued is 32, which is too small under
heavy load. That number is doubled.

Signed-off-by: Larry Finger <[email protected]>
Cc: Stable <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: rtl8192c: Add new definitions in the dm_common header

Changes in the gain-control mechanism will require some changes in the header.

Signed-off-by: Larry Finger <[email protected]>
Cc: Stable <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: Set the link state

The present code fails to set the linked state when an interface is
added.

Signed-off-by: Larry Finger <[email protected]>
Cc: Stable <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: Redo register save locations

The initial USB driver did not use some register save locations in the
private data storage. To save some memory, a union was used to overlay these
variables with USB I/O components. In an update of the gain-control code,
these register save locations are now needed for USB drivers.

Signed-off-by: Larry Finger <[email protected]>
Cc: Stable <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: rtl8192cu: Add new firmware

Vendor driver rtl8188C_8192C_8192D_usb_linux_v3.4.2_3727.20120404 introduced
new firmware for these chips. The code try for the new file, and fall back to
the original firmware if the new file is not available.

Signed-off-by: Larry Finger <[email protected]>
Cc: Stable <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: rtl8192c: Prevent reconnect attempts if not connected

This driver has a watchdog timer that attempts to reconnect when beacon frames
are not seen for 6 seconds. This patch disables that reconnect whenever the
device has never been connected.

Signed-off-by: Larry Finger <[email protected]>
Cc: Stable <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rt2x00: rt2800lib: no need to write RF register 3 twice for RT5592

In rt2800_init_rfcsr_5592(), there's no need to write RF register 3 twice.

Signed-off-by: Kevin Lo <[email protected]>
Acked-by: Gertjan van Wingerde <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: Remove unused calls to rtl_is_special_data()

When routine rtl_is_special_data() is called with false as its last argument,
and the returned value is not tested, the call is essentially an extended
no-op. Accordingly, these calls may be removed.

Signed-off-by: Larry Finger <[email protected]>
Reported-by: Mark Cave-Ayland <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

wcn36xx: Fix logging macro with unnecessary semicolon

The wcn36xx_err macro should not end in a semicolon as
there are 2 consecutive semicolons in the preprocessed
output.

Signed-off-by: Joe Perches <[email protected]>
Signed-off-by: Eugene Krasnikov <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

wcn36xx: enable beacon change using BSS_CHANGED_BEACON

Enable the beacon changed using BSS_CHANGED_BEACON. This is
especially useful for mesh mode.

Signed-off-by: Chun-Yeow Yeoh <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtl8187: Increase RX queue depth

Under heavy load, the relatively small number of RX queue entries are
completely filled. With an increase from 16 to 32 entries, this condition
rarely happens.

Signed-off-by: Larry Finger <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rtlwifi: fix usage of freq_reg_info()

freq_reg_info() expects KHz and not MHz, fix this. In
this case we'll now be getting the no-ir flags cleared
on channels for any channel when the country IE trusts
that channel.

@@
struct ieee80211_channel *ch;
struct wiphy *wiphy;
const struct ieee80211_reg_rule *rule;
@@

-rule = freq_reg_info(wiphy, ch->center_freq);
+rule = freq_reg_info(wiphy, MHZ_TO_KHZ(ch->center_freq));

Generated-by: Coccinelle SmPL
Cc: Julia Lawall <[email protected]>
Cc: Peter Senna Tschudin <[email protected]>
Cc: Larry Finger <[email protected]>
Reported-by: Mihir Shete <[email protected]>
Signed-off-by: Luis R. Rodriguez <[email protected]>
Acked-by: Larry Finger <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

brcm80211: fix usage of freq_reg_info()

freq_reg_info() expects KHz and not MHz, fix this. In
this case we'll now be getting the no-ir flags cleared
on channels for any channel when the country IE trusts
that channel.

@@
struct ieee80211_channel *ch;
struct wiphy *wiphy;
const struct ieee80211_reg_rule *rule;
@@

-rule = freq_reg_info(wiphy, ch->center_freq);
+rule = freq_reg_info(wiphy, MHZ_TO_KHZ(ch->center_freq));

Generated-by: Coccinelle SmPL
Cc: Julia Lawall <[email protected]>
Cc: Peter Senna Tschudin <[email protected]>
Cc: Seth Forshee <[email protected]>
Cc: Arend van Spriel <[email protected]>
Reported-by: Mihir Shete <[email protected]>
Signed-off-by: Luis R. Rodriguez <[email protected]>
Acked-by: Arend van Spriel <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

wireless: cw1200: Use consistent internal locking conventions

The cw1200_irq_handler() function expects the hwbus lock to be held when
it is called.  On the SDIO platform, this lock is implemented in terms
of sdio_claim_host/sdio_release_host.

This trivial patch makes it explicit that we are performing the hwbus
lock rather than something SDIO-specific.

Signed-off-by: Solomon Peachy <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

cw1200: Make the "scan failed" message into a warning

The reason that a scan failed for some reason (typically bad
parameters) should be logged even when debugging is turned off.

Signed-off-by: Solomon Peachy <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

iwl3945: do not print RFKILL message

We can mess logs if user space try to open device again and again if
RFKILL switch is on. Do not print message and return ERFKILL error
instead to indicate where the problem is.

Note that iwl4965 handle this problem differently, it allows to open
device when radio is disabled.

Reported-by: Dietmar Rudolph <[email protected]>
Signed-off-by: Stanislaw Gruszka <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

wl1251: return -ENOMEM if kzalloc fails

the original code used goto out if kzalloc fails,but the out include kfree,
so return -ENOMEM if kzalloc fails.

Signed-off-by: Jing Wang <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

libertas/sdio: make sure card enters power-off when firmware is loaded.

Since firmware loading became async it is possible that if_sdio_finish_power_of
is called with only one remaining runtime_pm reference, so it isn't safe
to call pm_runtime_put_noidle.  We must call pm_runtime_put().

Diagnosed-by: "Dr. H. Nikolaus Schaller" <[email protected]>
Signed-off-by: NeilBrown <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rt2x00: rt2800lib: update RF registers for RT5390

Mirror the latest MediaTek/Ralink driver with respect to RT5390 RF register
programming.  The PCI and USB devices use different init values.

Signed-off-by: Kevin Lo <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

mwifiex: improvements in mwifiex_ret_tx_power_cfg()

1) Move common code out of switch case handling
2) Return from the function if number of bytes left in response buffer
are less than tlv size
3) Pass pg_tlv_hdr directly instead of txp_cfg to mwifiex_get_power_level()

Signed-off-by: Amitkumar Karwar <[email protected]>
Signed-off-by: Bing Zhao <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

rt2x00: rt2800lib: Update BBP register initialization for RT53xx

Update bbp register initialization for RT53xx chips to match with the
latest MediaTek/Ralink driver.

Based on: NICInitRT5390BbpRegisters()
From: DPO_RT5572_LinuxSTA_2.6.1.3_20121022/chips/rt5390.c

Signed-off-by: Kevin Lo <kevlo.org>
Signed-off-by: John W. Linville <[email protected]>

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

Merge branch 'cxgb4'

Hariprasad Shenai says:

====================
Fixes T5 adapter init, due to incorrect FW version check

This patch series fixes, Chelsio T5 adapter initialization failure due to
incorrect firmware version check. This patch series modifies the firmware
flashing mechanism for T4/T5 adapter.

The patch series moves chip type from struct adapter to struct adapter_params.
It changes the references of chip type in cxgb4 and cxgb4vf drivers such that
build failure is avoided.

Patch 3/3 is dependent on patch 1/3
Patch 2/3 is also dependent on patch 1/3

We would like to request this patch series to get merged via David Miller's
'net' tree.

We have included all the maintainers of respective drivers. Kindly review the
change and let us know in case of any review comments.
====================

Signed-off-by: David S. Miller <[email protected]>

cxgb4: Add new scheme to update T4/T5 firmware

Signed-off-by: Hariprasad Shenai <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

cxgb4vf: added much cleaner implementation of is_t4()

Signed-off-by: Hariprasad Shenai <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

cxgb4: Much cleaner implementation of is_t4()/is_t5()

Signed-off-by: Hariprasad Shenai <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net/mlx4_core: destroy workqueue when driver fails to register

When driver registration fails, we need to clean up the resources allocated
before. mlx4_core missed destroying the workqueue allocated.

This patch destroys the workqueue when registration fails.

Signed-off-by: Wei Yang <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

rds: prevent BUG_ON triggered on congestion update to loopback

After congestion update on a local connection, when rds_ib_xmit returns
less bytes than that are there in the message, rds_send_xmit calls
back rds_ib_xmit with an offset that causes BUG_ON(off & RDS_FRAG_SIZE)
to trigger.

For a 4Kb PAGE_SIZE rds_ib_xmit returns min(8240,4096)=4096 when actually
the message contains 8240 bytes. rds_send_xmit thinks there is more to send
and calls rds_ib_xmit again with a data offset "off" of 4096-48(rds header)
=4048 bytes thus hitting the BUG_ON(off & RDS_FRAG_SIZE) [RDS_FRAG_SIZE=4k].

The commit 6094628bfd94323fc1cea05ec2c6affd98c18f7f
"rds: prevent BUG_ON triggering on congestion map updates" introduced
this regression. That change was addressing the triggering of a different
BUG_ON in rds_send_xmit() on PowerPC architecture with 64Kbytes PAGE_SIZE:
  BUG_ON(ret != 0 &&
      conn->c_xmit_sg == rm->data.op_nents);
This was the sequence it was going through:
(rds_ib_xmit)
/* Do not send cong updates to IB loopback */
if (conn->c_loopback
   && rm->m_inc.i_hdr.h_flags & RDS_FLAG_CONG_BITMAP) {
   rds_cong_map_updated(conn->c_fcong, ~(u64) 0);
     return sizeof(struct rds_header) + RDS_CONG_MAP_BYTES;
}
rds_ib_xmit returns 8240
rds_send_xmit:
  c_xmit_data_off = 0 + 8240 - 48 (rds header accounted only the first time)
     = 8192
  c_xmit_data_off < 65536 (sg->length), so calls rds_ib_xmit again
rds_ib_xmit returns 8240
rds_send_xmit:
  c_xmit_data_off = 8192 + 8240 = 16432, calls rds_ib_xmit again
  and so on (c_xmit_data_off 24672,32912,41152,49392,57632)
rds_ib_xmit returns 8240
On this iteration this sequence causes the BUG_ON in rds_send_xmit:
    while (ret) {
     tmp = min_t(int, ret, sg->length - conn->c_xmit_data_off);
     [tmp = 65536 - 57632 = 7904]
     conn->c_xmit_data_off += tmp;
     [c_xmit_data_off = 57632 + 7904 = 65536]
     ret -= tmp;
     [ret = 8240 - 7904 = 336]
     if (conn->c_xmit_data_off == sg->length) {
     conn->c_xmit_data_off = 0;
     sg++;
     conn->c_xmit_sg++;
     BUG_ON(ret != 0 &&
     conn->c_xmit_sg == rm->data.op_nents);
     [c_xmit_sg = 1, rm->data.op_nents = 1]

What the current fix does:
Since the congestion update over loopback is not actually transmitted
as a message, all that rds_ib_xmit needs to do is let the caller think
the full message has been transmitted and not return partial bytes.
It will return 8240 (RDS_CONG_MAP_BYTES+48) when PAGE_SIZE is 4Kb.
And 64Kb+48 when page size is 64Kb.

Reported-by: Josh Hunt <[email protected]>
Tested-by: Honggang Li <[email protected]>
Acked-by: Bang Nguyen <[email protected]>
Signed-off-by: Venkat Venkatsubra <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

xen-netback: clear vif->task on disconnect

xenvif_start_xmit() relies on checking vif->task for NULL to determine
whether the vif is ready to accept packets. The task thread is stopped in
xenvif_disconnect() but task is not set to NULL. Thus, on a re-connect the
check will give a false positive.

Also since commit ea732dff5cfa10789007bf4a5b935388a0bb2a8f (Handle backend
state transitions in a more robust way) it should not be possible for
xenvif_connect() to be called if the vif is already connected so change the
check of vif->tx_irq to a BUG_ON() and also add a BUG_ON(vif->task).

Signed-off-by: Paul Durrant <[email protected]>
Cc: Wei Liu <[email protected]>
Cc: Ian Campbell <[email protected]>
Cc: David Vrabel <[email protected]>
Acked-by: Wei Liu <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Revert "net: Handle CHECKSUM_COMPLETE more adequately in pskb_trim_rcsum()."

This reverts commit 018c5bba052b3a383d83cf0c756da0e7bc748397.

It causes regressions for people using chips driven by the sungem
driver.  Suspicion is that the skb->csum value isn't being adjusted
properly.

The change also has a bug in that if __pskb_trim() fails, we'll leave
a corruped skb->csum value in there.  We would really need to revert
it to it's original value in that case.

Signed-off-by: David S. Miller <[email protected]>

net: do not pretend FRAGLIST support

Few network drivers really supports frag_list : virtual drivers.

Some drivers wrongly advertise NETIF_F_FRAGLIST feature.

If skb with a frag_list is given to them, packet on the wire will be
corrupt.

Remove this flag, as core networking stack will make sure to
provide packets that can be sent without corruption.

Signed-off-by: Eric Dumazet <[email protected]>
Cc: Thadeu Lima de Souza Cascardo <[email protected]>
Cc: Anirudha Sarangi <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

IPv6: Fixed support for blackhole and prohibit routes

The behaviour of blackhole and prohibit routes has been corrected by setting
the input and output pointers of the dst variable appropriately. For
blackhole routes, they are set to dst_discard and to ip6_pkt_discard and
ip6_pkt_discard_out respectively for prohibit routes.

ipv6: ip6_pkt_prohibit(_out) should not depend on
CONFIG_IPV6_MULTIPLE_TABLES

We need ip6_pkt_prohibit(_out) available without
CONFIG_IPV6_MULTIPLE_TABLES

Signed-off-by: Kamala R <[email protected]>
Acked-by: Hannes Frederic Sowa <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ipv6: fix third arg of anycast_dst_alloc(), must be bool.

Signed-off-by: Francois-Xavier Le Bail <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net: fec_main: dma_map() only the length of the skb

On tx submit the driver always dma_map_single() FEC_ENET_TX_FRSIZE (=2048)
bytes. This works because we don't overwrite any memory after the data buffer,
we remove it from cache if it was there. So we hurt performace in case the
mapping of a smaller area makes a difference.
There is also a bug: If the data area starts shortly before the end of
RAM say 0xc7fffa10 and the RAM ends at 0xc8000000 then we have enough
space to fit the data area (according to skb->len) but we would map beyond
end of ram if we are using 2048. In v2.6.31 (against which kernel this patch
made) there is the following check in dma_cache_maint():

|BUG_ON(!virt_addr_valid(start) || !virt_addr_valid(start + size - 1));

Since the area starting at 0xc8000000 is no longer virt_addr_valid() we
BUG() during dma_map_single(). The BUG() statement was removed in v3.5-rc1 as
per 2dc6a016 ("ARM: dma-mapping: use asm-generic/dma-mapping-common.h").

This patch was tested on v2.6.31 and then forward-ported and compile
tested only against the net tree. I think it is still worth fixing
mainline even after the BUG() statement is gone.

Tested-by: Fugang Duan <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Sebastian Andrzej Siewior <[email protected]>
Acked-by: Fugang Duan <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

drivers: net: cpsw: fix dt probe for one port ethernet

When only one port of the two port is pinned out, then dt probe is failing
because second port phy is not found. fixing this by checking the number of
slaves and breaking the loop.

Signed-off-by: Mugunthan V N <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

PCI / tg3: Give up chip reset and carrier loss handling if PCI device is not present

Modify tg3_chip_reset() and tg3_close() to check if the PCI network
adapter device is accessible at all in order to skip poking it or
trying to handle a carrier loss in vain when that's not the case.
Introduce a special PCI helper function pci_device_is_present()
for this purpose.

Of course, this uncovers the lack of the appropriate RTNL locking
in tg3_suspend() and tg3_resume(), so add that locking in there
too.

These changes prevent tg3 from burning a CPU at 100% load level for
solid several seconds after the Thunderbolt link is disconnected from
a Matrox DS1 docking station.

Signed-off-by: Rafael J. Wysocki <[email protected]>
Acked-by: Michael Chan <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ipv6: judge the accept_ra_defrtr before calling rt6_route_rcv

when dealing with a RA message, if accept_ra_defrtr is false,
the kernel will not add the default route, and then deal with
the following route information options. Unfortunately, those
options maybe contain default route, so let's judge the
accept_ra_defrtr before calling rt6_route_rcv.

Signed-off-by: Duan Jiong <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

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

brcmfmac: expose chip information through debugfs

Adding the debugfs file <debugfs_mnt>/brcmfmac/<devid>/chipinfo
which contains the chip number and revision.

Reviewed-by: Hante Meuleman <[email protected]>
Reviewed-by: Pieter-Paul Giesberts <[email protected]>
Signed-off-by: Arend van Spriel <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

brcmfmac: move firmware event related defines to fweh.h

The dhd.h file contained a number of definitions that are
related to events received from the firmware. Those are
processed and dispatched in the driver by fweh. Hence the
definitions are moved to its include file.

Reviewed-by: Hante Meuleman <[email protected]>
Reviewed-by: Pieter-Paul Giesberts <[email protected]>
Signed-off-by: Arend van Spriel <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

brcmfmac: remove some unused definitions

Removing WLC_PHY_TYPE and some BRCMF_E_.* definitions as these
are not used in the driver sources.

Reviewed-by: Hante Meuleman <[email protected]>
Reviewed-by: Pieter-Paul Giesberts <[email protected]>
Signed-off-by: Arend van Spriel <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

brcmfmac: Use consistent naming for BCDC.

The BCDC protocol layer is using a mix of naming of CDC, BDC and
BCDC. Use the name BCDC consistenly over all functions, defines
and variables. This patch does not change code functionality.

Reviewed-by: Arend Van Spriel <[email protected]>
Reviewed-by: Pieter-Paul Giesberts <[email protected]>
Signed-off-by: Hante Meuleman <[email protected]>
Signed-off-by: Arend van Spriel <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

brcmfmac: move firmware command code definitions

Move the command codes to the firmware interface module as
that makes a bit more sense.

Reviewed-by: Hante Meuleman <[email protected]>
Reviewed-by: Pieter-Paul Giesberts <[email protected]>
Signed-off-by: Arend van Spriel <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

brcmfmac: move firmware related structures to fwil_types.h

The firmware control interface is provided by fwil source file, but
a number of structures used to communicate with the firmware still
resided in dhd.h. The patch moves them to fwil_types.h.

Reviewed-by: Hante Meuleman <[email protected]>
Reviewed-by: Pieter-Paul Giesberts <[email protected]>
Signed-off-by: Arend van Spriel <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

brcmfmac: remove unnecessary EXPORT_SYMBOL() usage

In bcmsdh.c the functions brcmf_sdio_probe() and brcmf_sdio_remove()
were exported, but that is not needed. The functions are linked into
the driver module, which is the only one needing to call these.

Reviewed-by: Franky Lin <[email protected]>
Reviewed-by: Hante Meuleman <[email protected]>
Reviewed-by: Pieter-Paul Giesberts <[email protected]>
Reviewed-by: Daniel (Deognyoun) Kim <[email protected]>
Signed-off-by: Arend van Spriel <[email protected]>
Signed-off-by: John W. Linville <[email protected]>

brcmfmac: provide firmware version in ethtool driver info

Knowing the firmware version is pretty useful information when
looking at issues. It is retrieved during initialization so
store it in driver data structure to fill the ethtool driver
info when requested.

Reviewed-by: Hante Meuleman <[email protected]>
Reviewed-by: Franky Lin <[email protected]>
Reviewed-by: Pieter-Paul Giesberts <[email protected]>
Signed-off-by: Arend van Spriel <[email protected]>
Signed-off-by: John W. Linville <[email protected]>