packet: Enhance AF_PACKET implementation to not require high order contiguous memory allocation (v4)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Version 4 of this patch.

Change notes:
1) Removed extra memset.  Didn't think kcalloc added a GFP_ZERO the way kzalloc did :)

Summary:
It was shown to me recently that systems under high load were driven very deep
into swap when tcpdump was run.  The reason this happened was because the
AF_PACKET protocol has a SET_RINGBUFFER socket option that allows the user space
application to specify how many entries an AF_PACKET socket will have and how
large each entry will be.  It seems the default setting for tcpdump is to set
the ring buffer to 32 entries of 64 Kb each, which implies 32 order 5
allocation.  Thats difficult under good circumstances, and horrid under memory
pressure.

I thought it would be good to make that a bit more usable.  I was going to do a
simple conversion of the ring buffer from contigous pages to iovecs, but
unfortunately, the metadata which AF_PACKET places in these buffers can easily
span a page boundary, and given that these buffers get mapped into user space,
and the data layout doesn't easily allow for a change to padding between frames
to avoid that, a simple iovec change is just going to break user space ABI
consistency.

So I've done this, I've added a three tiered mechanism to the af_packet set_ring
socket option.  It attempts to allocate memory in the following order:

1) Using __get_free_pages with GFP_NORETRY set, so as to fail quickly without
digging into swap

2) Using vmalloc

3) Using __get_free_pages with GFP_NORETRY clear, causing us to try as hard as
needed to get the memory

The effect is that we don't disturb the system as much when we're under load,
while still being able to conduct tcpdumps effectively.

Tested successfully by me.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Maciej Żenczykowski <zenczykowski@gmail.com>
Reported-by: Maciej Żenczykowski <zenczykowski@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/isdn/mISDN: Use printf extension %pV

Using %pV reduces the number of printk calls and
eliminates any possible message interleaving from
other printk calls.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

netlink: let nlmsg and nla functions take pointer-to-const args

The changed functions do not modify the NL messages and/or attributes
at all. They should use const (similar to strchr), so that callers
which have a const nlmsg/nlattr around can make use of them without
casting.

While at it, constify a data array.

Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv6: fix missing in6_ifa_put in addrconf

Fix ref count bug introduced by

commit 2de795707294972f6c34bae9de713e502c431296
Author: Lorenzo Colitti <lorenzo@google.com>
Date:   Wed Oct 27 18:16:49 2010 +0000

ipv6: addrconf: don't remove address state on ifdown if the address
is being kept

Fix logic so that addrconf_ifdown() decrements the inet6_ifaddr
refcnt correctly with in6_ifa_put().

Reported-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-next-2.6

net: Export netif_get_vlan_features().

ERROR: "netif_get_vlan_features" [drivers/net/xen-netfront.ko] undefined!

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

enic: Fix build warnings

Fix data type of argument passed to pci_alloc_consistent and pci_free_consistent routines.

Signed-off-by: Vasanthy Kolluri <vkolluri@cisco.com>
Signed-off-by: Roopa Prabhu <roprabhu@cisco.com>
Signed-off-by: David Wang <dwang2@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

hso: Fix unused variable warning

Fallout from the TIOCGICOUNT work

Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bridge: add RCU annotations to bridge port lookup

br_port_get() renamed to br_port_get_rtnl() to make clear RTNL is held.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bridge: fix RCU races with bridge port

The macro br_port_exists() is not enough protection when only
RCU is being used. There is a tiny race where other CPU has cleared port
handler hook, but is bridge port flag might still be set.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

netdev: add rcu annotations to receive handler hook

Suggested by Eric's bridge RCU changes.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bridge: add proper RCU annotation to should_route_hook

Add br_should_route_hook_t typedef, this is the only way we can
get a clean RCU implementation for function pointer.

Move route_hook to location where it is used.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bridge: add RCU annotation to bridge multicast table

Add modern __rcu annotatations to bridge multicast table.
Use newer hlist macros to avoid direct access to hlist internals.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/ipv6/mcast.c: Remove unnecessary semicolons

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

include/net/caif/cfctrl.h: Remove unnecessary semicolons

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Sjur Braendeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

include/linux/if_macvlan.h: Remove unnecessary semicolons

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/cnic.c: Remove unnecessary semicolons

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ixgbe: Remove unnecessary semicolons

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/e1000e: Remove unnecessary semicolons

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/bnx2x: Remove unnecessary semicolons

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/isdn: Remove unnecessary semicolons

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'dccp' of git://eden-feed.erg.abdn.ac.uk/net-next-2.6

net: Simplify RX queue allocation

This patch move RX queue allocation to alloc_netdev_mq and freeing of
the queues to free_netdev (symmetric to TX queue allocation).  Each
kobject RX queue takes a reference to the queue's device so that the
device can't be freed before all the kobjects have been released-- this
obviates the need for reference counts specific to RX queues.

Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: Move TX queue allocation to alloc_netdev_mq

TX queues are now allocated in alloc_netdev_mq and freed in
free_netdev.

Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

xfrm: use gre key as flow upper protocol info

The GRE Key field is intended to be used for identifying an individual
traffic flow within a tunnel. It is useful to be able to have XFRM
policy selector matches to have different policies for different
GRE tunnels.

Signed-off-by: Timo Teräs <timo.teras@iki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>

vlan: Fix build warning in vlandev_seq_show()

net/8021q/vlanproc.c: In function 'vlandev_seq_show':
net/8021q/vlanproc.c:283:20: warning: unused variable 'fmt'

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

carl9170: use generic sign_extend32

This patch replaces the handcrafted
sign extension cruft with a generic
bitop function.

Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

bitops: Provide generic sign_extend32 function

This patch moves code out from wireless drivers where two different
functions are defined in three code locations for the same purpose and
provides a common function to sign extend a 32-bit value.

Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

wl1251: use wl12xx_platform_data to pass data

Make use the newly added method to pass platform data for wl1251 too.
This allows to eliminate some redundant code.

Cc: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Acked-by: Kalle Valo <kvalo@adurom.com>
Acked-by: Luciano Coelho <luciano.coelho@nokia.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

wl1251: add runtime PM support for SDIO

Add runtime PM support, similar to how it's done for wl1271.
This allows to power down the card when the driver is loaded but
network is not in use.

Cc: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Acked-by: Kalle Valo <kvalo@adurom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

wl1251: add power callback to wl1251_if_operations

Call interface specific power callback before calling board specific
one. Also allow that callback to fail. This is how it's done for
wl1271 and will be used for runtime_pm support.

Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Acked-by: Kalle Valo <kvalo@adurom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlagn: enable shadow register

For 6000 series devices and up, enable automatic update MAC's register
for better power usage in PSP mode

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwifi: disable disconnected antenna for advanced bt coex

Disconnected antenna algorithm is used to find out which antennas are
disconnected. It should be disabled for devices that support advanced
bluetooth coexist.

Signed-off-by: Shanyu Zhao <shanyu.zhao@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwifi: seperate disconnected antenna function

Disconnected antenna algorithm is seperated into its own function from chain noise
calibration routine for better code management.

Signed-off-by: Shanyu Zhao <shanyu.zhao@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlagn: fix RXON HT

When the HT information is changed due to
BSS changes (like legacy stations joining)
we need to recalculate HT RXON parameters.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlagn: re-enable calibration

During the RXON rewrite, this code got lost.
When we've just associated, we need to enable
all calibrations and see if some were already
finished. Add back the missing code.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlagn: fix RXON issues

The RXON rework resulted in a massive loss of
throughput because we weren't programming the
device completely correctly -- the BSSID has
to be programmed into the device before the
AP station is uploaded. To fix this, simply
always send the unassoc RXON, i.e. even when
it was already unassoc so that the BSSID and
some other parameters are updated properly.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlagn: fix needed chains calculation

Garen noticed that this was wrong. Fix
the calibration -- default to multiple
chains and fall back to single where
possible.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwifi: resending QoS command when HT changes

"mac80211: Fix WMM driver queue configuration"
inadvertedly broke iwlwifi, because now mac80211
configures the QoS settings before assoc, and
therefore before HT. Thus, iwlwifi no longer told
the device about the HT setting, which it needs
to -- and thus throughput went down a lot. Fix
this by resending the QoS command to the device
not only when QoS/WMM settings change, but also
when HT changes.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwlifi: update rx write pointer w/o request mac access in the CAM mode

In iwl_rx_queue_update_write_ptr function
replace iwl_write_direct32 with iwl_write32 when not in power save mode.
We don't have to go through grab nic access as the NIC is already awake.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwifi: legacy tx_cmd_protection function

Legacy (4965 and 3945) devices has different tx_cmd_protection routine.
Move to iwl-legacy.c

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwifi: put all the isr related function under ops

There were two type of isr supported by iwlwifi devices.
  legacy isr - only used by legacy devices (3945 & 4965)
  ict isr - used by all new generation of iwlwifi devices

Move all the isr related functions into ops, the ict type of isr
supports only needed for newer devices.

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwifi: Legacy isr only used by legacy devices

Move iwl_isr_legacy function to iwl_legacy.c since it only used
by legacy devices.

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlagn: fix non-5000+ build

When building 4965 without 5000+ there were a
lot of build errors due to functions being used
that weren't even compiled in. To fix this move
some code around and only compile the HCMD code
for 5000+ series as it's not used for 4965.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlagn: update PCI ID for 100 series devices

Update the supported PCI ID list for 100 series devices

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlagn: update PCI ID for 6000g2a series devices

Update the supported PCI ID list for 6000g2a series devices

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlagn: update PCI ID for 6000g2b series devices

Update the supported PCI ID list for 6000g2b series devices

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath5k: Print stats as unsigned ints.

The debugfs code for ath5k was printing some unsigned int
stats with %d instead of %u.  This meant that you could see
negative numbers instead of a clean wrap.

Signed-off-by: Ben Greear <greearb@candelatech.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Add watchdog functions for HW queue

Add watchdog functions for managing the Queues inside the hardware.
Normally the driver doesn't have much to do with these queues
directly, but the Ralink drivers did implement watchdog functions
for these. These watchdog functions are not triggered that often,
compared to the other watchdog functions, but I have at least
seen them trigger once or twice during a long stresstest run.

v2: Add extra documentation for register fields

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Remove rt2x00lib_toggle_rx

As part of the queue refactoring, the rt2x00lib_toggle_rx
can be removed and replaced with the call directly to
the set_device_state callback function.

We can remove the STATE_RADIO_RX_ON_LINK and
STATE_RADIO_RX_OFF_LINK, as it was only used for
special behavior inside rt2x00lib rather then the
drivers.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Fix rt2x00queue_kick_tx_queue arguments

The queue_entry argument to rt2x00queue_kick_tx_queue,
doesn't make sense due to the function name (it is called
kick QUEUE)... But neither do we need the queue_entry, since
we need the data_queue.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Acked-by: Helmut Schaa <helmut.schaa@googlemail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Rename queue->lock to queue->index_lock

The queue->lock is only used to protect the index
numbers. Rename the lock accordingly.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt73usb.c

rt73usb.c:43: ERROR: do not initialise statics to 0 or NULL

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt2x00queue.c

rt2x00queue.c:804: ERROR: space prohibited after that open parenthesis '('
rt2x00queue.c:805: ERROR: space prohibited before that close parenthesis ')'

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt2x00link.c

rt2x00link.c:70: ERROR: space prohibited before that close parenthesis ')'

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt2x00lib.h

rt2x00lib.h:60: ERROR: space prohibited after that open parenthesis '('
rt2x00lib.h:60: ERROR: space prohibited before that close parenthesis ')'

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt2x00dev.c

rt2x00dev.c:689: ERROR: spaces required around that '=' (ctx:WxV)

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt2x00config.c

rt2x00config.c:136: ERROR: space required before the open parenthesis '('

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt2800usb.h

rt2800usb.h:43: ERROR: space prohibited after that open parenthesis '('
rt2800usb.h:43: ERROR: space prohibited before that close parenthesis ')'
rt2800usb.h:44: ERROR: space prohibited after that open parenthesis '('
rt2800usb.h:44: ERROR: space prohibited before that close parenthesis ')'

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt2800usb.c

 rt2800usb.c:48: ERROR: do not initialise statics to 0 or NULL

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt2800pci.h

rt2800pci.h:41: ERROR: Macros with complex values should be enclosed in parenthesis
rt2800pci.h:42: ERROR: Macros with complex values should be enclosed in parenthesis
rt2800pci.h:43: ERROR: Macros with complex values should be enclosed in parenthesis
rt2800pci.h:44: ERROR: Macros with complex values should be enclosed in parenthesis
rt2800pci.h:55: ERROR: space prohibited after that open parenthesis '('
rt2800pci.h:55: ERROR: space prohibited before that close parenthesis ')'
rt2800pci.h:56: ERROR: space prohibited after that open parenthesis '('
rt2800pci.h:56: ERROR: space prohibited before that close parenthesis ')'

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt2800lib.c

rt2800lib.c:831: ERROR: inline keyword should sit between storage class and type

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt2800.h

rt2800.h:1511: ERROR: space prohibited after that open parenthesis '('
rt2800.h:1511: ERROR: space prohibited before that close parenthesis ')'
rt2800.h:1513: ERROR: space prohibited after that open parenthesis '('
rt2800.h:1513: ERROR: space prohibited before that close parenthesis ')'
rt2800.h:1515: ERROR: space prohibited after that open parenthesis '('
rt2800.h:1515: ERROR: space prohibited before that close parenthesis ')'
rt2800.h:1517: ERROR: space prohibited after that open parenthesis '('
rt2800.h:1517: ERROR: space prohibited before that close parenthesis ')'
rt2800.h:1519: ERROR: space prohibited after that open parenthesis '('
rt2800.h:1519: ERROR: space prohibited before that close parenthesis ')'
rt2800.h:1521: ERROR: space prohibited after that open parenthesis '('
rt2800.h:1521: ERROR: space prohibited before that close parenthesis ')'
rt2800.h:1661: ERROR: space prohibited after that open parenthesis '('
rt2800.h:1661: ERROR: space prohibited before that close parenthesis ')'
rt2800.h:1662: ERROR: space prohibited after that open parenthesis '('
rt2800.h:1662: ERROR: space prohibited before that close parenthesis ')'
rt2800.h:1663: ERROR: space prohibited before that close parenthesis ')'
rt2800.h:2013: ERROR: space prohibited after that open parenthesis '('
rt2800.h:2013: ERROR: space prohibited before that close parenthesis ')'
rt2800.h:2014: ERROR: space prohibited after that open parenthesis '('
rt2800.h:2014: ERROR: space prohibited before that close parenthesis ')'

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt2500usb.c

rt2500usb.c:42: ERROR: do not initialise statics to 0 or NULL

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt2500pci.h

rt2500pci.h:1091: ERROR: space prohibited after that open parenthesis '('
rt2500pci.h:1091: ERROR: space prohibited before that close parenthesis ')'
rt2500pci.h:1092: ERROR: space prohibited after that open parenthesis '('
rt2500pci.h:1092: ERROR: space prohibited before that close parenthesis ')'

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: checkpatch.pl error fixes for rt2400pci.h

rt2400pci.h:812: ERROR: space prohibited after that open parenthesis '('
rt2400pci.h:812: ERROR: space prohibited before that close parenthesis ')'
rt2400pci.h:813: ERROR: space prohibited after that open parenthesis '('
rt2400pci.h:813: ERROR: space prohibited before that close parenthesis ')'
rt2400pci.h:950: ERROR: Macros with complex values should be enclosed in parenthesis

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

carl9170: tx path review

This patch fixes a few shortcomings in the tx path.

 * move temp. ampdu_[ack]_len out of txinfo->pad.

 * fix WARN_ON from tx.c:line 300 when tx_ampdu_queue
   fails to queue the frame.

 * In tx_prepare, we already have a local pointer
   to the station's ieee80211_sta struct.

 * remove a second !sta check, tx_prepare already takes
   care of that.

Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

mac80211: Remove redundant checks for NULL before calls to crypto_free_cipher()

crypto_free_cipher() is a wrapper around crypto_free_tfm() which is a
wrapper around crypto_destroy_tfm() and the latter can handle being passed
a NULL pointer, so checking for NULL in the
ieee80211_aes_key_free()/ieee80211_aes_cmac_key_free() wrappers around
crypto_free_cipher() is pointless and just increase object code size
needlesly and makes us execute extra test/branch instructions that we
don't need.
Btw; don't we have to many wrappers around wrappers ad nauseam here?
Anyway, this patch removes the redundant conditionals.

Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Fix comments in rt73usb.h and rt61pci.h

This patch fixes a few comments in rt73usb.h and rt61pci.h.

Signed-off-by: Lalith Suresh <suresh.lalith@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Fix hw crypto in AP mode for some devices

The BSSID register shouldn't be set in AP mode on some older devices (like
rt73usb) as it breaks hw crypto on these. However, rt2800 devices explicitly
need the BSSID register set to the same value as our own MAC address (only
in AP mode).

Hence, don't set the BSSID from rt2x00lib but move it down into rt2800 to
avoid problems on older devices.

This fixes a regression (at least for rt73usb) and avoids a new regression
for rt2800 devices in 2.6.36.

Reported-by: Johannes Stezenbach <js@sig21.net>
Reported-by: Lee <lee-in-berlin@web.de>
Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Fix crash on USB unplug

By not scheduling the TX/RX completion worker threads
when Radio is disabled, or hardware has been unplugged,
the queues cannot be completely cleaned.

This causes crashes when the hardware has been unplugged while
the radio is still enabled.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Fix MCU_SLEEP arguments

Legacy driver uses 0xff as the second argument for the MCU_SLEEP
command. It is still unknown what the values actually mean, but
this will at least keep the command in-sync with the original
driver.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Implement flush callback

Implement a basic flush callback function, which simply loops
over all TX queues and waits until all frames have been transmitted
and the status reports have been gathered.

At this moment we don't support dropping any frames during the
flush, but mac80211 will only send 'false' for this argument anyway,
so this is not important at this time.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Acked-by: Helmut Schaa <helmut.schaa@googlemail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Remove failsave from rt2x00usb_watchdog_tx_dma

When the TX status handler failed to clear the queue
in rt2x00usb_watchdog_tx_dma() we shouldn't use a failsave
to use the rt2x00usb txdone handler.

If a driver has overriden the txdone handler it must make
sure the txdone handler is capable of cleaning up the queue itself.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Rename rt2x00queue_timeout

Rename rt2x00queue_timeout to rt2x00queue_status_timeout to
better describe what is actually timing out (note that
we already have a rt2x00queue_dma_timeout).

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Acked-by: Helmut Schaa <helmut.schaa@googlemail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Optimize rt2x00debug_dump_frame when frame dumping is not active

When rt2x00 is compiled with debugging but frame dumping is currently
not active we can avoid the call to do_gettimeofday. Furthermore,
frame dumping is not the default case, mark it as unlikely.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Reduce tx descriptor size

The tx descriptor values qid, cw_min, cw_max and aifs are directly
accessible through the tx entry struct. So there's no need to copy
them into the tx descriptor and passing them to the indiviual drivers.
Instead we can just get the correct value from the tx entry.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Wait up to one second on rt2800 for WPDMA to be ready

At least some devices need such a long time to inititalize WPDMA. This
only increases the maximum wait time and shouldn't affect devices that
have been working before.

Reported-by: Joshua Smith <jesmith@kaon.com>
Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Sync Tx and RX ring sizes with legacy drivers

All rt2x00 devices used the same Tx and Rx ring size (24 entries) till
now. Newer devices (like rt2800) can however make use of a larger TX and
RX ring due to 11n capabilities (AMPDUs of size 64 for example).

Hence, bring rt2x00 in sync with the legacy drivers and use the same TX
and RX ring sizes. Also remove the global defines RX_ENTRIES, TX_ENTRIES,
BEACON_ENTRIES and ATIM_ENTRIES and use per driver values.

That is 24 entries for rt2400pci, 32 entries for rt2500pci, rt2500usb,
rt61pci and rt73usb and 128 (RX) and 64 (TX) for rt2800pci and rt2800usb.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rt2x00: Add TXOP_CTRL_CFG register definition

Remove the magic value initialisation of the TXOP_CTRL_CFG register by
defining its fields and using them during intialisation. The field
RESERVED_TRUN_EN is referred to as reserved, however it is set to 1 by
the legacy drivers. Hence, do the same.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: rework tx queue selection and fix queue stopping/waking

The current ath9k tx queue handling code showed a few issues that could
lead to locking issues, tx stalls due to stopped queues, and maybe even
DMA issues.

The main source of these issues is that in some places the queue is
selected via skb queue mapping in places where this mapping may no
longer be valid. One such place is when data frames are transmitted via
the CAB queue (for powersave buffered frames). This is made even worse
by a lookup WMM AC values from the assigned tx queue (which is
undefined for the CAB queue).

This messed up the pending frame counting, which in turn caused issues
with queues getting stopped, but not woken again.

To fix these issues, this patch removes an unnecessary abstraction
separating a driver internal queue number from the skb queue number
(not to be confused with the hardware queue number).

It seems that this abstraction may have been necessary because of tx
queue preinitialization from the initvals. This patch avoids breakage
here by pushing the software <-> hardware queue mapping to the function
that assigns the tx queues and redefining the WMM AC definitions to
match the numbers used by mac80211 (also affects ath9k_htc).

To ensure consistency wrt. pending frame count tracking, these counters
are moved to the ath_txq struct, updated with the txq lock held, but
only where the tx queue selected by the skb queue map actually matches
the tx queue used by the driver for the frame.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Reported-by: Björn Smedman <bjorn.smedman@venatech.se>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

b43legacy: rfkill: use status register based on core revision (not PHY's)

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

zd1201: Add missing id

The Mandriva patch seems to stem from 2.6.14, so much for their
upstreaming effort.

Didn't find another Linux reference of it, just an omnious
"USB\VID_1044&PID_8004" from GigabyteZD1201U.INF for
Gigabyte GN-WLBZ101 802.11b USB Adapter, which matches the
Mandriva patch comment.

Aboves file also lists an "USB\VID_1044&PID_8006", which I have
kept appart as this "Gigabyte GN-WBZB-M 802.11b USB Adapter"
didn't show up in googling.

Signed-off-by: maximilian attems <max@stro.at>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

zd1211rw: add 2 missing usb id's

"These USB ID came from Palnex <http://www.planex.co.jp/>
Worked fine." says Mandriva patch for their 2.6.32 and earlier.

Web has evidence for both id's to work, so just add them upstream:
http://www.mail-archive.com/zd1211-devs@lists.sourceforge.net/msg00507.html
http://ubuntuforums.org/showthread.php?t=473046

Signed-off-by: Go Taniguchi <go@turbolinux.co.jp>
Signed-off-by: maximilian attems <max@stro.at>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rtl8187: restore anaparam registers after reset with 8187B

Current 8187B initialization misses anaparam registers restore after
8187 reset. This causes ANAPARAM register to stay zeroed out (ANAPARAM2
kept its value on my tests). To avoid this, call rtl8187_set_anaparam
right after chip reset (to be on the safe side, as it makes sure we
restore all ANAPARAM registers).

Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Cc: seno <senada@t-online.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rtl8187: remove uneeded setting of anaparam write

Usually you set RTL818X_CONFIG3_ANAPARAM_WRITE when you are going to
change/write ANAPARAM registers. But in current initialization of
RTL8187B there is a place where ANAPARAM_WRITE bit is set without any
ANAPARAM register being written, without reason, so remove it.

Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Cc: seno <senada@t-online.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rtl8187: consolidate anaparam on/off write sequences

There are repeated calls for anaparam on/off sequence in the code.
Consolidate the common code in rtl8187_set_anaparam and use it where
needed.

Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rtl8187: don't set RTL818X_CONFIG3_GNT_SELECT flag on 8187B

The GNTSel bit should only concern pci devices by looking at RTL8180
spec, which is not the case of 8187B. Also testing shows that trying to
set this bit fails, a subsequent read from the register after trying to
set it shows that the bit isn't set, seems the hardware ignores it,
which makes sense. This setting was a left over from Realtek sources.

Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rtl8187: move pll reset at start out of ANAPARAM write

On 8187B start, comment about pll reset, and move it out of ANAPARAM
write sequence, so that code is more readable.

Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rtl8187: avoid redundant write to register FF72 (RFSW_CTRL)

The table with misc register initialization was setting it, and later
on we would set it again with a explicity call to rtl818x_iowrite16_idx.

Remove duplicate initialization from the register table.

Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rtl8187: fix wrong register initialization in 8187B

We were using wrong address for BRSR (Basic Rate Set Register) while
initializing its value, comparing with Realtek sources, for 8187B case.

Also, the same register is initialized in rtl8187b_reg_table, so remove
the duplicate initialization from the table.

Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rtl8187: remove setting of beacon/atim registers from initialization

On 8187B path, we set a initial value for beacon interval and atim
window on initialization. But this isn't needed, since same setup is
done on rtl8187_config.

Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

rtl8187: remove redundant initialization of ARFR

This removes redundant write to Auto Rate Fallback Register on RTL8187B.
The same value was being written twice in the same function. Avoid this
removing the duplicate initialization on rtl8187b_reg_table, and also
add comment for this write (information from Realtek source).

Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

mwl8k: force AP mode to use non-AMPDU frames

AP firmware uses xmitcontrol to differentiate between AMPDU
and non-AMPDU frames. As the support for AMPDU is not yet
added, set xmitcontrol to non-AMPDU for all tx frames for AP
firmware. This field will be set to indicate ampdu/non-ampdu
frames when tx AMPDU support is added.

Signed-off-by: Pradeep Nemavat <pnemavat@marvell.com>
Signed-off-by: Brian Cavagnolo <brian@cozybit.com>
Acked-by: Lennert Buytenhek <buytenh@wantstofly.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

carl9170: configurable beacon rates

Previously, the beacon rate was fixed to either:
 * 1Mb/s [2.4GHz band]
 * 6Mb/s [5GHz band]

This limitation has been addressed and now the
beacon rate is selected by ieee80211_tx_info's
rate control info, almost like any ordinary
data frame.

Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

carl9170: stop stale uplink BA sessions

This patch fixes a possible lengthy stall if the device
is operating as an experimental 11n AP and an STA
[during heavy txrx action] suddenly signalized to go
off-channel (old NetworkManager), or (sleep - which is
unlikely, because then it wouldn't be *active* at all!?).

Because the driver has to manage the BA Window, the
sudden PSM transition can leave active uplink BA
sessions to the STA in a bad state and a proper
cleanup is needed.

Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

carl9170: fix spurious restart due to high latency

RX Stress tests of unidirectional bulk traffic with
bitrates of up to 220Mbit/s have revealed that the
fatal-event recovery logic [which was solely triggered
by an out-of-rx-buffer situation] is too aggressive.

The new method now "pings" the device and then
decides - based on the response - whenever
a restart is needed or not.

Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

carl9170: initialize HW aMPDU parameters properly

This patch changes the initial aMPDU density and
factor settings to match those of Otus.

Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

carl9170: import hw/fw header updates

This patch imports all shared header changes
from carl9170fw.git.

 * add some strategic __aligned(4).
   This allows the compiler generate optimized code for
   architectures which can't access (unaligned/packed)
   data efficiently.
   ("ath9k_hw: optimize all descriptor access functions")

 * add a forgotten __CARL9170FW__ ifdef around
   a private firmware-internal struct.

 * GET_VAL macro helper
   Very useful for extracting data out of the
   bit-packed PHY registers.

 * cosmetic changes
   e.g.: _CCA_MINCCA_ to just _CCA_MIN_.

 * version bump 1.8.8.3 -> 1.9.0.

Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

mac80211_hwsim: Incorporate txpower into rssi

Up to now mac80211_hwsim has been reporting an rssi of -50. This patch
improves the model slightly by returning txpower-50. This makes it
easy to stimulate tests that need to see a varying rssi.

Signed-off-by: Blaise Gassend <blaise@willowgarage.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>