hostapd: Add uapsd option to netifd.sh

The uapsd option sets the uapsd_advertisement_enabled flag in hostapd.

The check for phy support is already implemented here in hostapd since 2011:
http://w1.fi/cgit/hostap/commit/?id=70619a5d8a3d32faa43d66bcb1b670cacf0c243e

So this can be safely set to 1 as default.

Signed-off-by: Vittorio Gambaletta <openwrt@vittgam.net>
SVN-Revision: 43846

ar8216: factor out AR8327/AR8337-specific driver code into ar8327.c

Move all AR8327/AR8337-specific driver code into a separate source file
ar8327.c and adjust patches so that ar8327.c is compiled if
CONFIG_AR8216_PHY is set.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43845

ar8216: move definitions from ar8216.c to ar8216.h and introduce ar8327.h

Move several structure definitions and #defines from ar8216.c
to ar8216.h and move AR8327/AR8337 header stuff into a new
header file ar8327.h.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43844

kernel/hwmon: package pwm-fan kernel module

pwm-fan was introduced in kernel 3.18

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
SVN-Revision: 43843

all targets: remove all =m kernel config symbols

modules build during the kernel compile phase are ignored anyway,
all modules should be built using KernelPackage in
package/kernel/linux/modules/*
selecting the appropriate config symbols there rather than in
target/linux/*/config-*

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
SVN-Revision: 43842

kernel: bridge: multicast: backport a few more fixes for 3.10

The following patches unfortunately didn't hit the kernel stable
branches yet, therefore cherrypicking them for OpenWRT here:

* bridge: fix netfilter/NF_BR_LOCAL_OUT for own, locally generated queries
* bridge: multicast: enable snooping on general queries only
* bridge: multicast: add sanity check for general query destination

Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
SVN-Revision: 43841

octeon: enable VLAN support

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
SVN-Revision: 43840

octeon: remove config-default

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
SVN-Revision: 43839

kernel/modules: add kmod-crypto-gcm

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
SVN-Revision: 43838

kernel/modules: add kmod-crypto-ghash

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
SVN-Revision: 43837

kernel/modules: kmod-crypto-xts: depend on kmod-crypto-gf128

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
SVN-Revision: 43836

kernel/modules: add kmod-crypto-gf128

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
SVN-Revision: 43835

kernel/modules: add kmod-crypto-ctr

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
SVN-Revision: 43834

kernel/modules: add kmod-crypto-seqiv

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
SVN-Revision: 43833

kernel/modules: remove CRYPTOMGR_MODULES

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
SVN-Revision: 43832

kernel/modules: kmod-crypto-manager: depend on kmod-crypto-aead

SVN-Revision: 43831

kernel/modules: add kmod-crypto-aead

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
SVN-Revision: 43830

kernel/modules: remove superfluous backslash

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
SVN-Revision: 43829

util-linux: fix packaging issues

Signed-off-by: Gergely Kiss <mail.gery@gmail.com>
Tested-by: Gergely Kiss <mail.gery@gmail.com>
SVN-Revision: 43828

et131x is out of staging in 3.18

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 43827

upgrade to latest revision

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 43826

add tmp421 to the device tree instead of relying on autodetect

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 43825

openvpn: backport an upstream fix for a regression in using --cipher none (fixes #18676)

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43823

kernel: update 3.18 to 3.18.1

Signed-off-by: Luka Perkov <luka@openwrt.org>
SVN-Revision: 43821

kernel: update 3.14 to 3.14.27

Signed-off-by: Luka Perkov <luka@openwrt.org>
SVN-Revision: 43820

ramips: fix WSR-600DHP flash layout

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43819

ramips: fix WSR-1166DHP flash layout

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43818

ramips: fix WSR-{600,1166} names

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43817

ramips: fix wsr-1166 sysupgrade support

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43816

ramips: on mt7621 only enable xhci for the eval board (still needs patch cleanup to take effect)

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43815

ramips: fix WSR-600 image format

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43814

ramips: use CONFIG_MTD_SPLIT_UIMAGE_FW instead of CONFIG_MTD_UIMAGE_SPLIT

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43813

ramips: fix image format for WSR-1166

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43812

ramips: disable ethernet checksum offloads for mt7621, they are broken

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43811

ramips: use netif_receive_skb instead of napi_gro_receive when rx csum offload is unavailable

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43810

kernel: add mtd split support for trx images

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43809

ramips: rework and fix m25p80 chunked-io support

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43808

ramips: split mt7621 spi into a separate driver, increase maximum transfer size

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43807

kernel: add a mtdsplit implementation for tp-link firmware headers

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43806

kernel: move mtdsplit files to drivers/mtd/mtdsplit/ to simplify maintenance, unify patches across kernel versions

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43805

honor the bootloader's bootargs

Only overwrite the bootloader's bootargs if "chosen" exists in the dtb

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 43802

enable wpad-mini by default

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 43801

bcm53xx: support flashing CHK and CyberTAN images

SVN-Revision: 43800

brcm47xx: support for flashing CHK and CyberTAN images

This will allow simple sysupgrade-s including simpler reverting to the
original firmware.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 43799

brcm47xx: verify if CHK/CyberTAN contains valid TRX

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 43798

mpc83xx: add support for kernel 3.14

This is compile tested only, please run test and report back.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 43797

mpc52xx: add support for kernel 3.14

This is compile tested only, please run test and report back.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 43796

ep93xx: add support for kernel 3.14

This is compile tested only, please run test and report back.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 43795

kernel: add some missing config options

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 43794

tools: build libtool after autoconf, not before (fixes #18659)

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43793

ipq806x: refresh patches

This makes the patchs apply again.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 43792

upgrade to new rev, fixes possible compile error

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 43791

ath10k: fix tx rate selection for ad-hoc mode with HT

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43790

oxnas: re-enable ubifs

After the changes make ubinize.cfg optional in image.mk, ubifs
generation can now safely be enabled again.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
SVN-Revision: 43789

image.mk: make ubinize.cfg optional

When using UbinizeImage with ubifs rootfs, ubinize.cfg is no longer
needed. Yet, the absance of ubinize.cfg would make the build process
abort with an error.
Fix that by checking if ubinize.cfg is present and do no not call the
"classic" ubinize image generation if it isn't.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
[missing new-line before UbinizeImage added itentionally]

SVN-Revision: 43788

oxnas: clk-oxnas: rework pllb enable function

kernel lock debugging unveiled that we should not call
of_reset_control_get inside a clock's enable operation (see below)

move of_reset_control_* previously used in pllb_clk_enable to new
pllb_clk_prepare and pllb_clk_unprepare functions.
use a container to carry runtime information.

------------[ cut here ]------------
WARNING: CPU: 0 PID: 1 at kernel/locking/lockdep.c:2742 lockdep_trace_alloc+0xb8/0xfc()
DEBUG_LOCKS_WARN_ON(irqs_disabled_flags(flags))
Modules linked in:
CPU: 0 PID: 1 Comm: swapper/0 Not tainted 3.14.26 #6
[<c001a6ac>] (unwind_backtrace) from [<c0016dec>] (show_stack+0x10/0x14)
[<c0016dec>] (show_stack) from [<c0194f68>] (dump_stack+0x7c/0x94)
[<c0194f68>] (dump_stack) from [<c0021b50>] (warn_slowpath_common+0x68/0x8c)
[<c0021b50>] (warn_slowpath_common) from [<c0021ba4>] (warn_slowpath_fmt+0x30/0x40)
[<c0021ba4>] (warn_slowpath_fmt) from [<c0061b30>] (lockdep_trace_alloc+0xb8/0xfc)
[<c0061b30>] (lockdep_trace_alloc) from [<c00cb740>] (kmem_cache_alloc+0x1c/0xf8)
[<c00cb740>] (kmem_cache_alloc) from [<c01d33c8>] (of_reset_control_get+0xe8/0x12c)
[<c01d33c8>] (of_reset_control_get) from [<c0269228>] (pllb_clk_enable+0x14/0xbc)
[<c0269228>] (pllb_clk_enable) from [<c0265738>] (__clk_enable+0x54/0xa0)
[<c0265738>] (__clk_enable) from [<c0265acc>] (clk_enable+0x18/0x2c)
[<c0265acc>] (clk_enable) from [<c04325f8>] (oxnas_pcie_probe+0x3b8/0x6a0)
[<c04325f8>] (oxnas_pcie_probe) from [<c01f2510>] (platform_drv_probe+0x18/0x48)
[<c01f2510>] (platform_drv_probe) from [<c01f1070>] (driver_probe_device+0xd8/0x24c)
[<c01f1070>] (driver_probe_device) from [<c01f1298>] (__driver_attach+0x70/0x94)
[<c01f1298>] (__driver_attach) from [<c01ef728>] (bus_for_each_dev+0x4c/0x98)
[<c01ef728>] (bus_for_each_dev) from [<c01f0818>] (bus_add_driver+0xcc/0x1e8)
[<c01f0818>] (bus_add_driver) from [<c01f169c>] (driver_register+0xa0/0xe8)
[<c01f169c>] (driver_register) from [<c01f2568>] (platform_driver_probe+0x20/0xa4)
[<c01f2568>] (platform_driver_probe) from [<c0013a3c>] (do_one_initcall+0x90/0x140)
[<c0013a3c>] (do_one_initcall) from [<c0421d38>] (kernel_init_freeable+0x1e4/0x2c0)
[<c0421d38>] (kernel_init_freeable) from [<c000c214>] (kernel_init+0x8/0x104)
[<c000c214>] (kernel_init) from [<c0008768>] (ret_from_fork+0x14/0x2c)
---[ end trace 5f17ed2f61e0683f ]---

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
SVN-Revision: 43787

kernel/generic: fix warnings if libata-ledtrig is disabled

accidentally #if was used instead of #ifdef, thus gcc warned
about CONFIG_ATA_LEDS being undefined...

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
SVN-Revision: 43786

binutils: run install twice instead of overwriting the final toolchain with the initial one

fixes compiler error after a binutils rebuild/reinstall without
rebuilding the rest of the toolchain

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43785

build: use gcc-provided ar, nm and ranlib where appropriate

Since GCC 4.7, GCC provides its own wrappers around ar, nm and ranlib, which
should be used for builds with link-time optimization. Since GCC 4.9, using them
actually necessary for LTO builds using convenience libraries to succeed.

There are some packages which try to automatically detect if gcc-{ar,nm,ranlib}
exist (one example is my package "fastd" in the package repository, which tries
to use LTO). This breaks because the OpenWrt build system explicitly sets the
binutils versions of these tools.

As it doesn't cause any issues to use gcc-{ar,nm,ranlib} instead of
{ar,nm,ranlib} even without LTO, this patch just makes OpenWrt use the
GCC-provided versions by default, which fixes the build of such packages with
GCC 4.9.

(I know that builds fail though when clang is used with -flto and
gcc-{ar,nm,ranlib}, but as all OpenWrt toolchains are based on GCC, this isn't
a real issue.)

Completely cleaning the tree (or at least `make clean toolchain/clean`) is
necessary to get a consistent state after the binutils plugins support patch and
this one (as trying to use gcc-{ar,nm,ranlib} with a binutils built without
plugin support will definitely lead to a build failure).

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
SVN-Revision: 43784

binutils: enable plugin support

--enable-plugin is necessary for gcc-ar, gcc-nm and gcc-ranlib to work, which
must be used with GCC 4.9 for LTO to work.

Without this option, gcc-ar etc. will just fail with

    sorry - this program has been built without plugin support

Using the normal ar from binutils with GCC 4.9 and -flto will cause linking
with static "convenience" libraries to fail.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
SVN-Revision: 43783

hostapd: add support for configuring supported rates

patch by Wilco Baan Hofman from #18627

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43782

fix mvneta vlan tagging

This is a 3.14.x backport of:
"mvneta: Fix TSO and checksum for non-acceleration vlan traffic"

Upstream commits:

817dbfa5d1bc276a72c1a577310382008e8aca0a
2d39d120781a5770573dc6ed672a5a562f541aea

Fixes vlan tagging on SoC interfaces for the WRT1900AC.  Useful
when using mvsw6171 driver.

Signed-off-by: Mark Milburn <mark.a.milburn@gmail.com>
Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 43781

don't copy initramfs.elf unconditionally

targets that need it should do it in their image Makefile

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 43780

only call UbinizeImage and BuildSysUpgrade for boards we support them on

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 43779

The ubinized filesystems should be used for generating the final images, so store them in KDIR instead.

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 43778

ar71xx: add flush reset register writes, should fix restart issues (#17839)

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43777

kernel: backport a few PHY layer fixes from upstream

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43776

ar71xx: rename/refresh patch

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43775

we've changed places with Luka, reflect that :)

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 43774

use mwlwifi by default

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 43773

Marvell 88W8864 open source mac80211 driver

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 43772

netifd: minor fixes, add mldversion option

thanks to Hans Dedecker

Signed-off-by: Steven Barth <steven@midlink.org>
SVN-Revision: 43771

mac80211: fix HT mode selection for ad-hoc

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43769

oxnas: sata_oxnas: make irq handler more readable

safed one level of indention by using 'continue' instead of a
lengthy if-clause.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
v2: use logic-AND instead of '?' operator when checking for hw bug 6320

SVN-Revision: 43768

oxnas: sata_oxnas: refactoring phase 2

 - replaced // comments by /* comments */
 - added line-breaks where needed
 - fixed white-space according to kernel style
 - fixed some obvious spelling mistakes in comments and printks
 - removed some unneeded left-overs imported from vendor code-base
 - replaced printk(...) by libata macros where possible

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
SVN-Revision: 43767

oxnas: sata_oxnas: refactoring phase 1

locking for 2nd port and hwraid was added from vendor's GPL code which
doesn't comply with current kernel coding style.

 - moved all global variables into host_priv
 - renamed locks
 - sanetized acquire() and release() parameter list

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
SVN-Revision: 43766

oxnas: refresh kernel config-3.1[48]

and no longer compile-in ext4 filesystem now that sata_oxnas is
also built as a module.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
SVN-Revision: 43765

kernel: remove openwrt micrel.c (replaced by upstream driver)

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43762

ubi-utils: add mirror md5sum

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43761

tools/mtd-utils: add mirror md5sum

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43760

dnsmasq: also add the actual patches...

Signed-off-by: Steven Barth <steven@midlink.org>
SVN-Revision: 43759

dnsmasq: backport some dnssec fixes

Signed-off-by: Steven Barth <steven@midlink.org>
SVN-Revision: 43758

kernel: disable soft-reset for generic PHY devices to avoid accidentally clearing preinitialized state

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43757

buffalo-enc: pass the longstate option to decryption

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43756

buffalo-enc: accept the longstate command line option

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43755

bcm53xx: support WAN connected to switch port 0

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 43754

mvsw6171: note support for 88E6172 switches

The '6171 and '6172 are similar enough to work
without any changes to the code.

Signed-off-by: Claudio Leite <leitec@staticky.com>
SVN-Revision: 43753

kirkwood: add *plugs to uci-defaults

Set defaults health led to a timer heartbeet (200/800)
and populate network for:

- GuruplugServerPlus
- SheevaPlug
- SheevaPlug with eSATA

Signed-off-by: Gerald Kerma <dreagle@doukki.net>
SVN-Revision: 43752

kikwood: populate kirkwood.sh with new boards

- SheevaPlug
- SheevaPlug with eSATA
- GuruplugServerPlus

Signed-off-by: Gerald Kerma <dreagle@doukki.net>
SVN-Revision: 43751

kirkwood: set default EA4500 switch config

This maps the LAN ports to eth0 and WAN port to eth1.
Since there are two interfaces the 802.1q VLAN mode
is unnecessary and left disabled. Port-based VLANs
are used instead.

Signed-off-by: Claudio Leite <leitec@staticky.com>
SVN-Revision: 43750

kernel: bgmac: fix some problems

There are some problems in the napi usage that are fixed now.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 43749

bcm53xx: PCIe: add link check again

When there is no device connected to the controller we should not start
the controller. This also changes between PCIe 1.0 and PCIe 2.0 speed
based on the device. This code is based on code from the new Broadcom
PCIe controller driver.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 43748

uboot-envtools: add more kirkwood boards

- SheevaPlug
- SheevaPlug with eSATA
- GuruplugServerPlus

Signed-off-by: Gerald Kerma <dreagle@doukki.net>
SVN-Revision: 43747

ar7: remove leftover 3.10 patches

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 43746

binutils: break-out libbfd, libopcodes and ar

Package libbfd, libopcodes and ar seperately.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
SVN-Revision: 43745

ar71xx: add support for qca956x soc

This patch adds soc support for QCA9561 and TP9343.
TP9343 is a reduced version of QCA9561, which can be found in TP-LINK routers in China.
The qca956x_wmac has not yet been supported by ath9k.

tested on TL-WDR6500 and TL-WR882N v1 (Chinese version)

Signed-off-by: Weijie Gao <hackpascal@gmail.com>
SVN-Revision: 43744

ar8216: Inline function ar8xxx_create_mii

Inline function ar8xxx_create_mii.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43743

ar8216: Remove read/write/rmw member functions from ar8xxx_priv

Remove read/write/rmw member functions from ar8xxx_priv

There seems to be no real benefit of the ar8xxx_priv member functions
read/write/rmw as one implementation exists for each of them only.
Especially ar8xxx_mii_rmw is assigned to priv->rmw first and then
mapped to ar8xxx_rmw.
Rename the ar8xxx_mii_.. functions to ar8xxx_.. and use them directly.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43742

ar8216: Create helpers mii_read32 / mii_write32 for 32 bit MII ops

Create helpers mii_read32 / mii_write32 for 32 bit MII ops.
Rename r3 variable to page in ar8xxx_mii_write to make it consistent
with the other ar8xxx_mii_xxxx functions.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43741

ar8216: Factor out chip-specific parameters from ar8xxx_probe_switch

Factor out chip-specific parameters from ar8xxx_probe_switch.
Move the ar8xxx_chip definitions after the swops definitions.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 43740