compat-wireless: add sched.h on libertas/firmware.c

sched.h is required to compile users of wake_up(),
firmware.c somehow gets sched.h included on newer kernels
but on older kernels this is not happening so include
this on the file itself. This patch should and will
be sent upstream.

This fixes this compilation error:

/home/mcgrof/devel/compat-wireless/drivers/net/wireless/libertas/firmware.c: In function ‘lbs_fw_loaded’:
/home/mcgrof/devel/compat-wireless/drivers/net/wireless/libertas/firmware.c:26:29: error: ‘TASK_NORMAL’ undeclared (first use in this function)
/home/mcgrof/devel/compat-wireless/drivers/net/wireless/libertas/firmware.c:26:29: note: each undeclared identifier is reported only once for each function it appears in
/home/mcgrof/devel/compat-wireless/drivers/net/wireless/libertas/firmware.c: In function ‘lbs_wait_for_firmware_load’:
/home/mcgrof/devel/compat-wireless/drivers/net/wireless/libertas/firmware.c:110:260: error: ‘TASK_UNINTERRUPTIBLE’ undeclared (first use in this function)
/home/mcgrof/devel/compat-wireless/drivers/net/wireless/libertas/firmware.c:110:2: error: implicit declaration of function ‘schedule’ [-Werror=implicit-function-declaration]
cc1: some warnings being treated as errors

make[4]: *** [/home/mcgrof/devel/compat-wireless/drivers/net/wireless/libertas/firmware.o] Error 1

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: few more pr_fmt fixes for drivers

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches for next-20120423

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: add an extra include to ath6kl/htc_pipe.c

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: build alx for kernel > 2.6.27

When compiling against kernel <= 2.6.27 alx has build errors so use
atl1c there instead. The logic was wrong before.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: deactivate CONFIG_BT_RFCOMM_TTY for old kernel versions.

net/bluetooth/rfcomm/tty.c does not build on older kernel versions.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: ath6kl: add config options for SDIO and USB

ath6kl now supports USB and SDIO based devices, this patch adds the
config options for them.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: brcmfmac: add usb config option

This activates the usb part of brcmfmac.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: ti: fix some problems with the config options

After the ti drivers where reorganized some config options in config.mk
where wrong. This patch corrects them.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: ath6kl: deactivate suspend/resume only for SDIO

If ath6kl is used on a USB device suspend/resume should work, only on
SDIO based devices this causes some problems with older kernel versions.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches with next-20120419

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: address moving of TI drivers

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: split patches/01-netdev.patch up

This splits up patches/01-netdev.patch into the following
set of patches:

patches/0001-netdev_ops.patch
patches/0002-net-misc.patch
patches/0003-netdev-needed_headroom_tailroom.patch
patches/0004-wext-namespace.patch

This isolates the network device operations infrastructure
into a separate file to show one collateral evolution [0].
The next objective will be to extract SmPL out of the patch
patches/0001-netdev_ops.patch. If we can extract SmPL [1] out
of it, with the help of compat as a backend store to help
us backport the work required, we could automatically backport
this collateral evolution to any network driver. Backporting
this specific collateral evolution then on compat-wireless
would simply require adding the driver's directory to be
copied into the framework, a one line change.

If we keep doing this for more collateral evolutions the
implications are the ability to enhance automatically
backporting even more collateral evolutions and therefore
drivers.

mcgrof@flash ~/devel/compat-wireless (git::master)$ time ckmake ; time ckmake
Trying kernel                     3.2.2-030202-generic  [OK]
Trying kernel                    3.1.10-030110-generic  [OK]
Trying kernel                    3.0.18-030018-generic  [OK]
Trying kernel                  2.6.39-02063904-generic  [OK]
Trying kernel                  2.6.38-02063808-generic  [OK]
Trying kernel                  2.6.37-02063706-generic  [OK]
Trying kernel                  2.6.36-02063604-generic  [OK]
Trying kernel                  2.6.35-02063512-generic  [OK]
Trying kernel                  2.6.34-02063410-generic  [OK]
Trying kernel                  2.6.33-02063305-generic  [OK]
Trying kernel                  2.6.32-02063255-generic  [OK]
Trying kernel                  2.6.31-02063113-generic  [OK]
Trying kernel                  2.6.30-02063010-generic  [OK]
Trying kernel                  2.6.29-02062906-generic  [OK]
Trying kernel                  2.6.28-02062810-generic  [OK]
Trying kernel                    2.6.27-020627-generic  [OK]
Trying kernel                    2.6.26-020626-generic  [OK]
Trying kernel                    2.6.25-020625-generic  [OK]
Trying kernel                    2.6.24-020624-generic  [OK]

real    85m10.426s
user    258m22.297s
sys     22m55.582s

real    14m37.117s
user    32m18.885s
sys     7m4.599s

[0] http://coccinelle.lip6.fr/ce.php
[1] http://coccinelle.lip6.fr/

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches for next-20120410

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix patches/01-netdev.patch

This was caused by these patches:

Author: Francois Romieu <romieu@fr.zoreil.com>
Date:   Fri Mar 9 19:28:28 2012 +0100

    atl1e: stop using net_device.{base_addr, irq}.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Cc: Jay Cliburn <jcliburn@gmail.com>
Cc: Chris Snook <chris.snook@gmail.com>
Author: Francois Romieu <romieu@fr.zoreil.com>
Date:   Fri Mar 9 19:22:31 2012 +0100

    atl1c: stop using net_device.{base_addr, irq}.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Cc: Jay Cliburn <jcliburn@gmail.com>
Cc: Chris Snook <chris.snook@gmail.com>
I cannot find an explanation to the removal of the assignments:

netdev->base_addr = (unsigned long)adapter->hw.hw_addr;
netdev->irq  = pdev->irq;

even on the public mailing lists, and the commit logs do not
explain *why* these are being removed. For now just leave this
as is...

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: simplify mac80211 queue backport

With compat now having a define for netif_wake_subqueue
for kernels 2.6.23 and higher we can remove most of the
backport patch for multiqueue. This does give up on
kernels 2.6.22 and older though.

mcgrof@flash ~/devel/compat-wireless (git::master)$ time ckmake
Trying kernel                     3.2.2-030202-generic  [OK]
Trying kernel                    3.1.10-030110-generic  [OK]
Trying kernel                    3.0.18-030018-generic  [OK]
Trying kernel                  2.6.39-02063904-generic  [OK]
Trying kernel                  2.6.38-02063808-generic  [OK]
Trying kernel                  2.6.37-02063706-generic  [OK]
Trying kernel                  2.6.36-02063604-generic  [OK]
Trying kernel                  2.6.35-02063512-generic  [OK]
Trying kernel                  2.6.34-02063410-generic  [OK]
Trying kernel                  2.6.33-02063305-generic  [OK]
Trying kernel                  2.6.32-02063255-generic  [OK]
Trying kernel                  2.6.31-02063113-generic  [OK]
Trying kernel                  2.6.30-02063010-generic  [OK]
Trying kernel                  2.6.29-02062906-generic  [OK]
Trying kernel                  2.6.28-02062810-generic  [OK]
Trying kernel                    2.6.27-020627-generic  [OK]
Trying kernel                    2.6.26-020626-generic  [OK]
Trying kernel                    2.6.25-020625-generic  [OK]
Trying kernel                    2.6.24-020624-generic  [OK]

real    27m43.716s
user    69m9.115s
sys     13m8.241s

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: backport get_ts_info support

Linux 3.5 will have get_ts_info to support the Precision Time Protocol.

http://linuxptp.sourceforge.net/
http://en.wikipedia.org/wiki/Precision_Time_Protocol

We cannot backport this support given that this introduces
a data structure change on the ethtool_ops.

commit c8f3a8c31069137fe0100e6920558f1a7487ef3c
Author: Richard Cochran <richardcochran@gmail.com>
Date:   Tue Apr 3 22:59:17 2012 +0000

    ethtool: Introduce a method for getting time stamping capabilities.

    This commit adds a new ethtool ioctl that exposes the SO_TIMESTAMPING
    capabilities of a network interface. In addition, user space programs
    can use this ioctl to discover the PTP Hardware Clock (PHC) device
    associated with the interface.

    Since software receive time stamps are handled by the stack, the generic
    ethtool code can answer the query correctly in case the MAC or PHY
    drivers lack special time stamping features.

Signed-off-by: Richard Cochran <richardcochran@gmail.com>
Reviewed-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
mcgrof@flash ~/devel/compat-wireless (git::master)$ time ckmake; time ckmake
Trying kernel                     3.2.2-030202-generic  [OK]
Trying kernel                    3.1.10-030110-generic  [OK]
Trying kernel                    3.0.18-030018-generic  [OK]
Trying kernel                  2.6.39-02063904-generic  [OK]
Trying kernel                  2.6.38-02063808-generic  [OK]
Trying kernel                  2.6.37-02063706-generic  [OK]
Trying kernel                  2.6.36-02063604-generic  [OK]
Trying kernel                  2.6.35-02063512-generic  [OK]
Trying kernel                  2.6.34-02063410-generic  [OK]
Trying kernel                  2.6.33-02063305-generic  [OK]
Trying kernel                  2.6.32-02063255-generic  [OK]
Trying kernel                  2.6.31-02063113-generic  [OK]
Trying kernel                  2.6.30-02063010-generic  [OK]
Trying kernel                  2.6.29-02062906-generic  [OK]
Trying kernel                  2.6.28-02062810-generic  [OK]
Trying kernel                    2.6.27-020627-generic  [OK]
Trying kernel                    2.6.26-020626-generic  [OK]
Trying kernel                    2.6.25-020625-generic  [OK]
Trying kernel                    2.6.24-020624-generic  [OK]

real    70m16.460s
user    204m37.315s
sys     19m56.447s

real    15m35.818s
user    33m10.020s
sys     7m19.779s

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: add export to commented out config vars

This makes it easier to enable config items correctly.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: provide URL for code metrics archive

This uses the Google Fusion Tables to keep track of code
metrics accross releases. What would be nice is if someone
adds the code required to upload stats automatically for
each release.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: skip forcing .config include

We know that .config will exist but it will not exist on
the make clean target. So skip that.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix pr_fmt compile warnings

This took a while to figure out... Thought this was impossible
to fix, but the fix is quite simple: fix this upstream. The
assumption by drivers when upstream is that you should #define
pr_fmt *prior* to including <linux/printk.h> but -- the assumption
is incorrect given that if a subsystem includes printk.h for
anything that the subsystem needs then the module would be defining
the pr_fmt. The proper usage then for pr_fmt then is to always
undef it first, and also include <linux/printk.h>.

When test cross compiling against 2.6.24 - 3.3 I notice that including
<linux/kernel.h> also helps, so do this for now. There may be
another reason why that is required but we can figure that out later.

This patch should be broken down into chunks and submitted upstream.
This shaves off compilation time down from 1 hour to 1/2 hour!

mcgrof@flash ~/devel/compat-wireless (git::master)$ time ckmake
Trying kernel                     3.2.2-030202-generic  [OK]
Trying kernel                    3.1.10-030110-generic  [OK]
Trying kernel                    3.0.18-030018-generic  [OK]
Trying kernel                  2.6.39-02063904-generic  [OK]
Trying kernel                  2.6.38-02063808-generic  [OK]
Trying kernel                  2.6.37-02063706-generic  [OK]
Trying kernel                  2.6.36-02063604-generic  [OK]
Trying kernel                  2.6.35-02063512-generic  [OK]
Trying kernel                  2.6.34-02063410-generic  [OK]
Trying kernel                  2.6.33-02063305-generic  [OK]
Trying kernel                  2.6.32-02063255-generic  [OK]
Trying kernel                  2.6.31-02063113-generic  [OK]
Trying kernel                  2.6.30-02063010-generic  [OK]
Trying kernel                  2.6.29-02062906-generic  [OK]
Trying kernel                  2.6.28-02062810-generic  [OK]
Trying kernel                    2.6.27-020627-generic  [OK]
Trying kernel                    2.6.26-020626-generic  [OK]
Trying kernel                    2.6.25-020625-generic  [OK]
Trying kernel                    2.6.24-020624-generic  [OK]

real    29m17.269s
user    75m56.693s
sys     10m48.469s

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: backport assinging the tty dev to hci dev

This backports assinging the tty device to the hci parent
pointer. The goal was to make the devices link under /sys/
but on older kernels we cannot support this. One alternative
that occurs to me is to have a compat subsystem where userspace
applications can hunt for such newer devices, but this requires
quite a bit of work and consideration.

commit 7f4b2b04c88377af30c022f36c060190182850fb
Author: Andrei Warkentin <andreiw@motorola.com>
Date:   Fri Feb 11 17:19:26 2011 -0600

    Bluetooth: Make hci a child of the corresponding tty device.

    Make /sys/class/bluetooth/hciX a symlink to
    path under corresponding tty.

Signed-off-by: Andrei Warkentin <andreiw@motorola.com>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches for next-20120402

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches for next-20120321

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix patches/16-bluetooth.patch

Remove hunk 2 that modifies drivers/bluetooth/hci_ldisc.c
The hunk as trying to do this:

-+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,37)
-       hdev->parent = hu->tty->dev;
-+#endif

This patch makes this not necessary anymore:

commit 68478a9a06c9336fe0e2039a658092ac25935b0d
Author: David Herrmann <dh.herrmann@googlemail.com>
Date:   Fri Mar 9 15:53:42 2012 +0100

    Bluetooth: Remove redundant hdev->parent field

Turns out SET_HCIDEV_DEV() is defined on older kernels
since v2.6.18-rc1~21 (via a91f2e39). So all we have to do
is remove this hunk. For newer kernels this will covert
itself accordingly, for older kernels we have the old macro
helper.

mcgrof@tux ~/linux-stable (git::master)$ git describe --contains a91f2e39
v2.6.18-rc1~21^2~3

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: add more files to ignore

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix compilation of ath6kl on 2.6.27

Some drivers rely on some other part of the kernel to include
in.h but on older kernels this proves to not work. Explicitly
include that file. This fixes compilation of ath6kl on 2.6.27.
This patch will be sent upstream though.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix p54 pm ops backport typo again

Really..

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix p54 pm ops backport typo

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: reorder config includes

We need to include first the kernel config as otherwise
we end up trying to compile drivers which were disabled
for older kernels. This fixes compilation on 2.6.28 where
carl9170 was being compiled although we had disabled it
on config.mk.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: add pm ops backport for p54

This should fix compiling on 2.6.28.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: do not use broad export on makefiles

This inspired by Johannes' patch on compat which does the
same thing.

perl -pi -ne 's|^(\s*)CONFIG|$1export CONFIG|' config.mk

And then we also address the autoconf generation.

We shave off compile time by more than 1/2:

Before:
real 0m34.034s
user 0m33.760s
sys 0m10.580s

After:
real 0m16.617s
user 0m22.290s
sys 0m6.180s

This was against 2.6.38.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix compilation of ath6kl on 2.6.34

This fixes compilation of compat-wireless completely on 2.6.34

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix typo in patches/40-netdev-hw-features.patch

This fixes compilation for 2.6.35 for ath6kl.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: backport netdev->addr_assign_type

You cannot backport assignment of netdev->addr_assign_type
given that its part of the netdev data structure only in future
kernels, so we backport this through a new patch file:

patches/61-netdev-addr_assign_type.patch

mcgrof@tux ~/linux-next (git::master)$ git describe --contains c1f79426
v2.6.36-rc1~571^2~104

This fixes compilation of the Atheros Ethernet drivers down to 2.6.35.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix patches/40-netdev-hw-features.patch

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix patches/37-vsnprintk.patch

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh linux-next-pending patches

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix linux-next-pending/0002-backport-alx.patch

The last hunk was must be placed in context, patch
is stupid and will only accept hunks in order.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: backport addr_assign_type usage

The alx driver uses the addr_assign_type variable from
the netdev data structure which was added as of 2.6.35:

mcgrof@tux ~/linux-next (git::master)$ git describe --contains c1f79426
v2.6.36-rc1~571^2~104

Although compat.git backported dev_hw_addr_random() the patch
in compat.git cannot address usage of addr_assign_type and as
such requires an ifdef around it.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix jobserver

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: enable mac80211 tracing for kernels >= 2.6.33

Turns out this compiles now that we are in the future, so just
disable tracing for older kernels. This goes unstested though.

Requested-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix patch for disabling ath9k RC

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: add driver-select option ath9k_ap

Prefer using Minstrel for AP for ath9k. The minstrel
rate control algorihtm has been field tested more than
the ath9k rate control algorithm. To disable ath9k's RC
we renaem the config option to with a CONFIG_COMPAT prefix
and disable it later.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches

This runs: ./scripts/admin-update.sh refresh

This refreshes all hunk offests automatically.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix patches/09-threaded-irq.patch for wl12xx

Hunk offsets differ for wl12xx.h changes. This is now fixed.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: backport ath6kl hw_features / [set|fix]_features

This requires a manual backport, we throw it into
patches/40-netdev-hw-features.patch following previous
ethernet backport of Atheros Ethernet drivers.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: add Makefile to PHONY

Debugging Make shows that 18 variations of Makefile are checked
for targets on the Makefile prior to process the first real target.
Stop this nonsense.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: remove recurssive call on building modules

The recursion call to build modules is not required. It was added
due to the fact that we now included a new file, CONFIG_COMPAT
and we needed it at build time. We do not need the recursive call
given that the kernel will already call the same Makefile later
when building modules. We take advantage of this fact and simply
ensure we build CONFIG_COMPAT when processing our rules.

We do not define a direct target for CONFIG_COMPAT given that
we want to force a regeneration of the file and compat autoconf,
whenever the build environment changes. We still add a dummy
CONFIG_COMPAT target rule to avoid Make looking rules for it.

In order for this build to work we now just need to export
CFLAGS so that the variables defined on the first pass are
available later at module build time.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: use PWD for file version details

The CFLAGS were using some local files to add some definition
of some variables later used by the compat module. Lets be
careful to use only the locally defined files.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: use PWD for file definition

Instead of using PWD in random places just use it
to defint he direct file once. This will allow us
to later move variables that use PWD to another
place.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: Handle name change from iwlagn to iwlwifi in 3.2

With linux-3.2 the iwlagn driver is renamed iwlwifi.  For some kernels
(like 3.1.5) this causes the non-compat iwlagn.ko to try and run against
the compat mac80211 resulting in an oops.  This change prevents the
non-compat iwlagn.ko from being loaded.

Signed-off-by: Don Fry <donald.h.fry@intel.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches

Synchs with next-20120306

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: disable b44 for Atheros drivers via driver-select

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: remove -e from depmod on modlib.sh

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix make install

This issue was introduced via this patch:

commit 802ea47e6316aea285f2c30d95763dd7c320eec1
Author: Rick Farina <sidhayn@gmail.com>
Date:   Fri Feb 24 21:28:40 2012 -0500

    compat-wireless: replace modprobe -l with functional equivilent

Fix the regression by properly using brackets on the
conditional check and also guaranteeing we output something
upon success. Not sure how this could have worked.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: add alx Ethernet driver to linux-next-pending/

This is now posted so we merge it. We can remove this once
David merges the driver and it gets into linux-next.git.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: add MAINTAINERS to distribution

This adds the MAINTAINERS file to the distribution.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix directory copying in admin-update.sh

On my system the trailing slash leads to the directory contents being copied
to the target instead of the full directory, which results in breakage during
build.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: use portable #! for bash scripts

Some systems (e.g. FreeBSD) do not have bash in /bin
Based on a patch by Jo-Philipp Wich

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: export MAKE

This is usefulf to propagate the same MAKE
variable preference over scripts used.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix old kernel checks for 3.x

These ancient checks require a check against the 2 release
as the KERNEL_26SUBLEVEL is only defined when that kernel
is detected.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: replace modprobe -l with functional equivilent

modprobe -l is either deprecated or removed, so let's replace it
testing [ -z "$(DESTDIR)" ] appeared to only be need due to the fact
that *enable was run directly instead of being run with $(DESTDIR)/

Testing for module presence should now work properly on modern distros
and *enable will be run properly with or without $(DESTDIR) being set.

Signed-off-by: Rick Farina <sidhayn@gmail.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: use compat's CONFIG_COMPAT_KERNEL_* variables

compat already deals with defining CONFIG_COMPAT_KERNEL_* variables
for us for both makefiles and C / header files. The C / header file
usage was fixed in the last commit, this removes internal usage we
had for the Makefiles in preference over the same mechanism.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: start using CONFIG_COMPAT_KERNEL_ in code

This lets use the defined CONFIG_COMPAT_KERNEL_* in C code and
header files. Next step is to remove the Makefile hackery to
defines CONFIG_COMPAT_KERNEL_* stuff as we have a generated
file for us now.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix old checks on KERNEL_SUBLEVEL

KERNEL_SUBLEVEL used to be used on config.mk for th 2.6.x series
but after 3.x support was added the variable should have been
changed to KERNEL_26SUBLEVEL.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

cleanup modprobe calls

There are a lot of needless calls to "modprobe -l <drivername>"
None of this is needed, and it errors when modprobe -l is removed
(such as when KMOD replaces modutils). Since modprobe -l is already
legacy, let's just remove it.

Signed-off-by: Rick Farina <sidhayn@gmail.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches for backport alx

alx is not yet upstream but it will be soon I hope.
This updates the backport of it.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix patches/24-pcmcia.patch

Some hunk offsets were off.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix patches/16-bluetooth.patch

Some context lines were removed from the upstream driver.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix 08-rename-config-options.patch

Hunk conflicts due to changes on the Makefile for the wl1251 driver.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches

This synchs with next-20120217.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: update bluetooth patch to latest wireless-testing

There is a small change in the bt_sock_unregister() function in
af_bluetooth (an extra if) that cause the patch not to apply anymore.

Signed-off-by: Luciano Coelho <coelho@ti.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: remove olpc_ec_wakeup_* calls for pre-3.1 kernels

This section of the libertas driver calls functions that simply don't
exist before the release of 3.1.  This code in question was an addition,
not a change from any existing code.  It is safe to simply remove it for
older kernels.

Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: add alx driver and backport

This adds backport support for the alx driver.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches with next-20120210

Synchs with next-20120210

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: remove patches/50-function-namespace.patch

This is now upstream. Boy that was fast.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix patches/44-deactivate-mac80211-tracing.patch

Tracing stuff is now compiled-in if tracing is enabled in your
build, but we keep it disabled regardless for compat.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: rename path_lookup() to mpath_lookup()

This is required for compiling compat-wireless down to 2.6.24.
Don't ask me why, I just wanted to get this done. This patch
will be sent upstream.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: add missing =y to CONFIG_RT2X00_LIB_DEBUGFS

When editing config.mk with a sed script the missing =y on this line
causes issue. Since this line is commented out by default this means no
changes to the default behavior and sed scripts will be able to edit the
config.mk file much more easily.

Signed-of-By: Rick Farina <sidhayn@gmail.com>

compat-wireless: Calling make directly is not permitted

make was being called directly, fix to $(MAKE)

Signed-of-By: Rick Farina <sidhayn@gmail.com>

compat-wireless: add Ethernet drivers to driver-select

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: update language on scripts/admin-update.sh

Use the actual git tree name. The term "linux-next.git based"
does more justice than "bleeding edge based". It also helps educate
users / developers about the source of the code.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: refresh patches

This applies on top of next-20120131.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix patches/99-change-makefiles.patch

CONFIG_USB_NET_QMI_WWAN was added and therefore the
respective Makefile, drivers/net/usb/Makefile, needs
to be adjusted to remove that from our list of targets.

If we user our own generated Makefiles for drivers the
advantage is that we would not face this issue when upstream
gets a new driver that our hunk needs to ignore but -- the
disadvantage is that any changes upstream would require a
respective change into compat-wireless. Driver Makefiles
though typically are just copied as-is and never modified.
Its subsystem Makefiles that we tend to have to patch on
this file.

One option would be to somehow annotate upstream targets as
compat-wireless maintained so that we can script out what
targets to leave in.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireles: remove ath9k packet log crap patch

This has been sitting in crap/ directory for ages. Time
to go. Too bad. Someone will have to get this upstream.

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix some config options

Adds some missing config options interceded in the last time and remove
some old now unused config options.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: rename path_lookup

path_lookup in net/mac80211/mesh_pathtbl.c consolidates with
path_lookup in include/linux/namei.h used in old kernel versions.

This should also go into compat-wireless-stable-3.3.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: use skb_get_queue_mapping() in b43 and b43legacy

Use skb_get_queue_mapping() for getting the queue_mapping member of
skb. Some old kernels do not have the member queue_mapping with all
configuration options, but this function always returns something.

This should also go into compat-wireless-stable-3.3.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: backport dev_pm_ops for bcma

bcma now also uses dev_pm_ops which does not work with older kernel
versions. Just backport it like it was done for the other drivers.

This should also go into compat-wireless-stable-3.3.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: fix wl12xx driver-select

wl12xx is a possible choice for both group drivers and
singular modules (making it impossible to build only
the wl12xx module). change the group name to "ti" instead.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: use linux-stable instead of linux-2.6-allstable

I used to use hpa's linux-2.6-allstable.git tree to make
stable releases of compat-wireless. This allowed me to
make updates to the stable releases whenever a 2.6.x.y
extraversion got out -- but also to make updates whenever
Linus updated his own tree so I can make RC releases
based on the latest RC release. Greg has taken over of
that tree and its now called linux-stable.git. What I am
seeing though is that the linux-stable git tree now seems
to not update as often to Linus' tree. This is fine as all
I have to do is add Linus' tree as a remote and fetch
those objects and then reset the tree to say, v3.2-rc6
and that'll get what Linus' tree looks like today.

This patch just changes the path where we pick up stable
releases from but also documents this change in case
anyone that wants to make stable release must also add
Linus' tree as a remote and fetch those objects.

Cc: greg@kroah.com
CC: kvalo@qca.qualcomm.com
Cc: linville@tuxdriver.com
Cc: senthilb@qca.qualcomm.com
Cc: johannes@sipsolutions.net
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

Enable CONFIG_RT2800{PCI,USB}_RT53XX

Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wireless: check if modprobe exists before using it

If you're cross-compiling compat-wireless on a different host, you
can't reliably use modprobe.  The host system may not even have
modprobe, so it doesn't make sense to run modprobe, unless you're
installing the modules on the target machine.

The modprobe command was being used unconditionally to set the MADWIFI
and OLD_IWL variables.  This commit adds a check to see whether
modprobe exists before trying to use.  It also adds a dependency on
the modprobe commands to all recipes that rely on it, so we fail more
gracefully if modprobe doesn't exist in the system.

Reported by: William List <wlist@ti.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>

compat-wireless: build: Throw error if space in path

Because of make limitations, the build will fail if there are spaces in
the path leading to the compat-wireless directory.
I tried to make it work anyway for a while and, while it probably is
possible, it would involve rewriting part of kbuild and well...
This at least throws up nicely if there is a space in the path.

Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>

compat-wirless: refresh patches

Signed-off-by: Luis R. Rodriguez <mcgrof@frijolero.org>