backports: add missing include for is_valid_ether_addr()

is_valid_ether_addr() is used, but the header here it is declared was
not included.

This fixes a problem introduced in commit:
commit 521acfd92f168fc9b8b6a1cefe9a165243f3d872
Author: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Date:   Thu Aug 15 14:40:09 2013 +0200

    backports: add eth_prepare_mac_addr_change() and eth_commit_mac_addr_change()

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Cc: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: usb_translate_errors() is needed for < 3.2

usb_translate_errors() is needed for every kernel < 3.2 and not <
2.6.39.

This fixes a problem introduced in this commit:
commit aee2dc1d7fd03d3900e06e88793b519e21049de0
Author: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Date:   Thu Aug 15 14:40:20 2013 +0200

    backports: add usb_translate_errors()

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Cc: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: fix conditional include of arch_phys_wc_add()

This fixes a problem introduced in this commit:
commit c871780b5afa182878884bf5ccd8df4817a2660f
Author: Arik Nemtsov <arik@wizery.com>
Date:   Wed Aug 14 10:48:05 2013 +0300

    backports: rename some mem functions to not break custom kernels

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Cc: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: cw1200 needs the multicast API changes to support <=2.6.35

Also, update the dependencies:

 cw1200_wlan_sdio needs >=2.6.34 due to missing sdio_set_host_pm_flags()
 cw1200_wlan_spi needs >=2.6.35 due to missing request_any_context_irq()

Signed-off-by: Solomon Peachy <pizza@shaftnet.org>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: cw1200 utilizes kthreads for <= 2.6.35

The new workqueue APIs in 2.6.36 allowed us to scrap the old kthread worker
that had been used; unfortunately the backported API isn't complete so
in order to support older kernels, resurrect this patch.

Signed-off-by: Solomon Peachy <pizza@shaftnet.org>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: move the cw1200 sdio quirk into 26-sdio-quirks

Signed-off-by: Solomon Peachy <pizza@shaftnet.org>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backport: Add an implementation of get_random_int() for <3.10 kernels.

get_random_int() was present, but simply not exported for use in modules
until 3.10.  Implement it in terms of the more expensive get_random_bytes()

This is needed by the cw1200 driver.

Signed-off-by: Solomon Peachy <pizza@shaftnet.org>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: rename some mem functions to not break custom kernels

When custom patches are cherry-picked to a kernel, some symbols exported
by backports may clash with the built-in ones. Rename the backports
symbols using the standard backport_ prefix to prevent that.

The offending symbols were exported by the patch below:

commit 2ce5c22448bb45998318267c00b5d6ef9cff3170
Author: Hauke Mehrtens <hauke@hauke-m.de>
Date:   Thu Jun 6 13:48:04 2013 +0200

    backports: backport some memory functions

Signed-off-by: Arik Nemtsov <arik@wizery.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: Add patch for cw1200_sdio with kernels < 3.2

Without this patch, the hardware *will* hang on <3.2 kernels when a
512-byte SDIO transfer occurs.

3.2-rc added an SDIO quirk to work around this, so this code was
stripped from the driver when committed to the mainline.

Signed-off-by: Solomon Peachy <pizza@shaftnet.org>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: Include drivers/net/wireless/cw1200

The cw1200 WLAN driver was added in 3.11-rc; include it in backports.

Signed-off-by: Solomon Peachy <pizza@shaftnet.org>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add defconfig for wwan device drivers

Signed-off-by: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add wwan device drivers

Add cdc_mbim, cdc_ncm, sierra_net, qmi_wwan device drivers.

Signed-off-by: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: define phys_addr_t for ARM on kernel version < 2.6.25

phys_addr_t is not defined on ARM systems if kernel version < 2.6.25.

Signed-off-by: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: include headers directly if math64 header is missing

If math64 header is not available, include div64.h directly. It is required for
do_div() that is used inside the backport version of math64.h.

Signed-off-by: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: update enum usb_device_speed with USB_SPEED_SUPER

USB_SPEED_SUPER is missing for kernel versions < 2.6.30 but used by uvc_video and
usbnet.

Signed-off-by: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add usb_endpoint_maxp()

This function is missing for kernel version < 3.2 but required by uvc_video
and cdc-wdm device drivers.

Signed-off-by: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add usb_translate_errors()

This inline function is missing for version < 2.6.39 but required by em28xx
and cdc-wdm device driver.

Signed-off-by: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: define PM_EVENT_AUTO

This macro is missing for < 2.6.25

Signed-off-by: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add eth_prepare_mac_addr_change() and eth_commit_mac_addr_change()

These functions are required by qmi_wwan device driver.

Signed-off-by: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: fix build errors for old kernel versions without PCI support

Signed-off-by: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: fix device_release_driver

The new macro, introduced by

commit 9f5876fa9cfb8cd6f450c200043ca7e31915f976
Author: Luis R. Rodriguez <mcgrof@do-not-panic.com>
Date:   Thu Jul 18 16:40:55 2013 -0700

    backports: backport drvdata = NULL core driver fixes

evaluates its argument multiple times, causing crashes as
there are potential side-effects of the called code and/or
the arguments.

Fix this by converting it to static inlines.

Also remove the driver_probe_device() macro that has the
same issue, but tries to wrap a function that isn't even
available outside the driver core code in any kernel.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: refresh patches for next-20130719

== ckmake-report.log ==

1   2.6.24              [  OK  ]
2   2.6.25              [  OK  ]
3   2.6.26              [  OK  ]
4   2.6.27              [  OK  ]
5   2.6.28              [  OK  ]
6   2.6.29              [  OK  ]
7   2.6.30              [  OK  ]
8   2.6.31              [  OK  ]
9   2.6.32              [  OK  ]
10  2.6.33              [  OK  ]
11  2.6.34              [  OK  ]
12  2.6.35              [  OK  ]
13  2.6.36              [  OK  ]
14  2.6.37              [  OK  ]
15  2.6.38              [  OK  ]
16  2.6.39              [  OK  ]
17  3.0.89              [  OK  ]
18  3.1.10              [  OK  ]
19  3.10.5              [  OK  ]
20  3.11-rc4            [  OK  ]
21  3.2.50              [  OK  ]
22  3.3.8               [  OK  ]
23  3.4.56              [  OK  ]
24  3.5.7               [  OK  ]
25  3.6.11              [  OK  ]
26  3.7.10              [  OK  ]
27  3.8.13              [  OK  ]
28  3.9.11              [  OK  ]

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: deactivate REGULATOR_WM8400

This needs regulator_map_voltage_linear_range() and
regulator_list_voltage_linear_range() which are not exported by the
kernel regulator framework.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add of_find_node_by_name() if !CONFIG_OF

The old kernel only exports this function when CONFIG_OF is set, but not
when CONFIG_OF is not set.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add hid log functions

This adds some hid log functions like hid_warn().

This backports the following commit:
commit 4291ee305e9bb0699504a66f0e2b7aefcf0512a5
Author: Joe Perches <joe@perches.com>
Date:   Thu Dec 9 19:29:03 2010 -0800

    HID: Add and use hid_<level>: dev_<level> equivalents

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: try to backport acpi_video_register_with_quirks()

Original the i915 driver called acpi_video_register(), but that was
replaced with acpi_video_register_with_quirks(), which we can not
backport easily, replace it with the old call to
acpi_video_register() again.

Thos backports the following commit:
commit 8c5bd7adb2ce47e6aa39d17b2375f69b0c0aa255
Author: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Date:   Thu Jul 18 02:08:06 2013 +0200

    ACPI / video / i915: No ACPI backlight if firmware expects Windows 8

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: backport PTR_ERR_OR_ZERO

This backports this patch:
commit 6e8b8726ad503214ba66e34aed69aff41de33489
Author: Rusty Russell <rusty@rustcorp.com.au>
Date:   Mon Jul 15 11:19:32 2013 +0930

    PTR_RET is now PTR_ERR_OR_ZERO

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: update kernel version to test against

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: the shrinkers api did not made it into 3.11

Change the conditions to use the backported code also in 3.11

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backprots: fix logic in calling intel_enable_gtt()

The old patch changed the or condition to an and condition which is
wrong. This patch fixes the logic again.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: remove some remaining parts form old alx driver

This removed some old parts remaining from the old alx driver.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: refresh patches for next-20130712

1   2.6.24              [  OK  ]
2   2.6.25              [  OK  ]
3   2.6.26              [  OK  ]
4   2.6.27              [  OK  ]
5   2.6.28              [  OK  ]
6   2.6.29              [  OK  ]
7   2.6.30              [  OK  ]
8   2.6.31              [  OK  ]
9   2.6.32              [  OK  ]
10  2.6.33              [  OK  ]
11  2.6.34              [  OK  ]
12  2.6.35              [  OK  ]
13  2.6.36              [  OK  ]
14  2.6.37              [  OK  ]
15  2.6.38              [  OK  ]
16  2.6.39              [  OK  ]
17  3.0.79              [  OK  ]
18  3.1.10              [  OK  ]
19  3.10-rc1            [  OK  ]
20  3.2.45              [  OK  ]
21  3.3.8               [  OK  ]
22  3.4.46              [  OK  ]
23  3.5.7               [  OK  ]
24  3.6.11              [  OK  ]
25  3.7.10              [  OK  ]
26  3.8.13              [  OK  ]
27  3.9.3               [  OK  ]

real    34m34.417s
user    933m25.264s
sys     129m41.580s

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: refresh patches for next-20130709

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: copy over s2mps11.h

This is required by s2mps11 as of at least next-20130709.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: fix patches for next-20130709

A hunk context was off for alx.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: refresh patches for next-20130703

1   2.6.24              [  OK  ]
2   2.6.25              [  OK  ]
3   2.6.26              [  OK  ]
4   2.6.27              [  OK  ]
5   2.6.28              [  OK  ]
6   2.6.29              [  OK  ]
7   2.6.30              [  OK  ]
8   2.6.31              [  OK  ]
9   2.6.32              [  OK  ]
10  2.6.33              [  OK  ]
11  2.6.34              [  OK  ]
12  2.6.35              [  OK  ]
13  2.6.36              [  OK  ]
14  2.6.37              [  OK  ]
15  2.6.38              [  OK  ]
16  2.6.39              [  OK  ]
17  3.0.79              [  OK  ]
18  3.1.10              [  OK  ]
19  3.10-rc1            [  OK  ]
20  3.2.45              [  OK  ]
21  3.3.8               [  OK  ]
22  3.4.46              [  OK  ]
23  3.5.7               [  OK  ]
24  3.6.11              [  OK  ]
25  3.7.10              [  OK  ]
26  3.8.13              [  OK  ]
27  3.9.3               [  OK  ]

real    32m59.877s
user    880m31.524s
sys     124m11.996s

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: backport cross-device reservation support

This backports cross-device reservation support.
Given that this feature is built around the
CONFIG_DMA_SHARED_BUFFER and given that some older kernels
will have DMA_SHARED_BUFFER without cross device reservation
support we can't use the c-file and h-file backports Kconfig
trick to automatically backport this feature from the
target git tree.

commit 786d7257e537da0674c02e16e3b30a44665d1cee
Author: Maarten Lankhorst <m.b.lankhorst@gmail.com>
Date:   Thu Jun 27 13:48:16 2013 +0200

    reservation: cross-device reservation support, v4

    This adds support for a generic reservations framework that can be
    hooked up to ttm and dma-buf and allows easy sharing of reservations
    across devices.

    The idea is that a dma-buf and ttm object both will get a pointer
    to a struct reservation_object, which has to be reserved before
    anything is done with the contents of the dma-buf.

    Changes since v1:
     - Fix locking issue in ticket_reserve, which could cause
       mutex_unlock
       to be called too many times.
    Changes since v2:
     - All fence related calls and members have been taken out for now,
       what's left is the bare minimum to be useful for ttm locking conversion.
    Changes since v3:
     - Removed helper functions too. The documentation has an example
       implementation for locking. With the move to ww_mutex there is no
       need to have much logic any more.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Cc: maarten.lankhorst@canonical.com
Cc: jglisse@redhat.com
Cc: airlied@redhat.com
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: backport ww_mutex support

This backports the kernel's wound/wait style locks 040a0a371,
using the linux-stable v3.11-rc2 as a base for development.
Given the complexity to support debugging mutexes this backport
implementation is simplified by only making this feature availabe
if you to have DEBUG_MUTEXES and DEBUG_LOCK_ALLOC disabled.

Given that ww mutex is required for DRM this also means we must
update the kconfig for DRM and require you to also not be able to build
DRM if you have either of these options enabled. Support for
DEBUG_MUTEXES and DEBUG_LOCK_ALLOC can be added later by anyone
daring. This uses the new dependencies file kconfig language
extension to specify the backport feature build restrictions
for DRM.

Part of the ww mutex addition to the kernel required modifying
the fast path mutex locking scheme by requiring you to deal
with the slow path alternatives on your own (refer to a41b56ef).
The reason for this change was that the mutex fastpath implementation
assumed your slowpath alternative can only be passed one argument
and the addition of ww mutexes requires dealing with the slow
path with a context passed.

It'd be painful to backport all asm for an optimized fastpath
implementation so we penalize the backport ww mutex fast path
by using the generic atomic_dec_return().

To backport a clean our own mutex_lock_common() with the least
amount of changes against upstream commits 2bd2c92c and 41fcb9f2
also needed to be backported. Commit 2bd2c92c dealt with adding
support for queue mutex spinners with an MCS lock, since this
cannot be backported for older kernels we provide empty inlines.
Commit 41fcb9f2 just removed SCHED_FEAT_OWNER_SPIN as it was an
early hack, the only thing required to backport this commit was
to provide an alternative declaration for mutex_spin_on_owner()
as it was declared non-inline for older kernels.

Finally c5491ea7 required backporting schedule_preempt_disabled()
as well but that just consisted of carrying over the original
implementation. Since its not exported we need to reimplement
it to make it available to our internal core ww mutex port.

mcgrof@frijol ~/linux-stable (git::master)$ git describe --contains 040a0a371
v3.11-rc1~147^2~5

mcgrof@frijol ~/linux-stable (git::master)$ git describe --contains a41b56ef
v3.11-rc1~147^2~6

mcgrof@frijol ~/linux-stable (git::master)$ git describe --contains 2bd2c92c
v3.10-rc1~200^2~3

mcgrof@frijol ~/linux-stable (git::master)$ git describe --contains 41fcb9f2
v3.10-rc1~200^2~5

mcgrof@frijol ~/linux-stable (git::master)$ git describe --contains c5491ea7
v3.4-rc1~3^2~27

commit 040a0a37100563754bb1fee6ff6427420bcfa609
Author: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Date:   Mon Jun 24 10:30:04 2013 +0200

    mutex: Add support for wound/wait style locks

    Wound/wait mutexes are used when other multiple lock
    acquisitions of a similar type can be done in an arbitrary
    order. The deadlock handling used here is called wait/wound in
    the RDBMS literature: The older tasks waits until it can acquire
    the contended lock. The younger tasks needs to back off and drop
    all the locks it is currently holding, i.e. the younger task is
    wounded.

    For full documentation please read Documentation/ww-mutex-design.txt.

    References: https://lwn.net/Articles/548909/
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Rob Clark <robdclark@gmail.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: dri-devel@lists.freedesktop.org
Cc: linaro-mm-sig@lists.linaro.org
Cc: rostedt@goodmis.org
Cc: daniel@ffwll.ch
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/51C8038C.9000106@canonical.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
commit a41b56efa70e060f650aeb54740aaf52044a1ead
Author: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Date:   Thu Jun 20 13:31:05 2013 +0200

    arch: Make __mutex_fastpath_lock_retval return whether fastpath succeeded or not

    This will allow me to call functions that have multiple
    arguments if fastpath fails. This is required to support ticket
    mutexes, because they need to be able to pass an extra argument
    to the fail function.

    Originally I duplicated the functions, by adding
    __mutex_fastpath_lock_retval_arg. This ended up being just a
    duplication of the existing function, so a way to test if
    fastpath was called ended up being better.

    This also cleaned up the reservation mutex patch some by being
    able to call an atomic_set instead of atomic_xchg, and making it
    easier to detect if the wrong unlock function was previously
    used.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: dri-devel@lists.freedesktop.org
Cc: linaro-mm-sig@lists.linaro.org
Cc: robclark@gmail.com
Cc: rostedt@goodmis.org
Cc: daniel@ffwll.ch
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20130620113105.4001.83929.stgit@patser
Signed-off-by: Ingo Molnar <mingo@kernel.org>
commit 2bd2c92cf07cc4a373bf316c75b78ac465fefd35
Author: Waiman Long <Waiman.Long@hp.com>
Date:   Wed Apr 17 15:23:13 2013 -0400

    mutex: Queue mutex spinners with MCS lock to reduce cacheline contention

    <-- snip -->

commit 41fcb9f230bf773656d1768b73000ef720bf00c3
Author: Waiman Long <Waiman.Long@hp.com>
Date:   Wed Apr 17 15:23:11 2013 -0400

    mutex: Move mutex spinning code from sched/core.c back to mutex.c

    <-- snip -->

commit c5491ea779793f977d282754db478157cc409d82
Author: Thomas Gleixner <tglx@linutronix.de>
Date:   Mon Mar 21 12:09:35 2011 +0100

    sched/rt: Add schedule_preempt_disabled()

    <-- snip -->

Cc: maarten.lankhorst@canonical.com
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: dri-devel@lists.freedesktop.org
Cc: linaro-mm-sig@lists.linaro.org
Cc: rostedt@goodmis.org
Cc: daniel@ffwll.ch
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: enable kconfig language on dependencies file

Certain complex features that are backported may be be
limitted to a certain target build configuration. An example
can be if a backported feature is not yet backported with
support for lockdep. In order to avoid build failures with
these types of restrictions allow for specifying build
configuration dependencies on backported upstream kconfig
symbols other than just kernel versioning contstraints.

This adds support for specifying upstream kconfig constaints
other than kernel versioning by adding kconfig language
extensions on the dependencies file. This will update the
copied over upstream Kconfig file for the symbol specified
with the kconfig constraints specified.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: backport of_get_child_by_name() support

This was added via 9c19761a. While at it clean up the backported
header a bit to make backporting more OF stuff more manageable.

mcgrof@frijol ~/linux-stable (git::master)$ git describe --contains 9c19761a
v3.7-rc1~123^2~4

commit 9c19761a7ecdc86abb2fba0feb81e8952eccc1f1
Author: Srinivas Kandagatla <srinivas.kandagatla@st.com>
Date:   Tue Sep 18 08:10:28 2012 +0100

    dt: introduce of_get_child_by_name to get child node by name

    This patch introduces of_get_child_by_name function to get a child node
    by its name in a given parent node.

    Without this patch each driver code has to iterate the parent and do
    a string compare, However having of_get_child_by_name libary function would
    avoid code duplication, errors and is more convenient.

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@st.com>
Signed-off-by: Rob Herring <rob.herring@calxeda.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: copy over mfd/max8998.h mfd/max8998-private.h

This is required by CONFIG_REGULATOR_MAX8998
which builds drivers/regulator/max8998.c as of
next-20130703.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backport: disable unused automatic backports

When an automatic backport isn't included due to not
being used (see commit 6e0475b599217eceb8e01a1e572c,
"gentree: add automatic backports only if needed"),
selecting BACKPORT_USERSEL_BUILD_ALL will make the
build fail. Avoid this by disabling such symbols in
the Kconfig.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: fix DMI_EXACT_MATCH() backport

DMI_EXACT_MATCH uses struct dmi_strmatch's new member exact_match:

@@ -456,7 +456,8 @@ enum dmi_field {
 };

 struct dmi_strmatch {
-       unsigned char slot;
+       unsigned char slot:7;
+       unsigned char exact_match:1;
char substr[79];
 };

Prior to 5017b285 we only had slot so to use DMI_EXACT_MATCH with its
intent we'd have to do something like slot |= 1 if its called. This
however assumes though that older code has the sanity check as changed
in 5017b285 on drivers/firmware/dmi_scan.c. dmi_scan.o gets linked
with CONFIG_DMI. At least for x86 that gets objects sprinkled on
arch/x86/, but more importantly CONFIG_DMI is bool. I've argued how I
envision us being able to backport core components before (see
0935deab for the hint) but as it is right now we can't. We only
backport things we can throw in as modular.

Since we only backport modularly for now we can't backport DMI_EXACT_MATCH()
and as such all entries defined with DMI_EXACT_MATCH() should be ifdef'd
out for usage only on kernels >= v3.11 but to help reduce code churn we
can also just force such entries to be ignored for now. We therefore
backport DMI_EXACT_MATCH() for now to match something that will not be
found.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: fix wq_name_list initialization

As noted by Johannes this wasn't being initialized
without this.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: simplify space regexp for src_line

The regexp doesn't require the extra brackets.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: lib/kconfig.py - fix src line regexp

The source line can be prefixed by spaces and tabs. Without
this a slew of kconfig options were not being parsed for
dependency checking and at compile time they'd obviously
fail.

Tested with ckmake against next-20130627

1   2.6.24              [  OK  ]
2   2.6.25              [  OK  ]
3   2.6.26              [  OK  ]
4   2.6.27              [  OK  ]
5   2.6.28              [  OK  ]
6   2.6.29              [  OK  ]
7   2.6.30              [  OK  ]
8   2.6.31              [  OK  ]
9   2.6.32              [  OK  ]
10  2.6.33              [  OK  ]
11  2.6.34              [  OK  ]
12  2.6.35              [  OK  ]
13  2.6.36              [  OK  ]
14  2.6.37              [  OK  ]
15  2.6.38              [  OK  ]
16  2.6.39              [  OK  ]
17  3.0.79              [  OK  ]
18  3.1.10              [  OK  ]
19  3.10-rc1            [  OK  ]
20  3.2.45              [  OK  ]
21  3.3.8               [  OK  ]
22  3.4.46              [  OK  ]
23  3.5.7               [  OK  ]
24  3.6.11              [  OK  ]
25  3.7.10              [  OK  ]
26  3.8.13              [  OK  ]
27  3.9.3               [  OK  ]

real    34m5.379s
user    921m5.112s
sys     128m8.156s

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: enable REGULATOR_MAX77693 on >= 3.6

There is a symbol missing otherwise that is currently
only provided by the core, ie, non modular.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add clock enable/disable on soc_camera_power_[on|off]()

v4l2-core now supports enabling and disabling its own clock
on turning the camera on / off. We backport v4l2-core fully
so just backport the soc calls appropriately to let SOC cameras
to turn the clock on / off when the v4l2-core clock ops are
implemented on a device driver.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: revert DMI_EXACT_MATCH() for older kernels

struct dmi_strmatch was modified on v3.11 with a
exact_match member. The DMI_EXACT_MATCH() macro was
added as a mod version of DMI_MATCH() to account for
dmi strings that look closely like each other. An
example is D510MO Vs D510MOV.

Since exact_match is is not available on older kernels
we can't use it and must revert to the old DMI_MATCH().
Addressing the desired behaviour will require a separate
change to deal with addressing the difference
in the behaviour.

Chris Wilson recommends that if we can't use this we should
also rever the DMI related fixes. I believe these are:

dcf6d294830d46b0e6901477fb4bf455281d90c8 - drm/i915: quirk away phantom LVDS on Intel's D525MW mainboard
e5614f0c2d0f4d7f0b8ef745d34593baf2c5dbf8 - drm/i915: quirk away phantom LVDS on Intel's D510MO mainboard

Once we get confirmation we can do that.

Proof of why we can't backport DMI_EXACT_MATCH():

DMI_EXACT_MATCH uses struct dmi_strmatch's new
member exact_match:

@@ -456,7 +456,8 @@ enum dmi_field {
 };

 struct dmi_strmatch {
-       unsigned char slot;
+       unsigned char slot:7;
+       unsigned char exact_match:1;
char substr[79];
 };

Prior to 5017b285 we only had slot member so to use
DMI_EXACT_MATCH() with its intent we'd have to do something
like:
slot |= 1

if its called. This however assumes though that older code has
the sanity check as changed in 5017b285 on drivers/firmware/dmi_scan.c.
dmi_scan.o gets linked with CONFIG_DMI. At least for x86 that gets
objects sprinkled on arch/x86/, but more importantly CONFIG_DMI is
bool. I've argued how I envision us being able to backport core
components before (see 0935deab for the hint) but as it is right
now we can't. We only backport things we can throw in as modular.

mcgrof@frijol ~/linux-stable (git::master)$ git describe --contains 5017b285
v3.11-rc1~99^2~254

commit 5017b2851373ee15c7035151853bb1448800cae2
Author: Jani Nikula <jani.nikula@intel.com>
Date:   Wed Jul 3 15:05:02 2013 -0700

    dmi: add support for exact DMI matches in addition to substring matching

    dmi_match() considers a substring match to be a successful match.  This is
    not always sufficient to distinguish between DMI data for different
    systems.  Add support for exact string matching using strcmp() in addition
    to the substring matching using strstr().

    The specific use case in the i915 driver is to allow us to use an exact
    match for D510MO, without also incorrectly matching D510MOV:

      {
        .ident = "Intel D510MO",
        .matches = {
                DMI_MATCH(DMI_BOARD_VENDOR, "Intel"),
                DMI_EXACT_MATCH(DMI_BOARD_NAME, "D510MO"),
        },
      }

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Cc: <annndddrr@gmail.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Cornel Panceac <cpanceac@gmail.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: jani.nikula@intel.com
Cc: annndddrr@gmail.com
Cc: chris@chris-wilson.co.uk
Cc: cpanceac@gmail.com
Cc: daniel.vetter@ffwll.ch
Cc: greg@kroah.com
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: copy the max77693-private.h header file

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: simplify swiotlb_nr_tbl() backport

swiotlb_nr_tbl() was available since 3.2 but was exported since 3.3.
Since it uses an internal global state variable, it is impossible
to backport it to. Instead of reverting the changes added just
let this return 0 as the code will not do anything. This simplifies
the backport.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: refresh patches for next-20130627

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: fix patches required for next-20130627

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backport: backport sprintf-style workqueue naming

Since kernel version 3.3, workqueue names could be
sprintf'ed instead of just pointing to a name. This
wasn't used a lot so never needed to be backported,
but now it's used everywhere. Backport this API.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
[mcgrof: add WQ_HIGHPRI and WQ_MEM_RECLAIM defines]
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: backport drvdata = NULL core driver fixes

The Linux kernel had tons of code which at times cleared the
drvdata upon probe failure or release. There are however a bunch
of drivers that didn't clear this.

Commit 0998d063 implmented clearing this upon device_release_driver()
and dealt with probe failure on driver_probe_device(). After this the
kernel was cleaned up separately with *tons* of patches to remove all
these driver specific settings given that the clearing is now done
internally by the device core.

Instead of ifdef'ing code back in for older code where it was properly
in place backport this by piggy backing the new required code upon the
calls used in place. There is a small race here upon device_release_driver()
but we can live with that theoretical race.

Due to the way we hack this backport we can't use a separate namespace
as we have with other symbols.

mcgrof@frijol ~/linux-stable (git::master)$ git describe --contains \
0998d0631001288a5974afc0b2a5f568bcdecb4d
v3.6-rc1~99^2~14^2~17

I count 65 patches implemented after this:

mcgrof@frijol ~/linux-stable (git::master)$ git format-patch \
--grep="device-core: Ensure drvdata = NULL when no driver is bound" \
 -o null-drv-fix v3.6-rc1~99^2~14^2~17..

  TL;DR

Alan Stern argued that perhaps applying this to backports wasn't a good
idea given that evidence shows that the original patch actually exposed
tons of bugs in driver code where they were doing the wrong thing.
While this may be true if the original patch was a bad idea it should
be reverted, and if a bug is found upstream, then by all means
finding it through backports will only accelerate the pace at which
we fix these exposed bugs. That is, if a bug is found due to this on
backports then a respective fix for it should go upstream, not to
backports. This is the benefit of providing backports releases: keep
your users engaged on upstream fixes.

Furthermore I am in hopes that perhaps we can SmPL'ify the bugs
instead and in the future perhaps require SmPL to proove that
the what the original patch was doing won't affect the inverse
of what the patch was trying to do -- that is drivers doing the
wrong thing.

commit 0998d0631001288a5974afc0b2a5f568bcdecb4d
Author: Hans de Goede <hdegoede@redhat.com>
Date:   Wed May 23 00:09:34 2012 +0200

    device-core: Ensure drvdata = NULL when no driver is bound

    1) drvdata is for a driver to store a pointer to driver specific data
    2) If no driver is bound, there is no driver specific data associated with
       the device
    3) Thus logically drvdata should be NULL if no driver is bound.

    But many drivers don't clear drvdata on device_release, or set drvdata
    early on in probe and leave it set on probe error. Both of which results
    in a dangling pointer in drvdata.

    This patch enforce for drvdata to be NULL after device_release or on probe
    failure.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Tested with ckmake against next-20130618:

1   2.6.24              [  OK  ]
2   2.6.25              [  OK  ]
3   2.6.26              [  OK  ]
4   2.6.27              [  OK  ]
5   2.6.28              [  OK  ]
6   2.6.29              [  OK  ]
7   2.6.30              [  OK  ]
8   2.6.31              [  OK  ]
9   2.6.32              [  OK  ]
10  2.6.33              [  OK  ]
11  2.6.34              [  OK  ]
12  2.6.35              [  OK  ]
13  2.6.36              [  OK  ]
14  2.6.37              [  OK  ]
15  2.6.38              [  OK  ]
16  2.6.39              [  OK  ]
17  3.0.79              [  OK  ]
18  3.1.10              [  OK  ]
19  3.10-rc1            [  OK  ]
20  3.2.45              [  OK  ]
21  3.3.8               [  OK  ]
22  3.4.46              [  OK  ]
23  3.5.7               [  OK  ]
24  3.6.11              [  OK  ]
25  3.7.10              [  OK  ]
26  3.8.13              [  OK  ]
27  3.9.3               [  OK  ]

real    32m2.332s
user    860m23.688s
sys     121m20.840s

Cc: Hans de Goede <hdegoede@redhat.com>
Cc: Julia Lawall <julia.lawall@lip6.fr>
Cc: linux-usb@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Cc: Jiri Slaby <jslaby@suse.cz>
Cc: Jiri Kosina <jkosina@suse.cz>
Cc: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add gplizer for EXPORT_SYMBOL_GPL() help

If we forget to use EXPORT_SYMBOL_GPL() we can run this.
We may later use this for other things.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: Fix and document EXPORT_SYMBOL_GPL() preference

In terms of project naming the backports project has evolved
as follows:

compat-wireless --> compat-drivers --> backports

All along the design and intent behind the initial project
has been to provide a framework for delivery of *upstream*
drivers using a backported infrastructure. I've made it
clear that in no way shape or form did I ever want any
proprietary driver to make use of the framework. As the
project has grown I'm not alone with this sentiment so
lets document that and also fix a few symbols that have
slipped along the way.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: update README to include new subsystems

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add make help and defconfig-help targets

Throw users a bone by helping them compile specific target
drivers. Hauke did some good work in providing tons of
defconfigs, provide a command to let users query these
easily with 'make defconfig-help'. Also provide a more
generic 'make help' target.

Screenshot of make defconfig-help:

mcgrof@frijol ~/backports-20130618 $ make defconfig-help
Driver or subsystem configuration targets:
  defconfig-alx
  defconfig-ar5523
  defconfig-ath5k
  defconfig-ath6kl
  defconfig-ath9k
  defconfig-ath10k
  defconfig-b43
  defconfig-b43legacy
  defconfig-brcmfmac
  defconfig-brcmsmac
  defconfig-carl9170
  defconfig-drm
  defconfig-ieee802154
  defconfig-iwlwifi
  defconfig-media
  defconfig-nfc
  defconfig-regulator
  defconfig-rtlwifi
  defconfig-wifi
  defconfig-wil6210

Running 'make help' will provide the above and in addition to that
something similar to what the Linux kernel provides with the
list of supported targets backports provides.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add ath10k defconfig

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add SPEED_UNKNOWN and DUPLEX_UNKNOWN

This gets us in synch with next-20130618.

This was added via uapi/linux/ethtool.h but using
that file causes some odd issues I simply cannot
resolve right now. I stuff these definitions into
the regular include/linux/ethtool.h for now.

This backports commit 589665f5 added via v3.2.

mcgrof@frijol ~/linux-stable (git::master)$ git describe --contains \
589665f5a6008dbce1d0af2cb93e94a80bf78151
v3.2-rc1~4^2~10

commit 589665f5a6008dbce1d0af2cb93e94a80bf78151
Author: Dan Carpenter <dan.carpenter@oracle.com>
Date:   Fri Nov 4 08:21:38 2011 +0000

    bonding: comparing a u8 with -1 is always false

    slave->duplex is a u8 type so the in bond_info_show_slave() when we
    check "if (slave->duplex == -1)", it's always false.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
1   2.6.24              [  OK  ]
2   2.6.25              [  OK  ]
3   2.6.26              [  OK  ]
4   2.6.27              [  OK  ]
5   2.6.28              [  OK  ]
6   2.6.29              [  OK  ]
7   2.6.30              [  OK  ]
8   2.6.31              [  OK  ]
9   2.6.32              [  OK  ]
10  2.6.33              [  OK  ]
11  2.6.34              [  OK  ]
12  2.6.35              [  OK  ]
13  2.6.36              [  OK  ]
14  2.6.37              [  OK  ]
15  2.6.38              [  OK  ]
16  2.6.39              [  OK  ]
17  3.0.79              [  OK  ]
18  3.1.10              [  OK  ]
19  3.10-rc1            [  OK  ]
20  3.2.45              [  OK  ]
21  3.3.8               [  OK  ]
22  3.4.46              [  OK  ]
23  3.5.7               [  OK  ]
24  3.6.11              [  OK  ]
25  3.7.10              [  OK  ]
26  3.8.13              [  OK  ]
27  3.9.3               [  OK  ]

real    15m7.262s
user    335m33.818s
sys     82m36.486s

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add new alx upstream backport

This backports the new upstream alx driver and
properly splits up each required backports into
its respective collateral evolution bucket.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: remove alx extra patch

The alx driver was upstreamed by Johannes, so this is
no longer required. Note that this driver is now GPLv2
or later, BSD folks will have to look at the old unified
tree for a permissive licensed port.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: re-introduce make uninstall

This approach only uninstalls modules you have built
previously on your current build directory for backports.
If this requires adjustments then we'll have to just have
to use our own new backports KMODDIR.

Signed-off-by: Felix Bitterli <ic.felix@gmail.com>
[mcgrof: extended the commit log a bit]
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backport: include printk.h into backport-3.11.c

This is needed for pr_warn(), otherwise it's not defined on
all kernels (it is sometimes pulled in by other headers.)

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: update output template .gitignore for kconfig move

I clearly forgot this during the Kconfig move, the .gitignore
that is in the output template directory (and used with the
git-tracker) needs to be updated for the kconf directory move.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: refresh patches on next-20130617

== ckmake-report.log ==

1   2.6.24              [  OK  ]
2   2.6.25              [  OK  ]
3   2.6.26              [  OK  ]
4   2.6.27              [  OK  ]
5   2.6.28              [  OK  ]
6   2.6.29              [  OK  ]
7   2.6.30              [  OK  ]
8   2.6.31              [  OK  ]
9   2.6.32              [  OK  ]
10  2.6.33              [  OK  ]
11  2.6.34              [  OK  ]
12  2.6.35              [  OK  ]
13  2.6.36              [  OK  ]
14  2.6.37              [  OK  ]
15  2.6.38              [  OK  ]
16  2.6.39              [  OK  ]
17  3.0.79              [  OK  ]
18  3.1.10              [  OK  ]
19  3.10-rc1            [  OK  ]
20  3.2.45              [  OK  ]
21  3.3.8               [  OK  ]
22  3.4.46              [  OK  ]
23  3.5.7               [  OK  ]
24  3.6.11              [  OK  ]
25  3.7.10              [  OK  ]
26  3.8.13              [  OK  ]
27  3.9.3               [  OK  ]

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

backprots: ath10k needs at least kernel version 2.6.30

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

backprots: add rounddown

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

backports: add PCI_DEVICE_SUB

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

backports: add USB_DEVICE_INTERFACE_NUMBER

The original version also checked for bInterfaceNumber, which is only
available in more recent kernel versions.

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

backports: remove useless include in USB_VENDOR_AND_INTERFACE_INFO

This also changes the #if <kernel version> to a ifndef, now it is also
compatible with kernel already containing this backport.

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

backports: deactivate FireDTV DVB Driver for kernel < 3.11

This would need the following commit backported:
commit 94a87157cde95d38b9cdf1116e4f0fd93f6d25df
Author: Stefan Richter <stefanr@s5r6.in-berlin.de>
Date:   Sun Jun 9 18:15:00 2013 +0200

    firewire: introduce fw_driver.probe and .remove methods

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

backports: add ieee802154

This adds the ieee802154 subsystem.

== ckmake-report.log ==

1   2.6.24              [  OK  ]
2   2.6.25              [  OK  ]
3   2.6.26              [  OK  ]
4   2.6.27              [  OK  ]
5   2.6.28              [  OK  ]
6   2.6.29              [  OK  ]
7   2.6.30              [  OK  ]
8   2.6.31              [  OK  ]
9   2.6.32              [  OK  ]
10  2.6.33              [  OK  ]
11  2.6.34              [  OK  ]
12  2.6.35              [  OK  ]
13  2.6.36              [  OK  ]
14  2.6.37              [  OK  ]
15  2.6.38              [  OK  ]
16  2.6.39              [  OK  ]
17  3.0.79              [  OK  ]
18  3.1.10              [  OK  ]
19  3.10-rc1            [  OK  ]
20  3.2.45              [  OK  ]
21  3.3.8               [  OK  ]
22  3.4.46              [  OK  ]
23  3.5.7               [  OK  ]
24  3.6.11              [  OK  ]
25  3.7.10              [  OK  ]
26  3.8.13              [  OK  ]
27  3.9.3               [  OK  ]

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

backports: add cast in class_find_device()

Also type of the 3 parameter of class_find_device() changed from void *
to const void *, add a cast to prevent a warning.

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

backports: add module_spi_driver

module_spi_driver was added in the following commit to mainline:
commit 3acbb0142d48713a8f65cde678a54f419801c189
Author: Lars-Peter Clausen <lars@metafoo.de>
Date:   Wed Nov 16 10:13:37 2011 +0100

    SPI: Add helper macro for spi_driver boilerplate

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

backports: use the given parameter in NETLINK_CB_PORTID()

NETLINK_CB_PORTID() ignored the given parameter, this was wrong.

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

backports: add NET_CORE config option

NET_CORE was added with kernel 3.2 and before this patch no Ethernet
driver was build on older kernel versions.

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

backports: build ATL1E only on kernel > 2.6.26

The ATL1E driver needs netif_set_gso_max_size() and this is not
available in older kernel versions.

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

backports: alx driver needs at least kernel 2.6.31

The alx driver uses drivers/net/mdio.c which was added with kernel
2.6.31. Build the alx driver only on this kernel version or a more
recent one.

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

backports: move netif_set_real_num_rx_queues() to compat-2.6.37.c

netif_set_real_num_rx_queues() was added to kernel version 2.6.37 and
not to kernel 2.6.35 therefore it should also be added to the
corresponding compat-*.c file.

This also adds a missing EXPORT_SYMBOL, documentation and a check.

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

backports: rename compat file descriptions

Make these descriptions generic.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backprots: add default config with all wifi drivers

This default config contains all wifi drivers.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: fix tracing on wil6210

This includes the trace.h file like it is done for the other drivers.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add default config for alx

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: alx driver now uses NETIF_F_HW_VLAN_CTAG_*

This was/will be renamed in the alx driver, this patch makes backports
work with alx again.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: add defconfigs for Atheros wifi drivers

This adds default configs for Atheros wifi drivers.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: adapt shrinkers api patch to next-20130607

The shrinkers api changed a little in the last version, this patch
makes the patches apply again.

== ckmake-report.log ==

1   2.6.24              [  OK  ]
2   2.6.25              [  OK  ]
3   2.6.26              [  OK  ]
4   2.6.27              [  OK  ]
5   2.6.28              [  OK  ]
6   2.6.29              [  OK  ]
7   2.6.30              [  OK  ]
8   2.6.31              [  OK  ]
9   2.6.32              [  OK  ]
10  2.6.33              [  OK  ]
11  2.6.34              [  OK  ]
12  2.6.35              [  OK  ]
13  2.6.36              [  OK  ]
14  2.6.37              [  OK  ]
15  2.6.38              [  OK  ]
16  2.6.39              [  OK  ]
17  3.0.79              [  OK  ]
18  3.1.10              [  OK  ]
19  3.10-rc1            [  OK  ]
20  3.2.45              [  OK  ]
21  3.3.8               [  OK  ]
22  3.4.46              [  OK  ]
23  3.5.7               [  OK  ]
24  3.6.11              [  OK  ]
25  3.7.10              [  OK  ]
26  3.8.13              [  OK  ]
27  3.9.3               [  OK  ]

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: refersh patches on next-20130606

There was a manual adjustment needed in
patches/collateral-evolutions/network/11-dev-pm-ops/drivers_net_wireless_iwlegacy_common.patch

1   2.6.24              [  OK  ]
2   2.6.25              [  OK  ]
3   2.6.26              [  OK  ]
4   2.6.27              [  OK  ]
5   2.6.28              [  OK  ]
6   2.6.29              [  OK  ]
7   2.6.30              [  OK  ]
8   2.6.31              [  OK  ]
9   2.6.32              [  OK  ]
10  2.6.33              [  OK  ]
11  2.6.34              [  OK  ]
12  2.6.35              [  OK  ]
13  2.6.36              [  OK  ]
14  2.6.37              [  OK  ]
15  2.6.38              [  OK  ]
16  2.6.39              [  OK  ]
17  3.0.79              [  OK  ]
18  3.1.10              [  OK  ]
19  3.10-rc1            [  OK  ]
20  3.2.45              [  OK  ]
21  3.3.8               [  OK  ]
22  3.4.46              [  OK  ]
23  3.5.7               [  OK  ]
24  3.6.11              [  OK  ]
25  3.7.10              [  OK  ]
26  3.8.13              [  OK  ]
27  3.9.3               [  OK  ]

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: use old shrinkers API on old kernels

The shrinkers API was changed in new kernel versions. This patch makes
the drm drivers use the old version on older kernel versions.

This backports the following commit from mainline kernel:
commit 1c1df1f6646854cca15fede54ec475b0e9f6a162
Author: Dave Chinner <dchinner@redhat.com>
Date:   Thu Jun 6 10:40:41 2013 +1000

    drivers: convert shrinkers to new count/scan API

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: backport some memory functions

This includes the following functions needed by some drm drivers:
arch_phys_wc_add()
arch_phys_wc_del()
phys_wc_to_mtrr_index()

This backports the following commit form mainline kernel:
commit d0d98eedee2178c803dd824bb09f52b0e2ac1811
Author: Andy Lutomirski <luto@amacapital.net>
Date:   Mon May 13 23:58:40 2013 +0000

    Add arch_phys_wc_{add, del} to manipulate WC MTRRs if needed

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

gentree/git-tracker: make python 3 compatible

Make the python scripting compatible with python 3
(as tested with python 3.3). The patch library was
a bit tricky.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: use check-lxdialog.sh

We ship it, so we should use it. This should fix issues
like one distro needing "-lncurses -ltinfo" and others
requiring just "-lncurses" and not having "tinfo".

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: do not backport struct watchdog_device in RHEL 6.4

This is already included in the RHEL 6.4 kernel.

1   2.6.24              [  OK  ]
2   2.6.25              [  OK  ]
3   2.6.26              [  OK  ]
4   2.6.27              [  OK  ]
5   2.6.28              [  OK  ]
6   2.6.29              [  OK  ]
7   2.6.30              [  OK  ]
8   2.6.31              [  OK  ]
9   2.6.32              [  OK  ]
10  2.6.33              [  OK  ]
11  2.6.34              [  OK  ]
12  2.6.35              [  OK  ]
13  2.6.36              [  OK  ]
14  2.6.37              [  OK  ]
15  2.6.38              [  OK  ]
16  2.6.39              [  OK  ]
17  3.0.79              [  OK  ]
18  3.1.10              [  OK  ]
19  3.10-rc1            [  OK  ]
20  3.2.45              [  OK  ]
21  3.3.8               [  OK  ]
22  3.4.46              [  OK  ]
23  3.5.7               [  OK  ]
24  3.6.11              [  OK  ]
25  3.7.10              [  OK  ]
26  3.8.13              [  OK  ]
27  3.9.3               [  OK  ]

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: protect pcmcia_enable_device

This is already defined in RHEL 6.4

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: protect irq_set_irq_type()

irq_set_irq_type is defined to set_irq_type in RHEL 6.4.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: to not backport skb_add_rx_fragi() on RHEL 6.4

The kernel in RHEL 6.4 already ships with skb_add_rx_frag() with 6
arguments.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: do not backport skb_tx_timestamp() on RHEL 6.4

The skb_tx_timestamp() implementation in RHEL 6.4 actually does
something.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: do not backport shmem_read_mapping_page_gfp() on RHEL 6.4

shmem_read_mapping_page_gfp() looks different in the RHEL 6.4 source
code and we should use that version instead.
This also adds a missing include to prefix this function with backport_.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: do not backport netif_set_real_num_rx_queues() on RHEL 6.4

The in kernel version of netif_set_real_num_rx_queues() on RHEL 6.4
accesses some other struct members than our backported version. We
should use that version instead.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>