backports: allow for different backport prefix and simple integration support

The way we backport when packaging is to minimize the amount of
changes required by taking advantage of the fact that Kconfig
can treat CONFIG_ as an environment variable with getenv()
when parsing menu entries. When doing integration we don't want
to do this so instead we'll rely on the CONFIG_BACKPORT prefix.
This requires a bit of work on our current parsers, so to do that
identify the backport and provide that to helpers.

This also provides initial basic support for integration where no
additional code was required other than prefixes or ensuring certain
things are not done for integration. Before we expose integration
support we'll need to add quite a bit more things so this just assumes
integration is never desired.

Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>

backports: keep track of original backported source symbols early on

We'll later want to do some modifications of symbols we backport
when integrating, but before we do any modification on kconfig
files we should keep track of the original source symbols.

This commit introduces no functional changes right now, it will
become more useful after we start adjusting kconfig entries
for integration.

Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>

backports: grant parsers access to kconfig config and menuconfig

Allow parsers to query the type of option used, either config
or menuconfig. This will be used later.

Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>

backports: move packaging kconfig hacks to its own file

Move packaging kconfig hacks to its own file, we'll be ignoring
this file later for postprocessing. Keeping this in a separate
file will make parsing much simpler.

Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>

backports: ifdef around module_init() module_exit() for modules

We only need this for modules.

Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>

backports: move legacy and SmPL patch application into helper

This allows us to extend how backports uses patches for
different types of applications. This will later be used
for kernel integration support, for example.

This should have no functional change.

Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>

backports: refresh patches on next-20141023

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.62              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.104             [  OK  ]
6   3.5.7               [  OK  ]
7   3.6.11              [  OK  ]
8   3.7.10              [  OK  ]
9   3.8.13              [  OK  ]
10  3.9.11              [  OK  ]
11  3.10.58             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.31             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.22             [  OK  ]
16  3.15.10             [  OK  ]
17  3.16.6              [  OK  ]
18  3.17.1              [  OK  ]
19  3.18-rc1            [  OK  ]

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

devel: update the test kernels

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

backports: add device coredump framework

As drivers will start requiring the new framework and it has
been merged to linux/master during the 3.18 merge window, we
can include it into backports to compile if needed.

On kernels that do already have it but didn't want it built
honor that, but if building otherwise don't. This may need
to be revisited in the future.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

backports: add kobj_to_dev()

This will be needed by the devcoredump backport.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

defconfig: add some new wifi drivers

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

defconfig: update media defconfig

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

backports: refresh patches on next-20141003

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.62              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.103             [  OK  ]
6   3.5.7               [  OK  ]
7   3.6.11              [  OK  ]
8   3.7.10              [  OK  ]
9   3.8.13              [  OK  ]
10  3.9.11              [  OK  ]
11  3.10.54             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.27             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.18             [  OK  ]
16  3.15.10             [  OK  ]
17  3.16.2              [  OK  ]
18  3.17-rc3            [  OK  ]

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

dependencies: deactivate IR_HIX5HD2 on < 3.7

IR_HIX5HD2 needs syscon, this has to be added to the Kconfig file and
on kernel < 3.7 it is completely missing.

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

dependencies: deactivate SOC_CAMERA on < 3.5

Deactivate SOC_CAMERA because of missing backport for vb2_expbuf() in
kernel < 3.5.

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

dependencies: deactivate pt3 on kernel < 3.3

pt3 needs kthread_freezable_should_stop() which is not available on
these old kernels.

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

dependencies: deactivate some media drivers on < 3.4

Deactivate all drivers using regmap_bulk_write() on kernel < 3.4.

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

dependencies: deactivate tracing in ath10k and ath with < 3.4

On kernel < 3.4 compiling ath10k with tracing results in huge compile
errors. Just deactivate it till someone fixes the problems. The same
compile error occurs in ath module with kernel 3.0.

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

backports: add freezable_schedule_hrtimeout_range()

This is used by the pt3 driver.

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

backports: add some missing header files

The hackrf driver needs PAGE_ALIGN() from mm.h.
The ir-hix5hd2 driver needs of_get_property() from of.h.

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

backports: add wil6210_uapi.h

This file is needed by the wil6210 driver now.

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

backports: refresh on next-20140919

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.62              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.103             [  OK  ]
6   3.5.7               [  OK  ]
7   3.6.11              [  OK  ]
8   3.7.10              [  OK  ]
9   3.8.13              [  OK  ]
10  3.9.11              [  OK  ]
11  3.10.54             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.27             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.18             [  OK  ]
16  3.15.10             [  OK  ]
17  3.16.2              [  OK  ]
18  3.17-rc3            [  OK  ]

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

backports: these drivers need a new function

These drivers need regmap_bulk_write(), which was added in 3.4

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

backports: add skb_clone_sk()

skb_clone_sk() is now used by mac80211 and this change also needs some
changes to skb_complete_wifi_ack(), so I replaced it with our own
version.

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

backports: add eth_get_headlen()

Instead of using the code from a recent kernel, I used the old code
from the igb driver to calculate the header length. The new code in the
kernel makes use of some __skb_flow_dissect() functions and headers not
available in 3.0.

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

backports: add S8_MIN

This is needed by the iwlwifi river.

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

defconfigs: enable PCIe support in brcmfmac defconfig

In 3.17 kernel the brcmfmac supports PCIe host interface. So enable
support when selecting defconfig-brcmfmac.

Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

backports: fix mconf compilation library assumptions

On OpenSUSE factory 'make menuconfig' fails to compile with:

cc -Wall -Wmissing-prototypes -Wstrict-prototypes -O2
-fomit-frame-pointer -DCURSES_LOC="<ncurses.h>" -DLOCALE   -c -o
lxdialog/yesno.o lxdialog/yesno.c
cc -Wl,--no-as-needed -lncursesw  mconf.o zconf.tab.o
lxdialog/checklist.o lxdialog/inputbox.o lxdialog/menubox.o
lxdialog/textbox.o lxdialog/util.o lxdialog/yesno.o   -o mconf
/usr/lib64/gcc/x86_64-suse-linux/4.8/../../../../x86_64-suse-linux/bin/ld:
/lib64/libncursesw.so.5: undefined reference to symbol 'acs_map'
/usr/lib64/gcc/x86_64-suse-linux/4.8/../../../../lib64/libtinfo.so:
error adding symbols: DSO missing from command line
collect2: error: ld returned 1 exit status

Although this can be fixed by linking with -ltinfo after a bit
further investigation I've traced this issue down to how we speak
to the linker and ask it to link and when. For instance if one
were to compile and link mconf by specifying the linker requirements
before the final object file things fail and if we ask for it after
things work. For instance if we ask for the libraries before:

gcc  -lncursesw -o
kconf/mconf
kconf/mconf.o
kconf/zconf.tab.o
kconf/lxdialog/checklist.o
kconf/lxdialog/util.o
kconf/lxdialog/inputbox.o
kconf/lxdialog/textbox.o
kconf/lxdialog/yesno.o
kconf/lxdialog/menubox.o  -lncursesw

Passing the linker requirements at the end makes the gcc happy:

gcc  -o kconf/mconf
kconf/mconf.o
kconf/zconf.tab.o
kconf/lxdialog/checklist.o
kconf/lxdialog/util.o
kconf/lxdialog/inputbox.o
kconf/lxdialog/textbox.o
kconf/lxdialog/yesno.o
kconf/lxdialog/menubox.o  -lncursesw

It seems that using -Wl,--no-as-needed does not do what we wish with regards
to the above requirements, this could be a bug but the kernel treats things
a bit differently so we need to adopt, fix this by being explicit by
using -Wl,--add-needed and letting it figure things out automagically.

This has been tested on Debian and OpenSUSE factory.

Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>

backports: refresh patches on next-20140905

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.62              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.103             [  OK  ]
6   3.5.7               [  OK  ]
7   3.6.11              [  OK  ]
8   3.7.10              [  OK  ]
9   3.8.13              [  OK  ]
10  3.9.11              [  OK  ]
11  3.10.54             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.27             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.18             [  OK  ]
16  3.15.10             [  OK  ]
17  3.16.2              [  OK  ]
18  3.17-rc3            [  OK  ]

manual changes were done in:
patches/collateral-evolutions/network/0006-ieee802154-6lowpan-namespace.patch

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

backports: remove usage of skbuff->xmit_more

This is a revet of these two upstream commits:
commit c1ebf46c1f72fe542853fc00f059a7d15259379d
Author: David S. Miller <davem@davemloft.net>
Date:   Fri Aug 22 17:24:49 2014 -0700

    igb: Support netdev_ops->ndo_xmit_flush()

commit 0b725a2ca61bedc33a2a63d0451d528b268cf975
Author: David S. Miller <davem@davemloft.net>
Date:   Mon Aug 25 15:51:53 2014 -0700

    net: Remove ndo_xmit_flush netdev operation, use signalling instead.

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

backports: bluetooth revert patch adding write_iter

This reverts this upstream patch for older kernel versions:

commit 9f87eaae842575ec7de51654127a9bf4ad07ad6d
Author: Al Viro <viro@zeniv.linux.org.uk>
Date:   Sat Aug 23 11:28:14 2014 -0400

    switch hci_vhci to ->write_iter()

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

backports: adapt api change of rchan_callbacks->create_buf_file in ath10k

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

backports: always use absolute path to output directory, less probability of pycocci failure.

While running gentree.py I got error messages such as 'Failed to
process SmPL patch
collateral-evolutions/generic/0001-group-attr/0001-group_attr_class.cocci',
due to incorrect path handling. This patch always makes gentree always
call pycocci with the absolute path to the output directory.

Signed-off-by: Jacob Siverskog <jacob@teenageengineering.com>

backports: refresh patches on next-20140822

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.62              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.103             [  OK  ]
6   3.5.7               [  OK  ]
7   3.6.11              [  OK  ]
8   3.7.10              [  OK  ]
9   3.8.13              [  OK  ]
10  3.9.11              [  OK  ]
11  3.10.54             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.27             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.18             [  OK  ]
16  3.15.10             [  OK  ]
17  3.16.2              [  OK  ]
18  3.17-rc3            [  OK  ]

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

backports: remove regulator drivers

It looks like no one is using the regulator drivers and maintaining
them costs some time because they often break. When someone needs
a specific driver please send a patch adding this one specific driver,
maintaining a small subset is not a problem, but maintaining all is a
big task.

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

backports: update test kernel versions

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

backports: fix build with Ubuntu mainline kernels

We only want the UTS_UBUNTU_RELEASE_ABI var when we are on a normal
Ubuntu distribution kernel and not when we are on a Ubuntu mainline
kernel. Some of the Ubuntu mainline kernel do have an invalid octal
number in this field like 031418 and we do not want to evaluate this at
all on the Ubuntu mainline kernels.  All Ubuntu distribution kernel
have CONFIG_VERSION_SIGNATURE set so this way we can detect the which
type of kernel we are on.

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

backports: remove ath5k section mismatch patch

This change was merge upstream in this commit:
commit 9baa3c34ac4e27f7e062f266f50cc5dbea26a6c1
Author: Benoit Taine <benoit.taine@lip6.fr>
Date:   Fri Aug 8 15:56:03 2014 +0200

    PCI: Remove DEFINE_PCI_DEVICE_TABLE macro use

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

backports: update test kernel versions

This updates the kernel versions to test against to the most recent
versions.
Do not use 3.15.9, but 3.15.7 because there are some problems with the
define UTS_UBUNTU_RELEASE_ABI in the more recent kernel version, see:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1347879

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

backports: refresh patches on next-20140808

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.62              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.102             [  OK  ]
6   3.5.7               [  OK  ]
7   3.6.11              [  OK  ]
8   3.7.10              [  OK  ]
9   3.8.13              [  OK  ]
10  3.9.11              [  OK  ]
11  3.10.52             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.26             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.16             [  OK  ]
16  3.15.7              [  OK  ]
17  3.16.0              [  OK  ]

Manual changes were done to:
patches/collateral-evolutions/network/0006-ieee802154-6lowpan-namespace.patch

This patch was removed:
patches/collateral-evolutions/network/0009-inet_frag_evictor.patch

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

backports: copy include/linux/platform_data/st21nfcb.h

This is needed by drivers/nfc/st21nfcb/i2c.c

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

backports: add list_next_entry()

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

backports: add devm_kmalloc_array()

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

backports: add ktime_get_raw()

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

backports: irq_get_trigger_type requires CONFIG_HAVE_GENERIC_HARDIRQS=y

Only backport irq_get_trigger_type on architectures that have
CONFIG_HAVE_GENERIC_HARDIRQS=y. Otherwise we end up with:

In file included from drivers/pci/msi.c:11:0:
backport/backport-include/linux/irq.h: In function 'irq_get_trigger_type':
backport/backport-include/linux/irq.h:9:9: error: implicit declaration
of function 'irq_get_irq_data' [-Werror=implicit-function-declaration]
  struct irq_data *d = irq_get_irq_data(irq);

Signed-off-by: Stefan Assmann <sassmann@kpanic.de>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

backports: refresh on next-20140725

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.60              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.93              [  OK  ]
6   3.5.7               [  OK  ]
7   3.6.11              [  OK  ]
8   3.7.10              [  OK  ]
9   3.8.13              [  OK  ]
10  3.9.11              [  OK  ]
11  3.10.43             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.22             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.7              [  OK  ]
16  3.15.0              [  OK  ]
17  3.16-rc1            [  OK  ]

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

backports: add missing linux/mm.h to airspy

Without this patch I get the following error message when compiling
against kernel 3.2, all other kernel versions work:

3.2.60/drivers/media/usb/airspy/airspy.c: In function ‘airspy_queue_setup’:
3.2.60/drivers/media/usb/airspy/airspy.c:506:2: error: implicit declaration of function ‘PAGE_ALIGN’ [-Werror=implicit-function-declaration]
sizes[0] = PAGE_ALIGN(s->buffersize);

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

backports: do not use dma_buf_vmap() on kernel < 3.5

The dma function dma_buf_vmap() is not available on kernel < 3.5,
remove the call from the media subsystem code.

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

backports: make media driver USB_MSI2500 depend on kernel 3.5

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

backports: add some more dev_*_ratelimited() functions

Some drivers now also call the other dev_*_ratelimited() functions.

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

backports: revert commit for mwifiex sido.

Revert this commit on kernel version < 3.7

    mwifiex: card reset: enable rescan of non-removable card

    mmc_rescan will scan for non-removable cards only once, hence the card
    will not be rediscovered.

Signed-off-by: Andreas Fenkart <afenkart@gmail.com>
Acked-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

backports: backport argument change in alloc_netdev_mqs()

alloc_netdev_mqs() got a new parameter  in kernel 3.17, this patch
removed it from the calls if an older kernel is used. This is safe,
because older kernel versions do not need it.

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

backports: do not build da9211 driver for kernel < 3.17

This driver needs a new header file, which is not in the kernel < 3.17.

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

backports: fix bit_wait() and bit_wait_io()

In kernel version < 3.17 wait_on_bit() takes an action as a parameter,
which contains a schedule() or io_schedule() call in most cases. For
such calls wait_on_bit() was changed to not take an action but execute
schedule() action. This patch backports the new behavior to old kernel
versions.

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

backports: refresh patches on next-20140710

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.60              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.93              [  OK  ]
6   3.5.7               [  OK  ]
7   3.6.11              [  OK  ]
8   3.7.10              [  OK  ]
9   3.8.13              [  OK  ]
10  3.9.11              [  OK  ]
11  3.10.43             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.22             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.7              [  OK  ]
16  3.15.0              [  OK  ]
17  3.16-rc1            [  OK  ]

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

backports: make AT86RF230 driver depend on kernel > 3.4

AT86RF230 needs the devm_regmap_init_spi() function.

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

backports: disable some more media drivers

These driver need clk_prepare_enable(), clk_disable_unprepare(),
devm_regulator_bulk_get() or devm_regmap_init_i2c()

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

backports: add some ubuntu trusty specific changes

It is now possible to detect a ubuntu kernel see:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1327619

This uses the new define and adds some code depending on the ubuntu
trusty kernel.

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

backports: build some regulator drivers only with kernel 3.17

They are depending in some headers only available in more recent kernel
versions.

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

backports: backport __QUEUE_STATE_{DRV,STACK}_XOFF

The old __QUEUE_STATE_XOFF was divided into __QUEUE_STATE_DRV_XOFF and
__QUEUE_STATE_STACK_XOFF, in this patch we map both to the old version
on old kernel vesions.

This backports this upstream commit:
commit 7346649826382b769cfadf4a2fe8a84d060c55e9
Author: Tom Herbert <therbert@google.com>
Date:   Mon Nov 28 16:32:44 2011 +0000

    net: Add queue state xoff flag for stack

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

backports: add request_firmware_direct()

request_firmware_direct() was newly added and in some places it is
better to use request_firmware_direct(), but request_firmware() will
also work and result in 60 seconds waiting if the file is not there in
some situations.

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

backports: add pci_zalloc_consistent()

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

backports: add cc2520 driver

This adds the cc2520.h and disables the driver on kernel < 3.5.

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

backports: add U16_MAX

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

backports: remove parameter from dma_buf_export()

dma_buf_export() got a new parameter which is not available in older
kernel versions.

This was once removed in this commit and is needed again:
commit 9981d1abf4b52e84a432fc9bdfbdd30d1838819d
Author: Hauke Mehrtens <hauke@hauke-m.de>
Date:   Mon Jun 16 21:02:25 2014 +0200

    backports: Revert "remove parameter from dma_buf_export()"

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

backports: bump minimum kernel version for ADV7604 to 3.13

While doing a make allmodconfig I noticed the following:
  CC [M]  /dev/shm/backports-next/drivers/media/i2c/adv7604.o
/dev/shm/backports-next/drivers/media/i2c/adv7604.c:31:33: fatal error: linux/gpio/consumer.h: No such file or directory

linux/gpio/consumer.h cannot be backported currently, as functions rely
on in-kernel functionality (GPIO_DEVRES). Bumping the minimum required
kernel version for ADV7604 to 3.13 where the header got introduced.

Signed-off-by: Stefan Assmann <sassmann@kpanic.de>
[add comment with reason]
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

Revert "backports: include gpio consumer.h"

This is provides in-kernel functionality, to include
it assuems we have provided a backport for this functionality
and that's not true.

This reverts commit 16472f0c6924f0fa54c68a024a55d89689b413a0.

backports: include gpio consumer.h

While doing a make allmodconfig I noticed the following:
  CC [M]  /dev/shm/backports-next/drivers/media/i2c/adv7604.o
/dev/shm/backports-next/drivers/media/i2c/adv7604.c:31:33: fatal error: linux/gpio/consumer.h: No such file or directory

Including the header fixes the error.

Signed-off-by: Stefan Assmann <sassmann@kpanic.de>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: make regulator palmas driver depend on v3.15

The regulator palmas driver makes use of the new struct regulator_desc
enable_val and disable_val added via commit ca5d1b352 on v3.15 to help
overcome an issue on assumptions on how regulator drivers enable /
disable the regulator with only one value / mask. We can't port this
so just require at least v3.15 for it.

mcgrof@ergon ~/linux (git::master)$ git describe --contains ca5d1b352
v3.15-rc1~148^2~4^3

commit ca5d1b3524b4d90a2e2f1f71583c1dca6b96fd92
Author: Carlo Caione <carlo@caione.org>
Date:   Wed Mar 5 22:11:29 2014 +0100

    regulator: helpers: Modify helpers enabling multi-bit control

    This patch extends the regulator helpers to account for device that use
    multiple bits for control when using regmap enable/disable/bypass ops.

    The actual regulator helpers wrongly assume that the regulator control
    is always performed using single bits, using in the regulator_desc
    struct only two parameters *_reg and *_mask defining register and mask
    for control.

    This patch extends this struct and introduces the helpers to take into
    account devices where control is performed using multiple bits and
    specific multi-bit values are used for enabling/disabling/bypassing the
    regulator.

Signed-off-by: Carlo Caione <carlo@caione.org>
Signed-off-by: Mark Brown <broonie@linaro.org>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

backports: update test kernel versions

This adds kernel 3.16 and updates some other kernel versions.

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

backports: refresh patches on next-20140616

== ckmake-report.log ==

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.60              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.93              [  OK  ]
6   3.5.7               [  OK  ]
7   3.6.11              [  OK  ]
8   3.7.10              [  OK  ]
9   3.8.13              [  OK  ]
10  3.9.11              [  OK  ]
11  3.10.43             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.22             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.7              [  OK  ]
16  3.15.0              [  OK  ]
17  3.16-rc1            [  OK  ]

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

backports: do not include net_ieee802154_lowpan in kernel >= 3.16

In kernel >= 3.16 this backport is not needed at all. When
CPTCFG_IEEE802154_6LOWPAN is not set it should not get build, we do not
have to check if we are kernel > 3.5, because in an other case
ieee802.15.4 will not get build. Kernel version 3.15 needs a different
backport than the older versions.

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

backports: Revert "remove parameter from dma_buf_export()"

The linux commit which made backports need this patch was removed from
linux-next.

This reverts commit 6e1b080e83f1537d8d198917180d0a8bdfd7e0c9.

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

backports: add __pskb_copy_fclone()

__pskb_copy() was renamed to __pskb_copy_fclone() in most places and
got an extra parameter, it should be safe to just ignore this extra
parameter.

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

backports: refresh patches on next-20140606

There were manual changes ineeded in the following files:
patches/collateral-evolutions/network/0034-ndo_set_vf_spoofchk/igb_ndo_set_vf_spoofchk.patch
patches/collateral-evolutions/network/0037-ethtool_get_module/igb_get_module.patch
patches/collateral-evolutions/network/0038-ethtool_rxfh_indir/igb_rxfh_indir.patch

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.59              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.91              [  OK  ]
6   3.5.7               [  OK  ]
7   3.6.11              [  OK  ]
8   3.7.10              [  OK  ]
9   3.8.13              [  OK  ]
10  3.9.11              [  OK  ]
11  3.10.40             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.20             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.4              [  OK  ]
16  3.15-rc7            [  OK  ]

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

backports: remove useless git patch header

When exporting a patch from git it adds a header to it, but this header
is not needed by backports, remove it from the patches.

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

backports: activate NFC_TRF7970A on kernel 3.4

The function devm_gpio_request_one() was backported some time ago.

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

backports: make some regulator drivers depend on more recent kernel versions

These drivers got new and not so easy backportable dependencies.

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

backports: add hashtable.h

This header file is used by net/mac802154/llsec.h and does not provide
any function prototype so it is save to copy it.

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

backports: remove const from pci error handlers on kernel < 3.7

This fixes a warning on these older kernel versions.

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

backports: do not access user_ns on struct net

The ieee802154 subsystem wants to access user_ns on struct net, but
this member was added in kernel 3.8, check the general capabilities
instead.

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

backports: handle rename of ndo_set_vf_tx_rate to ndo_set_vf_rate

This member of struct net_device_ops was renamed and the function
signature was also changed a bit.

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

backports: remove usage of get_ts_info in cdc_ncm.c

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

backports: copy include/linux/platform_data/st21nfca.h

This is needed by the st21nfca nfc driver.

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

backports: add split_page()

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

backports: add of_match_ptr()

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

backports: add of_node_put()

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

backports: remove parameter from dma_buf_export()

dma_buf_export() got a new parameter which is not available in older
kernel versions.

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

backports: add crc7_be()

This was added in:
commit 1836eea209546b870dd83f3f4ef234d6598a560d
Author: George Spelvin <linux@horizon.com>
Date:   Sat May 10 10:32:57 2014 -0400

    lib/crc7: Shift crc7() output left 1 bit

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

backports: add prandom_u32_max()

This was added in:
commit f337db64af059c9a94278a8b0ab97d87259ff62f
Author: Daniel Borkmann <dborkman@redhat.com>
Date:   Wed Jan 22 02:29:39 2014 +0100

    random32: add prandom_u32_max and convert open coded users

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

backports: add U32_MAX

This was added in:
commit 89a0714106aac7309c7dfa0f004b39e1e89d2942
Author: Alex Elder <alex.elder@linaro.org>
Date:   Thu Jan 23 15:54:00 2014 -0800

    kernel.h: define u8, s8, u32, etc. limits

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

backports: remove USHORT_MAX from kernel.h

This define is not in the kernel any more.

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

backports: remove upstream patches

These two patches are now in upstream linux kernel and not needed in
backports any more

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

backport: update drivers_net_wireless_iwlwifi_iwl-debug.patch

The driver was updated - need to update the patch accordingly

commit 7f715626a9ddfa7df13f076f59fc62e3c99066cf
Author: Liad Kaufman <liad.kaufman@intel.com>
Date:   Thu May 8 15:24:54 2014 +0300

    iwlwifi: use dev_printk instead of dev_dbg for debug logs

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

backports: activate some more media drivers

These new drivers are compile tested only.

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

backports: fix a compile warning in devnode

The mode parameter type was changed in the callback function devnode in
struct device_type.

commit 2c9ede55ecec58099b72e4bb8eab719f32f72c31
Author: Al Viro <viro@zeniv.linux.org.uk>
Date:   Sat Jul 23 20:24:48 2011 -0400

    switch device_get_devnode() and ->devnode() to umode_t *

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

backports: add I2C_CLIENT_SCCB

This is needed for some newly activated media driver.

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

backports: add i2c_add_mux_adapter()

This adds i2c_add_mux_adapter() also for kernel < 3.5. This is needed
for some newly activated media driver.

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

backports: remove backport of DMA shared buffers

This DMA shared buffer backport was used for kernel < 3.5, more recent
kernel versions already have a working version of the DMA shared
buffers.
The DMA shared buffers are only used by some media drivers, but non of
these media drivers, which are actually using this API, are compiled on
kernel < 3.5. The DMA shared buffer part would need some more attention
in the next backports version, because there are some bigger changes
and instead we removed it.

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

backports: disable DMA shared buffer support for kernel < 3.5 in v4l2

v4l2 uses the DMA shared buffers API of kernel >= 3.5. When the v4l2
backport is be compiled against older kernel versions a backport of the
DMA shared buffers is used. Instead of using this backport this patch
will remove usage of the DMA shared buffers on these old kernel
versions. This allows us to remove DMA shared buffers from backports.

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