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

This adds basic support for debian packages using the xz compression
algorithm.

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

This converts the backports-update-manager to use python3 instead of
python2. python3 is needed because new Debian packages are compressed
with the xz algorithm and only the tar module of python3 supports this
algorithm.

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

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.69              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.108             [  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  3.10.84             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.44             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.48             [  OK  ]
16  3.15.10             [  OK  ]
17  3.16.7              [  OK  ]
18  3.17.8              [  OK  ]
19  3.18.19             [  OK  ]
20  3.19.8              [  OK  ]
21  4.0.9               [  OK  ]
22  4.1.3               [  OK  ]
23  4.2-rc3             [  OK  ]

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

This was intoduced in this commit:
commit 431a9e1713c512bb96d570dc84e518384d1ac62b
Author: Loic Poulain <loic.poulain@intel.com>
Date:   Wed Jul 1 12:20:26 2015 +0200

    Bluetooth: hci_uart: Add basic support for Intel Lightning Peak devices

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

It seems new versions of python make the os walk recurse more
than we used to before, our ckmake implementation was picking up
more kernels targets than obviously we want to build then. Fix
this by replacing the os.walk() with a simple os.listdir()
which just keeps us at the top level directory of lib/modules.

While at it, this adds the --develdebug option which lets you
debug the the directory naming it will use for display and what
directories it will traverse without actually compiling anything
and without using ncurses. For instance to debug or ckmake without
compilation you can use:

$ ckmake --allyesconfig --develdebug
Paths for kernels:
3.4.103-0304103-generic
3.9.11-030911-generic
3.2.62-030262-generic
3.16.2-031602-generic
3.11.10-031110-generic
3.8.13-030813-generic
3.5.7-03050712-generic
3.0.101-0300101-generic
3.15.10-031510-generic
3.6.11-030611-generic
3.13.11-031311-generic
3.1.10-030110-generic
3.10.54-031054-generic
3.3.8-030308-generic
3.7.10-030710-generic
3.14.18-031418-generic
3.17.0-031700rc3-generic
3.12.27-031227-generic
IDX VERSION	DIRECTORY
1   3.0.101	/home/mcgrof/ksrc-backports/lib/modules/3.0.101-0300101-generic
2   3.1.10	/home/mcgrof/ksrc-backports/lib/modules/3.1.10-030110-generic
3   3.2.62	/home/mcgrof/ksrc-backports/lib/modules/3.2.62-030262-generic
4   3.3.8	/home/mcgrof/ksrc-backports/lib/modules/3.3.8-030308-generic
5   3.4.103	/home/mcgrof/ksrc-backports/lib/modules/3.4.103-0304103-generic
6   3.5.7	/home/mcgrof/ksrc-backports/lib/modules/3.5.7-03050712-generic
7   3.6.11	/home/mcgrof/ksrc-backports/lib/modules/3.6.11-030611-generic
8   3.7.10	/home/mcgrof/ksrc-backports/lib/modules/3.7.10-030710-generic
9   3.8.13	/home/mcgrof/ksrc-backports/lib/modules/3.8.13-030813-generic
10  3.9.11	/home/mcgrof/ksrc-backports/lib/modules/3.9.11-030911-generic
11  3.10.54	/home/mcgrof/ksrc-backports/lib/modules/3.10.54-031054-generic
12  3.11.10	/home/mcgrof/ksrc-backports/lib/modules/3.11.10-031110-generic
13  3.12.27	/home/mcgrof/ksrc-backports/lib/modules/3.12.27-031227-generic
14  3.13.11	/home/mcgrof/ksrc-backports/lib/modules/3.13.11-031311-generic
15  3.14.18	/home/mcgrof/ksrc-backports/lib/modules/3.14.18-031418-generic
16  3.15.10	/home/mcgrof/ksrc-backports/lib/modules/3.15.10-031510-generic
17  3.16.2	/home/mcgrof/ksrc-backports/lib/modules/3.16.2-031602-generic
18  3.17-rc3 	/home/mcgrof/ksrc-backports/lib/modules/3.17.0-031700rc3-generic

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

The 0002-no_dmabuf patch series creates the largest delta for
backports, and it turns out we can generalize it thorugh SmPL.
There's just a few changes we can't get rid of as they relate
to other build things.

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.68              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.107             [  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  3.10.75             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.40             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.39             [  OK  ]
16  3.15.10             [  OK  ]
17  3.16.7              [  OK  ]
18  3.17.8              [  OK  ]
19  3.18.12             [  OK  ]
20  3.19.5              [  OK  ]
21  4.0.0               [  OK  ]
22  4.1-rc8             [  OK  ]

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

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.68              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.107             [  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  3.10.75             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.40             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.39             [  OK  ]
16  3.15.10             [  OK  ]
17  3.16.7              [  OK  ]
18  3.17.8              [  OK  ]
19  3.18.12             [  OK  ]
20  3.19.5              [  OK  ]
21  4.0.0               [  OK  ]
22  4.1-rc8             [  OK  ]

manual changes done to:
patches/collateral-evolutions/media/0005-dma-no-sync/v4l2.patch

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

NFC_PN544_I2C uses devm_gpiod_get_index() now which is not available in
kernel < 3.17.

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

NFC_NXP_NCI_I2C uses the linux/gpio/consumer.h which is not available
in kernel version < 3.13

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

VIDEO_FB_IVTV uses arch_phys_wc_add(), arch_phys_wc_del() and
pat_enabled() now which are not available in older kernel versions.

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

The media framework now uses the frame vector code which is only
available from kernel 4.2 onwards. This backports the code.

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

The get_user_pages_locked() function is needed by the frame vector code
which will be added in the next patch. While adding that, this patch
also replaces the get_user_pages_unlocked() function with the original
version from the next kernel.

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

With kernel 3.12 a new locking was added to struct tty_struct. This
patch makes the code use the old locking.

The new locking was added in this commit:
commit 6a1c0680


Author: Peter Hurley <peter@hurleysoftware.com>
Date:   Sat Jun 15 09:14:23 2013 -0400

    tty: Convert termios_mutex to termios_rwsem

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

The led_trigger_remove() function is exported on older kernel versions,
but there is no pubic prototype for this function. Now the first driver
makes use of this function and this patch adds the prototype.
There is no LINUX_BACKPORT() guard here because we want it to point to
the original function which is exported normally.

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

It should be save to just copy it.
This was added in this upstream commit:
commit 9961127d


Author: Vincent Cuissard <cuissard@marvell.com>
Date:   Thu Jun 11 11:25:47 2015 +0200

    NFC: nci: add generic uart support

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

This is not used by any code in backports and was also removed from the
mainline kernel.

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

In the mainline kernel the kernel module parameter locking is not
protected by a global lock any more, but each module has its own lock
now. This resulted in some changes in the api. This patch converts
calls to the new api to the old one.

This backports the changes done in this commit:
commit b51d23e4


Author: Dan Streetman <ddstreet@ieee.org>
Date:   Wed Jun 17 06:18:52 2015 +0930

    module: add per-module param_lock

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

The new nfcmrvl driver needs a new platform data header file.

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

st21nfcb.h was renamed in the upstream kernel, do the same for
backports. This rename was done in this commit:
commit ed06aeef


Author: Christophe Ricard <christophe.ricard@gmail.com>
Date:   Tue Jun 9 22:26:05 2015 +0200

    nfc: st-nci: Rename st21nfcb to st-nci

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

The same include guard is also used in backport/backport-include
/asm-generic/barrier.h which prevents both files from being included at
the same time.

Reported-by: Martin Tippmann <martin.tippmann@gmail.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

The file net/bluetooth/hci_ldisc.c contains references to tty->termios.
These changed from pointer to instance in 3.7.0. This patch backports
all references by introducing kernel version defines.

Signed-off-by: Florian Grandel <fgrandel@gmail.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

Usage of the bt_tx_busylock variable was already patched but not the
declaration which caused a compile warning. This patch introduces a
define statement that also excludes the variable declaration.

Signed-off-by: Florian Grandel <fgrandel@gmail.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

The net/route.h header is missing when compiling against the android msm
kernel. Explicitly add it here to make the kernel compile.

Signed-off-by: Florian Grandel <fgrandel@gmail.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

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

Using expression is pretty flexible, in this case we
don't want that flexibility as it can override any
struct who also uses the same variable name. For a
specific example of the issue refer to the netdev-ops
demo git tree [0] which provide an example of the issue
with two SmPL rules.

We keep the expression on the last rule on NETLINK_CB(skb)
given that we know only an skb can possibly be used on it.

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.68              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.107             [  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  3.10.75             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.40             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.39             [  OK  ]
16  3.15.10             [  OK  ]
17  3.16.7              [  OK  ]
18  3.17.8              [  OK  ]
19  3.18.12             [  OK  ]
20  3.19.5              [  OK  ]
21  4.0.0               [  OK  ]
22  4.1-rc8             [  OK  ]

The failure on 4.1 is not related to this change,
its because we are using 4.1-rc1, it fails because of
the missing tty_set_termios() is not exported,
4.1-rc4 works.

[0] https://github.com/mcgrof/netdev-ops.git



Cc: cocci@systeme.lip6.fr
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>

When building the kernel there are lots of warnings like this:
drivers/net/wireless/ath/ath10k/ce.c:404:21: warning: unused variable ‘ar_pci’ [-Wunused-variable]
  struct ath10k_pci *ar_pci = ath10k_pci_priv(ar);

They are caused by an empty lockdep_assert_held() which does nothing
with this variable in kernel version < 3.9. This patch replaces this
macro with the version from a recent kernel and silence this warning.

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

1   3.0.101             [  OK  ]
2   3.1.10              [  OK  ]
3   3.2.68              [  OK  ]
4   3.3.8               [  OK  ]
5   3.4.107             [  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  3.10.75             [  OK  ]
12  3.11.10             [  OK  ]
13  3.12.40             [  OK  ]
14  3.13.11             [  OK  ]
15  3.14.39             [  OK  ]
16  3.15.10             [  OK  ]
17  3.16.7              [  OK  ]
18  3.17.8              [  OK  ]
19  3.18.12             [  OK  ]
20  3.19.5              [  OK  ]
21  4.0.0               [  OK  ]
22  4.1-rc1             [  FAIL  ]

Only kernel 4.1-rc1 will fail, because tty_set_termios() is not
exported, 4.1-rc4 works.

manual changes done to:
drivers/media/platform/vim2m.c

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

On kernel version < 3.4 we have to include linux/interrupt.h, because
it otherwise results in a long compile errors about missing HI_SOFTIRQ
for example.

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

The mt7601u driver makes use of __field_struct which was added in Linux
commit 4d4c9cc8

. It is hard to backport this feature, so just remove
the code using it on kernel version < 3.16.

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

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

This is needed by the mt7601u driver.

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

The mt7601u driver makes use of some of these functions and they are
not available on kernel version < 3.19.

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

tty_set_termios() was not exported any more in kernel 4.0-rc1 (commit
632f32e2) and exported again in kernel 4.1-rc2 (commit b00f5c2d

).
This patch adds the backport for kernel 4.0.X, we can not detect kernel
4.1-rc1, which would also need this fix, so bluetooth will have a
compile error on kernel 4.1-rc1, kernel 4.1-rc2 and more recent
versions will work.

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

More v4l drivers are now accessing the vidioc_expbuf function pointer
in struct v4l2_ioctl_ops which we can not backport to kernel < 3.5.
This adds some more checks around the places.

We should convert this to an spatch.

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

Before we only copied the hdmi.h header file and not the driver. Now
some other drivers need the hdmi driver code, also copy it now too.

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

The driver VIDEO_ADP1653 now includes linux/gpio/consumer.h which is
not available in kernel versions < 3.13. This patch makes the driver
depend on kernel version >= 3.13.

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

The VIDEO_ADV7604 driver now makes use of
devm_gpiod_get_index_optional() which gets less parameters in older
kernel versions. This makes the driver depend on kernel >= 3.17. In
addition also the VIDEO_COBALT has to depend on this kernel version
because it selects VIDEO_ADV7604.

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

The drivers MEDIA_TUNER_FC2580 and MEDIA_TUNER_TUA9001 are using
devm_regmap_init_i2c() which is only believable on kernel >= 3.3, add a
dependency to the driver.

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

The drivers DVB_M88DS3103 and DVB_TS2020 are accessing some member in
struct regmap_config which is not available in kernel version < 3.8.
These drivers are getting selected by some other drivers, which are now
also depending on kernel >= 3.8, otherwise KConfig would ignore the
dependency on kernel >= 3.8

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