package/mpd: bump to version 0.19.18

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Update for 2016.08-rc1

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

CHANGES: update with recent changes

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

binutils: fix Blackfin '.rofixup section size mismatch' problems

The bug has already been reported by Stuart Henderson to binutils
upstream: https://sourceware.org/bugzilla/show_bug.cgi?id=13391

Fixes:

  http://autobuild.buildroot.net/results/d5999ae91a1f4dee3d01a0bbc3d8a4c5939ad175/
  http://autobuild.buildroot.net/results/17232204249aeb04150ac43a2424aa26a6b6c807/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

transmission: fix build when using system libminiupnpc with libutp disabled

Fixes:

  http://autobuild.buildroot.net/results/63a3caac7e1e7b5ab268d9be3cd77617df619659/

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gst1-plugins-bad: fix broken dependency for the hls plugin

HLS plugin can be built with nettle or libgcrypt or openssl
cryptographic backend. But current dependency on gnutls is incorrect.
It has been working so far because gnutls depends on nettle.

gst-plugins-bad's build system for HLS allows user to choose which
cryptographic backend to use. If that is not specified, it internally
checks for nettle or libgcrypt or openssl in order. If none of the
cryptographic backend is available, HLS plugin gets disabled internally.

Select cryptographic backend according to which cryptographic packages
are available. If both libgcrypt or openssl are not available, choose
nettle by default.

Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Reviewed-by: Marcin Nowakowski <marcin.nowakowski@imgtec.com>
Signed-off-by: Rahul Bedarkar <rahul.bedarkar@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/libuio: select gettext when BR2_NEEDS_GETTEXT_IF_LOCALE is set

gettext is added to libuio dependency when BR2_NEEDS_GETTEXT_IF_LOCALE
is set without selecting it at Kconfig level.

Fixes:

  http://autobuild.buildroot.net/results/321688e16596086d1dad6f489d35edbb7fe7a955

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

xfsprogs: bump version to 4.7.0

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/oprofile: introduce a BR2_PACKAGE_OPROFILE_ARCH_SUPPORTS

operf_utils.h defines rmb() for a limited number of architectures, so
add this list to BR2_PACKAGE_OPROFILE_ARCH_SUPPORTS to disable any new
or unsupported architectures.

Doing so, this disable oprofile for m68k which lack of memory barrier
operations.

Remove nios2 dependency since it's not supported by oprofile even if
binutils could be built for nios2.

Fixes:

 http://autobuild.buildroot.net/results/1cc761d8a5715d0a2c6eaacfde7e44b225da1b36

Signed-off-by: Romain Naour <romain.naour@gmail.com>
[Thomas: remove BR2_sh64, use BR2_sh instead of BR2_sh4.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/kmsxx: fix static install

Currently, we only try to install the shared libraries.

However, when doing a static build, only the static libs are built,
obviously.

Fix this by only installing relevant libraries.

Fixes:
    http://autobuild.buildroot.org/results/82c/82cfb7451f933b222abe30b5d35d23e409a4af79/
    http://autobuild.buildroot.org/results/bb3/bb3840f14382b6ed77d5947eb3ac5c229286681e/

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Ripard <maxime.ripard@free-electrons.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/kmsxx: detect failure at install-staging

Currently, the staging-install commands do not detect failures to
install the libs.

That's because we use ';' to separate the commands, so we only get the
result of the last one.

Fix that by using multi-line commands, so they each are called on their
own and make would catch the failures.

Partially fixes (only detects the real failure):
    http://autobuild.buildroot.org/results/82c/82cfb7451f933b222abe30b5d35d23e409a4af79
    and a few others.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Ripard <maxime.ripard@free-electrons.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libsepol: fix static build

Commit a3ebe45dcb3e (libsepol: bump to version 2.5, 2016-06-26) didn't refresh
correctly the patch adding static build support. Do this now.

Fixes:
http://autobuild.buildroot.net/results/14b/14bbb7c9c59e7f68f3ae7ad63e075e27a0da2b1e/
http://autobuild.buildroot.net/results/97f/97f61d5e46a7f479e8de5dd19d49f7a3d84cc86e/
http://autobuild.buildroot.net/results/920/920e1ddac6bf93707a187e64a4ef7d95aff09e6c/

Cc: Adam Duskett <Aduskett@gmail.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ncurses: reinstall l/linux terminfo file

This file was installed prior to the refactoring done in commit
33ba5188351ece027455e460f6cba4245af49f3a ("ncurses: factorize terminfo
files installation").

Reported-by: Benoît Allard <benoit.allard@greenbone.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/kodi: Update upstream patch to support libcurl >= 7.50.1

An updated patch was committed to Kodi after Curl added a fix to address
the original problem:
https://github.com/curl/curl/commit/d6604524ad24daf4581efbe0020da058d2b3af84

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gnupg: disable assembly optimization for coldfire

Fixes:
http://autobuild.buildroot.net/results/d25425cff055a14b9285cefb0775bc138eb2f914/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libgcrypt: disable assembly for coldfire

Fixes:
http://autobuild.buildroot.org/results/374/374158bb0a0e6a16af5d0b909cd05ec52812aaa3/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

imx6ulpico: linux.fragment: Disable CONFIG_SMP

imx6ul has currently an issue on kernel 4.7 that causes a stall when running
the "reboot" command.

This issue has been reported in the linux-arm-kernel mailing list, but we
don't have a proper fix at the moment.

This problem is not seen when SMP is disabled, so let's disable it for now
until a proper fix becomes available.

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/sdl_image: remove trailing \ from the last item of {, HOST_}SDL_IMAGE_CONF_OPTS

Signed-off-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/libdvdread: needs host-pkgconf

Add host-pkgconf to the dependencies since pkg-config is used to find
libdvdcss.

Fixes:
http://autobuild.buildroot.net/results/e5f/e5fffdd51bdd6962149a836af1e7a6bbc6bae20c

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

gdb: fix simulator build failures

Following reports from me of build failures of the GDB simulator for the
Blackfin architecture, Waldemar cooked a
patch (0005-fix-sim-compile.patch) that removes the typedef of SIM_CPU,
because there was a redefinition of this typedef for Blackfin. This was
not causing an issue with recent compilers as redefining the same
typedef is valid with recent compilers, but was causing build failures
with gcc 4.4.x.

However, by removing the common definition of SIM_CPU, this patch broke
the build of the GDB simulator on other architectures, which did not had
an architecture-specific redefinition of SIM_CPU (unlike Blackfin).

The crux of the problem is in a commit from Mike Frysinger, that tries
to refactor the SIM_CPU definition into a common one. Except that it
leaves a redefinition of it for Blackfin. Removing this second
definition however doesn't easily work, due to include ordering
issues. The easiest solution is to simply revert the patch from Mike
Frysinger. This allows to fix the build for all architectures and all
compiler versions.

Fixes:

  http://autobuild.buildroot.net/results/3b82c44ee853fab0e0c63881f0705bb659412917/
  http://autobuild.buildroot.net/results/dafbb93ab38a4285ce42436219d552cceb14828b/

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/libvpx: bump version to 1.6.0

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/libvpx: Fix compilation when __uClinux__ is defined

<lddk.h> is a platform-specific header, which is not provided by any C
library today.

Fixes:

  http://autobuild.buildroot.net/results/ae7ba1d5979ebe39aace628fa8ff2c8f08a70a98/

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libdvdnav: fix nios2 cross-compile

The host is misdetected as OS/2.

Fixes:
http://autobuild.buildroot.net/results/d5b30d6b5463928ded1a1c1f1f5345a726f6ffaf/
http://autobuild.buildroot.net/results/917041c7165093f44a5d3b4a5ba56a3d1b59e456/
http://autobuild.buildroot.net/results/dcfc53b6fc25210cf70a1abf0e24b85f0d971101/
http://autobuild.buildroot.net/results/79109d06183efa17f6f68c1ea78cbd9927c5c46e/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Tested-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

eudev: add patch to fix getrandom() system call on ARM64

When the kernel headers in the toolchain don't provide __NR_getrandom,
eudev has a predefined set of values. However, the provided value for
ARM64 is incorrect, and causes a runtime problem on this architecture
when old kernel headers (not providing __NR_getrandom) are used.

This commit adds a patch to eudev to fix this issue.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Tested-by: Kevin Hilman <khilman@baylibre.com>

gcc: bump 4.9 series to 4.9.4

Two patches are removed, as they have been upstreamed:

 - 130-fix_build_with_gcc-6.patch (svn commit 233721, Git commit
   8c3fa311caa86f61b4e28d1563d1110b44340fb2)

 - 920-libgcc-remove-unistd-header.patch (svn commit 226092, Git commit
   e940d7953f06af11d09229a29ecbcc1ba25b378d)

All other patches have simply been refreshed, with no manual edit
needed.

A build+runtime test has been done with an ARM, Cortex-A8, EABIhf, musl
configuration, booted under Qemu.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

linux-headers: bump 3.18.x and 4.1.x series

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

shellinabox: add dependency on !BR2_STATIC_LIBS

shellinabox uses <dlfcn.h>, so a !BR2_STATIC_LIBS dependency is needed.

Fixes:

  http://autobuild.buildroot.net/results/59dd88e31a5f49231c4131b840fbf1bb76e777e7/

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

shellinabox: fix build on musl

This commit fixes the build of shellinabox with the musl C library,
which requires two changes:

 - A patch to include <sys/ttydefaults.h>, which is needed to get the
   definitions of TTYDEF_*

 - A hack to workaround what seems to be a problem in musl itself (musl
   does #define utmp utmpx, which causes some symbol conflicts down the
   road). Since anyway the utmpx implementation is just a set of stubs
   in musl, we simply make shellinabox believe that <utmpx.h> is not
   available by passing the appropriate variable. The musl issue has
   been reported at http://www.openwall.com/lists/musl/2016/08/04/8.

Fixes:

  http://autobuild.buildroot.net/results/1847cab964957da3c9bf4911a5ad3602b3c82431/

Signed-off-by: Olivier Singla <olivier.singla@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

shellinabox: convert patch to a Git formatted one

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ncdu: bump version from 1.9 to 1.11

Signed-off-by: Alexander Dahl <post@lespocky.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/mono: bump to version 4.4.2.11

This patch bumps mono to version 4.4.2.11

Signed-off-by: Angelo Compagnucci <angelo.compagnucci@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Revert "dtc: install libfdt library and headers for host variant"

This reverts commit 5ca9b7cf66941a66cff736aadaf8a31afe55f964. Installing
dtc headers in $(HOST_DIR) causes some build failures of
host-uboot-tools due to a conflicting libfdt.h header.

The possible solution to solve this problem, based on -isystem, has been
reverted has it was causing too many issues that we didn't want to solve
so close to cutting 2016.08-rc1.

Since the headers of libfdt for the host are currently not needed for
any package in the Buildroot tree, it is easier and safer to just revert
this patch.

Fixes:

  http://autobuild.buildroot.net/results/9fba77f4080bf6bc26e341f3f383cdba26c0e14f/

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/imagemagick: bump to version 7.0.2-6

Signed-off-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

ncurses: use foreach make loops instead of shell for loops

The main benefit of using make foreach loops is that they will abort
if one of the iteration of the loop fails. The current for loops will
continue, and only report a failure if the last iteration was a
failure, but will silently ignore other errors.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

ncurses: factorize terminfo files installation

This long list of commands to copy the terminfo files from staging to
target is really silly and can be factorized using a
NCURSES_TERMINFO_FILES variable, which is then iterated through.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

configs: nitrogen*: bump kernel and u-boot version for m4 support

Also adding the '-m4' device tree blobs which allow to disable the
peripherals that are used by the Cortex-M4.

The 6x_bootscript is already setup to load the '-m4' dtb if m4enabled
u-boot variable is set to 1.

The FreeRTOS BSP for both Nitrogen6SX and Nitrogen7 can be found here:
https://github.com/boundarydevices/freertos-boundary

Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

gst1-imx: bump to version 0.12.2

Changelog:
* imxv4l2videosrc:
  * Stricter checks for video device capabilities
  * Add support for more video formats like 8/16-bit grayscale

* pxp: Fix incorrect return code which broke blitter with GStreamer 1.8

* blitter:
  * Improvements in error logging
  * Fixes for RGBx framebuffer format detection
  * Fix for subtle tearing problem by implementing triple buffering

* audio:
  * Replace objdump-based scan during build configuration
    Codec filenames are instead moved to the C code; the libdir
    path of the libfslaudiocodec.pc file is used to know where on
    the target the files are placed
  * Explicitely add the libfslaudiocodec.pc libdir to the MP3 encoder
    Fixes build errors in buildroot
  * Minor logging fixes

* misc:
  * Proper check for region copies in phys_mem_meta
  * Add missing G_END_DECLS lines in headers

Tested with the following commands on i.MX6Q (IPU):
 # gst-launch-1.0 playbin uri=file:///root/tears_of_steel_1080p.webm
 # gst-launch-1.0 filesrc location=/root/tears_of_steel_1080p.webm ! \
   matroskademux ! imxvpudec ! imxipuvideosink
 # gst-launch-1.0 filesrc location=/root/tears_of_steel_1080p.webm ! \
   matroskademux ! imxvpudec ! imxeglvivsink

Tested with the following commands on i.MX7 (PXP):
 # gst-launch-1.0 imxv4l2videosrc device=/dev/video1 ! imxpxpvideosink

Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

imx-parser: add missing configure options

Enable hard float support when the configuration contains:
BR2_ARM_EABIHF=y

See Yocto equivalent:
https://github.com/Freescale/meta-fsl-arm/blob/krogoth/recipes-multimedia/imx-parser/imx-parser.inc

Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

imx-parser: add install hooks to fix libraries path

By default, all the libraries are installed under /usr/lib/imx-mm
which causes problems at runtime.

[Peter: drop -mindepth/-maxdepth, add comment explaining why this is done]
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

imx-codec: add missing configure options

Enable VPU and hard float support when the configuration contains:
BR2_ARM_EABIHF=y
BR2_PACKAGE_IMX_VPU=y

See Yocto equivalent:
https://github.com/Freescale/meta-fsl-arm/blob/krogoth/recipes-multimedia/imx-codec/imx-codec.inc

Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

imx-codec: add install hooks to fix libraries path

By default, all the libraries are installed under /usr/lib/imx-mm
which causes problems at runtime.

The hooks are inspired from the mechanism used in the Yocto recipe:
https://github.com/Freescale/meta-fsl-arm/blob/krogoth/recipes-multimedia/imx-codec/imx-codec.inc

Tested with the following commands:
 # gst-launch-0.10 audiotestsrc ! mfw_mp3encoder ! fakesink
 # gst-launch-1.0 audiotestsrc ! imxmp3audioenc ! fakesink

[Peter: drop -mindepth/-maxdepth, add comment explaining why this is done]
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

system: move locale purging options to the "System configuration" menu

The options to purge locales and to generate locale data are currently
located in the toolchain menu. However, these options are not really
related to the toolchain per-se, they are more system-level
configuration options, much like the timezone selection option we
already have in the "System configuration" menu.

Therefore, it makes more sense to have the locale-related options in
the "System configuration" menu as well.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

toolchain: reduce the list of locales to keep by default

Our current list of locales to keep by default is "C en_US de fr". It
doesn't make much sense to keep "de" and "fr" more than any other
language. So let's keep only the "C" and "en_US" locales by default,
and leave it to the user to specify other locales to keep if needed.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

toolchain: purge locales by default

Our current default is to keep all locales installed in
/usr/share/locale/. However, in practice, those locales take up a
significant amount of space, and most users do not need
locales. Therefore, it makes more sense to default to purging locales,
in order to keep only a few useful ones rather than keeping them all.

It helps in providing a small filesystem size by default, and still
allows advanced users who really need locales to tune their
configuration.

As an example, a very basic system with just util-linux enabled (not
even Busybox) weights 11 MB, including 6.4 MB of locales. With this new
default, the generated system is only 4.2 MB.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

config: update synopsys defconfigs

With this commit we update ARC defconfigs with the following:

  - "snps_axs101_defconfig" and "snps_axs103_defconfig":
     - bump linux kernel version to 4.7
     - set up host linux headers to 4.7
     - bump u-boot version to 2016.07

  - "snps_hs38_smp_vdk_defconfig":
     - bump linux kernel version to 4.7
     - set up host linux headers to 4.7

Also we delete "snps_hs38_vdk_defconfig" as we doesn't support
uni-processor HS38 configuration for vdk any longer.

Signed-off-by: Vlad Zakharov <vzakhar@synopsys.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libcurl: security bump version to 7.50.1

Contains fixes for CVE-2016-5419..5421:
https://curl.haxx.se/docs/vuln-7.50.0.html

[Peter: extend commit message with CVE info]
Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

alsa-utils: bump version to 1.1.2

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

alsa-lib: bump version to 1.1.2

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libsigsegv: update web page URL

The web page has moved.

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

initscripts: show status in S40network

The output of "/etc/init.d/S40network start/stop" was lacking a
newline due to the usage of printf. Fix it by echoing the status, like
we do in other init scripts.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

toolchain: libatomic is not built on "uclinux" platforms

The cairo package fails to build on some architectures:
BR2_TOOLCHAIN_HAS_LIBATOMIC is enabled, but libatomic is in fact not
available.

This happens because the gcc logic in libatomic/configure.tgt does not
recognize "uclinux" as a valid OS part of the target tuple, and
therefore it does not build libatomic.

The "uclinux" part of the tuple is used by Buildroot when
BR2_BINFMT_FLAT=y, so we make BR2_TOOLCHAIN_HAS_LIBATOMIC enabled only
if !BR2_BINFMT_FLAT.

It is worth mentioning that support for the uclinux tuple could most
likely very easily be added to gcc: it could rely on the generic
"posix" implementation of libatomic, which uses pthread locks,
available on all architectures where thread support is available.

Fixes:

 [arm]  http://autobuild.buildroot.net/results/3d8dc45e41a043d2c2c26bfb26c3617499fbe671
 [m68k] http://autobuild.buildroot.net/results/318e01406e3e92eb589ee5b2231c671a4dbb6da4

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
[Thomas: adjust dependency after analysis of the gcc code.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/boost: context module needs ARM instructions on ARM

The boost-context module contains ARM assembly code, which uses ARM
instructions. Therefore, on Thumb-2 only platforms (such as Cortex-M),
the following build error occurs:

libs/context/src/asm/make_arm_aapcs_elf_gas.S: Assembler messages:
libs/context/src/asm/make_arm_aapcs_elf_gas.S:33: Error: unshifted register required -- `bic a1,a1,#15'
libs/context/src/asm/make_arm_aapcs_elf_gas.S:42: Error: immediate value out of range

To avoid this error, we make the boost-context module depend on
BR2_ARM_CPU_HAS_ARM.

Fixes:

  http://autobuild.buildroot.net/results/8f6770b59a343b9f710e9363b43227ee9f026660

Signed-off-by: Romain Naour <romain.naour@gmail.com>
[Thomas: use BR2_ARM_CPU_HAS_ARM instead of a dependency on
!BR2_ARM_CPU_ARMV7M.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

readline: add project URL

Signed-off-by: Rahul Bedarkar <rahul.bedarkar@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ncurses: install PuTTY terminfo files

Signed-off-by: David Glessner <david.glessner@rockwellcollins.com>
Signed-off-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

lldpd: bump version to 0.9.4

With this new version, the generation of atom-glue fails with the
following error:

    Making all in src/lib
    make[3]: Entering directory '/mnt/data/src/sfl/zodiac/buildroot/output/build/lldpd-0.9.4/src/lib'
      GEN      atom-glue.c
    arm-buildroot-linux-uclibcgnueabi-cpp.br_real: fatal error: too many input files
    compilation terminated.
    arm-buildroot-linux-uclibcgnueabi-cpp.br_real: fatal error: too many input files
    compilation terminated.
    make[3]: *** [Makefile:899: atom-glue.c] Error 1

There is an upstream patch pushed after 0.9.4 which fixes the
corresponding Makefile.am, included in this commit.

Note that since the provided tarball ships the related Makefile.in file,
we need to tell Buildroot to autoreconfigure the package.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
[Thomas: add comment about AUTORECONF=YES in .mk file.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

lldpd: add sysv init script

Add a simple sysv init script to start the lldpd daemon at boot.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
[Thomas: cosmetic improvements.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/openpowerlink: bump to v2.4.1

Remove upstream patches.

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libseccomp: bump version to 2.3.1

Bump libseccomp version to 2.3.1 to enable power architectures.

Signed-off-by: Judd Meinders <judd.meinders@rockwellcollins.com>
Signed-off-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

i2c-tools: install binaries to /usr/sbin

Upstream i2c-tools executables are installed to /usr/sbin by default.
Make buildroot match this behavior.

Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

i2c-tools: add busybox to dependencies if enabled

Before changing the install directory of i2c-tools to /usr/sbin/ make
sure busybox is built first (if selected) in order to ensure that the
symlinks to the busybox-provided versions are overwritten by the
i2c-tools' install commands.

Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

tcpdump: bump version to 4.8.0

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

util-linux: use upstream patch to fix uClibc build issue

Upstream has this patch a while and other projects are using
it already. So better switch to this.

Signed-off-by: Waldemar Brodkorb <wbx@uclibc-ng.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

i2c-tools: depend on BUSYBOX_SHOW_OTHERS

As of version 1.25.0 busybox by default includes a nearly complete
implementation of i2c-tools. Hide the upstream version unless the
BR2_PACKAGE_BUSYBOX_SHOW_OTHERS option is selected.

Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

clapack: disable for m68k coldfire

Either a assembler bug or a platform limitation.
Fixes:
http://autobuild.buildroot.net/results/d24a2e9b35c98a10585cd28b3306a09c49163296/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

assimp: fix m68k uclinux compile

The GCC manual suggest when getting:
relocation truncated to fit: R_68K_GOT16O foobar
to use -mxgot.

https://gcc.gnu.org/onlinedocs/gcc/M680x0-Options.html

Fixes:
http://autobuild.buildroot.net/results/63898c00929b4546279bda52148b218f847714ed/
http://autobuild.buildroot.net/results/bb7a107d1ca6e8713e6ccffe6c61c43b777fb962/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

openssl: fix m68k uclinux compile

The GCC manual suggest when getting:
relocation truncated to fit: R_68K_GOT16O foobar
to use -mxgot.

Fixes:
http://autobuild.buildroot.net/results/455fd0f274bfa4bbd786bcd6740ecf960e47c1bd/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

uclibc: disable PIE for microblaze

PIE is not supported at the moment for uClibc-ng/microblaze.
Fixes:
http://autobuild.buildroot.net/results/a5848753eb0e2b0ddf55e226c45b85fd3a1c84f3/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

toolchain: Bump ARC tools to arc-2016.09-eng007 and enable PIE

As described at:
4520524ba055706236db9f00dd79f1b2e2e87fde
this commit continues a series of updates of ARC tools.
This time we're updating tools to arc-2016.09-eng007 tag plus a
couple of fixes on top of it that will all make its way in the
next engineering build.

We hope this patch will cure most buildroot ARC failures as it
contains important fixes:

1) PIE fix. We have added PIE support to ARC toolchain at last.
So that should prevent breakage of many packages. As ARC now
supports PIE we remove ARC from BR2_TOOLCHAIN_SUPPORTS_PIE
exclusion in toolchain/Config.in file.

2) Assembler fix. This patch also have changes that fixes frequent
assembler failures, e.g.:
http://autobuild.buildroot.net/results/543/5430b902d900943a34c1888e7e410bd5df367bc2//

We still keep GDB as it is of arc-2016.03 release because there're some
issues we'd like to resolve before releasing it to wider audience.

So again note this is next engineering builds of arc-2016.09 series
and it might have all kinds of breakages, please don't use it for
production builds.

Signed-off-by: Vlad Zakharov <vzakhar@synopsys.com>
[Thomas: remove uClibc PIE patch, since we have bumped uClibc in the
mean time, to a version that contains the PIE fix for ARC.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/luvi: bump to version 2.7.5

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/luv: bump to version 1.9.1-0

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

lighttpd: security bump version to 1.4.41

Security fixes and fixes for bugs introduced in 1.4.40.

Drop 0002-autobuild-clock_gettime-lrt-with-glibc-2.17.patch, which is now
upstream.

Due to the patch removal, autoreconf is no longer needed.

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
[Thomas: remove autoreconf, no longer needed.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

uclibc: update to 1.0.17

Many cleanups and ARC improvements including PIE support.
Patches included upstream and removed.

Signed-off-by: Waldemar Brodkorb <wbx@uclibc-ng.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

openssh: bump version to 7.3p1

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

configs: imx6ulpico: Add Wifi support

imx6ulpico has a BCM4339 Wifi chip. Add Wifi support by default
to allow a better customer experience.

The dts patch has already been sent to the linux-arm-kernel list
and we can remove it once it reaches a mainline kernel (in
version 4.9 probably).

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

configs: warpboard: Select wireless related packages

Let the wireless packages be selected by default to make the
usage of Wifi a bit simpler.

Suggested-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/xenomai: disable for musl toolchains

After fixing the <bits/local_lim.h> issue by using <limits.h>, the
build still fail with udefined __WORDSIZE.

Adding <sys/user.h> to define __WORDSIZE reveal that xenomai use
<error.h> which is not provided by musl.

The discussion on the musl mailing list [1] about glibc error reporting
functions concluded to not add this support in the C library.

For now, disable Xenomai user space support for musl toolchains since
it require several patches to build correctly and needs to remove each
glibc error reporting functions.

Users interested in musl support for Xenomai can work with upstream to
fix these issues and revert this patch.

[1] http://www.openwall.com/lists/musl/2014/06/29/8

Fixes:
http://autobuild.buildroot.net/results/e36/e36a21c1df33bdd3fbc61d516a3e8f7c5f7c41af

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/xenomai: disable for cortex-M cpus

Following include/asm-arm/features.h, Xenomai doesn't support cortex-M
cpus.

error "Could not find current ARM architecture"

Fixes:
http://autobuild.buildroot.net/results/06a/06a52b2c2c23df595f58feb6b95437fc65a86d31

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/xenomai: introduce a BR2_PACKAGE_XENOMAI_ARCH_SUPPORTS option

Introduce BR2_PACKAGE_XENOMAI_ARCH_SUPPORTS to hold the
architectures than can build and use Xenomai on the target.

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

configs/freescale: drop unneeded exit statement in post-image scripts

There is no need for the 'exit' statement in the post-image scripts,
so just remove it.

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

boost: disable wave support for coldfire

A big autogenerated C file is used, which does not fit
in the limited range provided for coldfire.
See here for a longer discussion about the type of issue:
http://uclinux-dev.uclinux.narkive.com/9HrogvzP/fatal-error-tried-to-convert-pc-relative

Fixes:
http://autobuild.buildroot.net/results/c0a8b73901956b393bebe7df3b7af26ade26fcbf/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

warpboard: README: Improve instructions about using wifi

Provide a more complete set of instructions on how to get Wifi
working on the warpboard.

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

uclibc: add missing defines for m68k/microblaze

POLLRDHUP supported since Linux 2.6.17.
http://man7.org/linux/man-pages/man2/poll.2.html

Fixes:
http://autobuild.buildroot.net/results/0f5ed287f1756fbec710454e9c694202c3206dbe/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Config.in: fix typo

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

strace: bump to version 4.13

Drop upstream patch.

Remove preseed of netlink.h and if_packet.h tests since these tests were
removed from configure.ac.

Add stronger sha256 hash.

Version 4.13 fixes a typo in configure.ac (commit 7c0e8875a432) that causes
build failure with kernel v4.7 headers, thus fixing:
http://autobuild.buildroot.net/results/aa6/aa6badae8b3d58d6d48161f261988103271bc2ef/
http://autobuild.buildroot.net/results/4ad/4ad1b0c42206243bd76e94b34454d4108fb31070/
http://autobuild.buildroot.net/results/1c6/1c6c88a7163529a666d1661969736f66e9aba7f7/

and more.

Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

wavemon: depends on a toolchain with NPTL

The software uses pthread_mutexattr_setrobust which isn't
available in uClibc-ng Linuxthreads support.

Fixes:
http://autobuild.buildroot.net/results/d81b199c74cb4251f0b0eb975f66cb5016eb464c/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

triggerhappy: unbreak build against 4.7+ kernel headers

Fixes:
http://autobuild.buildroot.org/results/3d0/3d07de291ec64a16d03e10f0cf1faeb19d546412/
http://autobuild.buildroot.org/results/b03/b03500e7d85786c6eee8f0adc38801340129aa14/
http://autobuild.buildroot.org/results/dd0/dd05caf0db318a9287d239cc6fa5470ccdc2ce7c/
http://autobuild.buildroot.org/results/76b/76bb000d041bd0828d15aa04138e222a4773ce94/

Triggerhappy used the wrong constants for array sizes, causing a compilation
error about out of array bounds initializers with the SW_PEN_INSERTED switch
event added in 4.7. Fix it by using the correct constants (<foo>_CNT) instead.

Patch submitted upstream: https://github.com/wertarbyte/triggerhappy/pull/12

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

grub2: fix build with BR2_STRIP_none

grub2 assumes the strip command will generate output and the output should
always be stripped - so, just use the $(TARGET_CROSS)strip to make sure that
the build succeeds regardless of the buildroot strip configuration.

Signed-off-by: Charles Hardin <ckhardin@exablox.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

yasm: add license

Signed-off-by: Julien Floret <julien.floret@6wind.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

iozone: remove dead code for !thread handling

iozone requires thread support since commit
06b53255cf639a5b033a2cdbfcf114770b94a83b, so having code to handle the
non-threaded case in the .mk file is useless.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

iozone: depends on toolchain with NPTL support

The existent patch is not enough, as pthread_barrier_wait is used, which
isn't implemented in uClibc-ng Linuxthreads.
Remove the patch.

Fixes:

  http://autobuild.buildroot.net/results/e5bbb80c81f4f170cf48d375e016e87a296ff754/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

autofs: add patch to force result of procfs check

The autofs configure script verifies on the build system if /proc is
provided by a Linux system. This might fail if /proc is not mounted,
but does ont mean that procfs will not be available in the target.

Therefore, this commit adds a patch improving the AF_LINUX_PROCFS
autoconf macro to allow passing an autoconf cache variable. autofs.mk
then uses that to force the fact that /proc will be available on the
target.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

tinycbor: add patch to fix installation with old make versions

This commit adds a patch to the tinycbor package that fixes the
installation process with old make version.

This fixes the build on autobuilders using an old make, where tinycbor
was creating empty directories instead of installing files.

Fixes:

  http://autobuild.buildroot.net/results/7605c301f460398c2435f6b73d6d07c7d1022e8c/

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

zynq: unify readme.txt for all Zynq boards

The readme.txt for the three boards are almost the same, so merge
them into board/zynq/readme.txt.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

zynq_zc706: bump U-Boot to xilinx-v2016.2

This is the latest release tag in the Xilinx repository, based on
U-Boot v2016.01 in the mainline.  It includes proper ps7_init_gpl.c/h
in it, so builds working SPL without any manual intervention.

BR2_TARGET_UBOOT_BOARD_NAME should be changed to zynq_zc706 since
U-Boot now has separate defconfig files for ZC702 and ZC706 boards.

BR2_TARGET_UBOOT_SPL_NAME should be changed to spl/boot.bin since
the Zynq image support for the mkimage tool was upstreamed and it
now generates boot.bin under spl/ directory.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

zynq_zc706: bump Linux to xilinx-v2016.2

This is the latest release tag in the Xilinx repository, based on
Linux v4.4 in the mainline.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

zynq_microzed: bump U-Boot to xilinx-v2016.2

This is the latest release tag in the Xilinx repository, based on
U-Boot v2016.01 in the mainline.  To sync with the Zedboard updates,
switch over to the Xilinx GitHub repository.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

zynq_microzed: bump Linux to xilinx-v2016.2

This is the latest release tag in the Xilinx repository, based on
Linux v4.4 in the mainline.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

zynq_zed: bump U-Boot to xilinx-v2016.2

This is the latest release tag in the Xilinx repository, based on
U-Boot v2016.01 in the mainline.

Commit 46d6a561be41 ("zedboard: Generate BOOT.BIN automatically")
switched from the Xilinx GitHub repository to the U-Boot mainline,
but this commit gets it back to the Xilinx GitHub repository for the
following reasons:

 - This defconfig still relies on the Xilinx's local repository for
   the kernel.  The vendor does the best test for the combination of
   U-Boot and the kernel with the same release tag (xilinx-v2016.2
   in this case).

 - At the time of commit 46d6a561be41, the u-boot-xlnx still needed
   manual copy of ps7_init(_gpl).c/h in order to build a working SPL
   image.  So, the mainline U-Boot had advantage in that point of
   time.  However, the improvement in the mainline U-Boot was merged
   into the u-boot-xlnx at the xilinx-v2015.3 release.  Now, the
   mainline and the u-boot-xlnx are even in this point of view.

 - The mainline U-Boot defaults to boot FIT, so something must be
   done; either patch environments around with a local patch to
   switch to uImage booting (current solution), or build an FIT
   with a post build script (chromebook snow does this).  On the
   other hand, the Xilinx repository defaults to uImage booting,
   so it is straightforward, and needs no addition care.

This commit does:

 - Switch to the Xilinx custom git repository, and stick to the
   xilinx-v2016.2 tag.

 - Delete the local patch board/avnet/zedboard/uboot/0001...
   since the Xilinx custom repository can boot uImage by default.

 - Enable BR2_TARGET_UBOOT_FORMAT_IMG instead of ..._FORMAT_DTB_IMG
   since all the Zynq boards in U-Boot enable CONFIG_OF_EMBED.

 - Replace BR2_TARGET_UBOOT_ZYNQ_IMAGE with BR2_TARGET_UBOOT_SPL_NAME
   since U-Boot can natively generate the Zynq boot image now. The
   Zynq image support for mkimage tool was upstreamed at v2016.01
   (so xilinx-v2016.1 as well), so no additional tool is needed
   any more.

 - Update readme.txt

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

zynq_zed: bump Linux to xilinx-v2016.2

This is the latest release tag in the Xilinx repository, based on
Linux v4.4 in the mainline.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Revert "package/Makefile.in should grab HOST_DIR headers using -isystem instead of -I."

This reverts commit 6f8162cf8c1abef7e0a4771fe0d6b26a28f5c2b6. This is
causing too many problems that are not easy to solve.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>