package/enlightenment: remove elementary dependency

elementary is now provided by efl package, so remove elementary package
dependency.

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

package/efl: bump to 1.18.0

Elementary, evas-generic-loaders and emotion-generic-loaders are now
part of efl 1.18 package [1].

Merge the elementery and evas-generic-loaders package to the efl one.

Rename librsvg option (srvg/libsrvg)

[1] https://phab.enlightenment.org/w/efl_and_elementary_1_18/

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

package/ffmpeg: add optional support for opencv3

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

package/opencv: Fix compiling VFP assembler code

Fixes build errors in ffmpeg
http://autobuild.buildroot.net/results/c32/c32a21240a9933796ee850349a62ff3c2314f25c/
http://autobuild.buildroot.net/results/22a/22a9769d83d2a398bb6b264d342b2077c41c9cc5/

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

package/opencv3: use BR2_TOOLCHAIN_HAS_LIBATOMIC

It uses __atomic_fetch_add_4 so libatomic must be pulled in if
necessary. Fixes
http://autobuild.buildroot.net/results/652/6521223b426c3cb4dfefab27bc6021900db6910e/

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

webkitgtk: bump to version 2.12.5

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

e2fsprogs: bump to version 1.43.3

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

mpd: bump to version 0.19.19

It fixes runtime issues with ffmpeg 3.1.x

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

mpd-mpc: bump to version 0.28

The download has a .sig file so use it.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

ncmpc: bump to version 0.25

The download has a .sig file so use it.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

mpv: bump to version 0.20.0

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libenca: bump to version 1.19

Switch to homepage download URL to get the smaller xz-compressed variant.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux-headers: bump 3.{12, 18}.x and 4.1.x series

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

btrfs-progs: bump to version 4.7.2

Drop all of the upstreamed patches (not precisely in the same form) and
rebase the only required patch for proper static builds with staging
libs required by other packages.
Build-tested for arm-glibc, arm-uclibc, arm-musl and arm-uclibc-static.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

lxc: use uptream patch to handle --disable-werror

This new patch does change behavior, but it was accepted by upstream and
could be removed on next version bump.

Signed-off-by: Jérôme Pouiller <jezz@sysmic.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libgtk3: fix typo

It's wayland, not waylandd, hence gtk3+ with the wayland backend went
completely poof.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/amd-catalyst: fix legal-info

Fixes
http://autobuild.buildroot.net/results/4f1/4f1e9b4bf43add4712e82d4a58e633abb2a7d2b8/

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

package/libva-intel-driver: bump version to 1.7.2

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

package/libva: bump version to 1.7.2

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

package/kodi: Fix compilation of kodi-pvr-vbox with gcc6

Fixes a compile error not yet caught by autobuilders:

In file included from /home/bernd/buildroot/br2/output/build/kodi-pvr-vbox-15e864d160da5a051e18aef06f3a53e49808be02/src/client.cpp:24:0:
/home/bernd/buildroot/br2/output/host/usr/i586-buildroot-linux-uclibc/sysroot/usr/include/kodi/xbmc_pvr_dll.h:423:63: error: 'SEEK_SET' was not declared in this scope
   long long SeekLiveStream(long long iPosition, int iWhence = SEEK_SET);
                                                               ^~~~~~~~
/home/bernd/buildroot/br2/output/host/usr/i586-buildroot-linux-uclibc/sysroot/usr/include/kodi/xbmc_pvr_dll.h:510:67: error: 'SEEK_SET' was not declared in this scope
   long long SeekRecordedStream(long long iPosition, int iWhence = SEEK_SET);
                                                                   ^~~~~~~~
CMakeFiles/pvr.vbox.dir/build.make:62: recipe for target 'CMakeFiles/pvr.vbox.dir/src/client.cpp.o' failed

using this defconfig

BR2_TOOLCHAIN_BUILDROOT_WCHAR=y
BR2_BINUTILS_VERSION_2_26_X=y
BR2_GCC_VERSION_6_X=y
BR2_TOOLCHAIN_BUILDROOT_CXX=y
BR2_INIT_NONE=y
BR2_PACKAGE_KODI=y
BR2_PACKAGE_KODI_PVR_VBOX=y
BR2_PACKAGE_MESA3D=y
BR2_PACKAGE_MESA3D_DRI_DRIVER_I965=y
BR2_PACKAGE_MESA3D_OPENGL_EGL=y
BR2_PACKAGE_XORG7=y

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

micropython: replace local fix with upstream one

The fix from commit 26248571b625 ("micropython: fix build failures") was
applied upstream. This replaces the local buildroot fix with the patch
that was accepted upstream.

Signed-off-by: Chris Packham <judge.packham@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

toolchain: Bump ARC tools to arc-2016.09-eng011

As described at:
4520524ba055706236db9f00dd79f1b2e2e87fde
this commit continues a series of updates of ARC tools.
This time we're updating tools to arc-2016.09-eng011.

This engenering build contains the following updates:
1. rebase binutils on top of the latest upstream master
2. update GCC to version 6.2

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

package/{mesa3d, mesa3d-headers}: bump version to 12.0.2

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

e2fsprogs: bump version to 1.43.2

- Upstream patches removed.
- Do not autoreconf since this was neecessary for one of those patches
  that are now upstream. Also remove host-gettext dependency for the
  same reason.

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

git: bump version to 2.10.0

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

mutt: bump version to 1.7.0

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

nano: bump version to 2.7.0

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

privoxy: bump version to 3.0.26

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

qemu: bump version to 2.7.0

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

package/binutils: Disable -Os for ARM cortex-M cpus

ERROR: reloc type R_ARM_THM_MOVW_ABS_NC unsupported in this context
ERROR: reloc type R_ARM_THM_MOVT_ABS unsupported in this context
2 bad relocscollect2: error: ld returned 1 exit status

Binutils build fine with -O1, -O2 or -O3.

Reported upstream:
https://sourceware.org/bugzilla/show_bug.cgi?id=20552

Fixes:
http://autobuild.buildroot.net/results/e72/e727e6e9cf361acc786acec192a88f5f5444a2f6

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

micropython: fix build failures

Fixes:

  http://autobuild.buildroot.net/results/061c66987e9c33a6641c8183f5e0badae516fc1d
  http://autobuild.buildroot.net/results/62e5b5c6d9dca0f41fb4e7d462ebfbb02f8d29da
  http://autobuild.buildroot.net/results/a6437a49659a7b2983269e758dba9fa5a29240d7

Signed-off-by: Chris Packham <judge.packham@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libwebsockets: bump to version 2.0.2

Signed-off-by: Danomi Manchego <danomimanchego123@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/jpeg: properly declare providers

libjpeg and jpeg-turo are providers of the jpeg virtual package, so they
must declare themselves as such.

We do not have any problem, because the dependency is computed from the
Kconfig-defined option BR2_PACKAGE_PROVIDES_JPEG, and we do not check
the Makefile-defined <FOO>_PROVIDES variable.

For the sake of correctness, make both libjpeg and jpeg-turo declare
themselves as providers for the jpeg virtual package.

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

package/openssl: depend on the virtual cryptodev package

Instead of depending on each provider of cryptodev, make openssl depend
on the virtual package.

This is easy because in both cases the openssl build systems needs the
same configuration options.

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

package/cryptodev: properly declare providers

cryptodev-linux and ocf-linux are providers of the cryptodev virtual
package, so they must declare themselves as such.

We do not have any problem for now, as the only consumer (openssl) of
both cryptodev-linux and ocf-linux did not depend on the virtual package
but on each provider. The other consumer is gnutls, but has only a
dependency on cryptodev-linux.

However, openssl can be turned into a proper consumer of the virtual
package, which would trigger the bug.

For the sake of correctness, make both cryptodev-linux and ocf-linux
declare themselves as providers for the cryptodev virtual package.

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

package/dosfstools: Fix undefined PATH_MAX under musl

Dosfstools doesn't build with musl only if BR2_PACKAGE_HAS_UDEV is set
due to a missing include for PATH_MAX (limits.h).

Fixes:
http://autobuild.buildroot.net/results/346/346feda6a4a83850dbb62919eb33482302053490

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

lxc: revive "drop-werror" patch

commit 5050a94ae9f448641e05ce29c931671ffdc62784 dropped patch that
remove '-Werror' from CFLAGS. However, it seems this patch is still
necessary:

  http://autobuild.buildroot.net/results/b1cbca6d0396863654b1d09ccc3163c5f6124ab8
  http://autobuild.buildroot.net/results/fcf2834ad74b95548d25dad2274704ea401f9665
  http://autobuild.buildroot.net/results/048be2fc702c9dba9ca4439ff687d71b30c10551
  http://autobuild.buildroot.net/results/939246c9a4f433dfd0dc414988f5957441b8e9b4

Signed-off-by: Jérôme Pouiller <jezz@sysmic.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

agentpp: bump to 4.0.6

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

snmppp: bump to 3.3.7

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

mpv: remove stray patch

It's remaining from the master-next merge and no longer applies.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Merge branch 'next'

Quite some conflicts, so here goes ..

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

Kickoff 2016.11 cycle

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

docs/website/news.html: add 2016.08 announcement link

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

Update for 2016.08

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

mpd: add optional support for libshout

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

libshout: add optional openssl support

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

package/libshout: add patches to fix static linking

This series includes two patches to fix static linking with libshout.

The first patch fixes two typos resulting in an uninterpretable library
propagation and is mainly needed for the follow-up patch.

The second patch makes sure that `@SHOUT_REQUIRES@` and therefore
`Requires.private` includes ssl as `Requires.private` defines a list of private
packages required by this package but not exposed to applications.

Note that autoreconf is enabled as the first patch touches configure.ac.

Fixes:
http://autobuild.buildroot.net/results/b82/b82408fccbf816575125dcc1a1de7260f66f62e8/
http://autobuild.buildroot.net/results/360/3605750023267e4cd35d98451bd283f9e70a6106/
http://autobuild.buildroot.net/results/03b/03b98b313dd6ed2909607cdc0c4f6fb87be3ccfe/
http://autobuild.buildroot.net/results/70b/70ba47e04d1fc0d2379bf1ac7aab4c52d6193001/

[Peter: fix libssl .pc reference, fix typo]
Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/luajit: disable for mips{32,64}r6

LuaJIT does not have support for MIPS Release 6 [1]. If LuaJIT is
invoked with the option `-b -a mips` to save the bytecode for the MIPS
architecture it generates object files with the MIPS32 Release 1
instruction set.

GCC added support for MIPS32R6 in version 5. However, object files for MIPS32R1
cannot be linked with object for MIPS32R6.

That's why building the package luvi fails for MIPS32R6, where object files
created with GCC are linked with object files created with LuaJIT with
the following linker error:

```
linking mips:isa32 module with previous mips:isa32r6 modules
```

Disable LuaJIT for MIPS32 and MIPS64 R6 to fix the build issue with the
luvi package.

Fixes:
http://autobuild.buildroot.net/results/e01/e017671c33314e5bf2d7fc4920f6c0e64a132383/
http://autobuild.buildroot.net/results/35f/35f558453c45a27367e80e9b8cbe3bd2dccbfd54/
http://autobuild.buildroot.net/results/8e0/8e001112b1678f9fb1c198c84184cd6f5fc3d004/

[1] http://www.freelists.org/post/luajit/Support-for-MIPS32r6,1

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

mplayer: fix detection of X.org

As reported in bug #8206, the mplayer configure script fails to detect
the availability of X11 header/library if the X.org development packages
are not installed on the build machine.

This is due to the logic used by the mplayer configure script, which
looks like this:

  for I in $(echo $extra_cflags | sed s/-I//g) /usr/include ; do
    if test -f "$I/X11/Xlib.h" ; then
      _x11_headers="yes"

So, in other words, it:

 1/ Parses the --extra-cflags option, and finds the -I options in there.

 2/ Looks in /usr/include

Since $(STAGING_DIR)/usr/include is in the compiler built-in search path
for headers, we currently don't explicitly pass it in --extra-cflags, so
mplayer only looks in /usr/include. If you have X11 headers there thanks
to being installed on your build machine, everything works fine (the
rest of the build logic really uses the headers and libraries of the
cross-compiler). But if you don't have X11 headers in /usr/include, the
configure scripts assumes X11 is not available.

Since fixing the hand-written configure script of mplayer, hosted in a
Subversion repository, is beyond sanity, we simply work around this
problem by passing the appropriate -I$(STAGING_DIR)/usr/include option
in --extra-cflags.

Before this patch, during the configure script:

Checking for X11 headers presence ... no (check if the dev(el) packages are installed)
Checking for X11 ... no (check if the dev(el) packages are installed)

And then, the mplayer binary:

 0x00000001 (NEEDED)                     Shared library: [librt.so.0]
 0x00000001 (NEEDED)                     Shared library: [libz.so.1]
 0x00000001 (NEEDED)                     Shared library: [libpthread.so.0]
 0x00000001 (NEEDED)                     Shared library: [libdl.so.0]
 0x00000001 (NEEDED)                     Shared library: [libm.so.0]
 0x00000001 (NEEDED)                     Shared library: [libc.so.0]

With this patch, during the configure script:

Checking for X11 headers presence ... yes
Checking for X11 ... yes

And then, the mplayer binary:

 0x00000001 (NEEDED)                     Shared library: [librt.so.0]
 0x00000001 (NEEDED)                     Shared library: [libz.so.1]
 0x00000001 (NEEDED)                     Shared library: [libpthread.so.0]
 0x00000001 (NEEDED)                     Shared library: [libdl.so.0]
 0x00000001 (NEEDED)                     Shared library: [libm.so.0]
 0x00000001 (NEEDED)                     Shared library: [libXext.so.6]
 0x00000001 (NEEDED)                     Shared library: [libX11.so.6]
 0x00000001 (NEEDED)                     Shared library: [libXinerama.so.1]
 0x00000001 (NEEDED)                     Shared library: [libXxf86vm.so.1]
 0x00000001 (NEEDED)                     Shared library: [libc.so.0]

Fixes bug #8206

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

gcc: remove BR2_GCC_ENABLE_TLS option

The current BR2_GCC_ENABLE_TLS can cause users to make incorrect
choices, and is not very useful. This options allows to decide whether
we pass --enable-tls or --disable-tls to gcc, to enable or disable
support for Thread Local Storage.

Its behavior is:

 - The option is default to "y" but only exists if we're using
   uClibc/NPTL or glibc.

 - When we're using uClibc, the option can be disabled.

So, in practice, this means that currently:

 - TLS support is always on for glibc

 - TLS support is on by default for uClibc/NPTL, but can be disabled in
   the configuration. This is in fact bad and causes the build failure
   reported in bug #7424 (this bug is still reproducible on master)

 - TLS support is always disabled for uClibc/no-thread and
   uClibc/linuxthreads.

 - TLS support is always disabled for musl. This does not cause any
   build failure, but musl can use TLS support, and therefore be more
   efficient. According to
   http://www.openwall.com/lists/musl/2012/10/04/1, "Note that if you've
   been building gcc with --disable-tls, __thread was already working
   but gets emulated (very poorly; it's slow and will abort() if it runs
   out of memory) through libgcc.".

So, this commit completely removes the BR2_GCC_ENABLE_TLS and instead
makes the right choice inside gcc.mk directly:

 - TLS support enabled for glibc, musl and uClibc/NPTL

 - TLS support in other cases, i.e uClibc/no-thread and
   uClibc/linuxthreads.

We have intentionally *not* added the option to
Config.in.legacy. Indeed, the new behavior is *exactly* the same as the
older behavior, with the exception of:

 - People can no longer disable TLS support in uClibc/NPTL, which was
   anyway causing a build failure and therefore was not used.

 - TLS support is now enabled on musl, but people using musl already had
   BR2_GCC_ENABLE_TLS not set, so they wouldn't get the legacy warning.

Fixes bug #7424.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

valgrind: let the valgrind configure script detect TLS availability

Back in 2005, in commit
a2c326396a43ecbc8d02c3d815d4010a7ba2e004 ("update valgrind to the latest
and greatest"), an explicit --disable-tls option was added. More
recently, in commit 31a3f4bd54e12b8d6de286ab8fb6d9651990e2f5 ("valgrind:
enable tls support") changed this to be conditional on
BR2_GCC_ENABLE_TLS.

However, the configure script of valgrind is perfectly capable of
detecting TLS support, even in a cross-compilation case: it tries to
compile a program that uses __thread and sees if it works.

Since we're about to modify how BR2_GCC_ENABLE_TLS is handled, we'd
better remove its usage from packages, and valgrind is the only package
using this config option.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

gupnp: bump to version 0.99.0

Signed-off-by: Fabrice Fontaine <fabrice.fontaine@orange.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

gssdp: bump to version 0.99.0

Signed-off-by: Fabrice Fontaine <fabrice.fontaine@orange.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/procps-ng: Bump to version 3.3.12

Signed-off-by: Jeroen Roovers <jer@airfi.aero>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

toolchain: Bump ARC tools to arc-2016.09-eng010

As described at:
4520524ba055706236db9f00dd79f1b2e2e87fde
this commit continues a series of updates of ARC tools.
This time we're updating tools to arc-2016.09-eng010.

This engenering build contains different fixes done to TLS and
PIE features. Appropriate custom patches are removed as they have
been added to eng010.

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

alljoyn-tcl: bump to version 16.04a

Signed-off-by: Fabrice Fontaine <fabrice.fontaine@orange.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

alljoyn: bump to version 16.04a

Signed-off-by: Fabrice Fontaine <fabrice.fontaine@orange.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

mosquitto: bump to version 1.4.10

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

libinput: bump to version 1.4.2

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

gvfs: bump to version 1.28.3

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Update for 2016.08-rc3

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

zynq: enable BR2_TARGET_UBOOT_BUILD_SYSTEM_KCONFIG

The recent U-Boot adopts Kconfig for its configuration system.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libaio: allow building on static-only configurations

This commit adds a simple patch to the libaio build system that allows
it to be used on static-only configurations. It is probably easier to
fix this than to propagate the !BR2_STATIC_LIBS dependency.

Fixes:

  http://autobuild.buildroot.net/results/8c86402bd14a0af1b82e6e07c06531a856dd37c6/

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Romain Naour <romain.naour@gmail.com>
Tested-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

ccache: bump to version 3.3

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

docs/website: remove reference to GMane

GMane is dead (at least its web interface), so it's no longer really
useful to reference it.

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

pango: bump to version 1.40.2

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

tftpd: disallow on nios with pre-6.x toolchains

Fixes:
http://autobuild.buildroot.org/results/349/349bc8668614b0f2c970ff33f25216e708ecdb46/
http://autobuild.buildroot.org/results/264/26486e88211d01ee49ad786e9a27a67b849ad1ab/

Older toolchains contain an issue with the linker script causing failures when linking tftpd:

tftpd.c:(.text.startup+0x5c): warning: Unable to reach (null) (at 0x00000000)
from the global pointer (at 0x00011a80) because the offset (-72320) is out of
the allowed range, -32678 to 32767.

This is a known issue which has been fixed from 6.x, so disallow on older
toolchains:

https://www.altera.com/support/support-resources/knowledge-base/solutions/rd02132012_291.html

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

package/openswan: not available on musl

After fixing the usual sys/cdefs.h and _P stuff the build still fails
because Openswan makes use of GLOB_ functions

/home/bernd/buildroot/br4/output/build/openswan-2.6.48/lib/libipsecconf/parser.l:200:29: error: ‘GLOB_BRACE’ undeclared (first use in this function)
      globresult = glob(try, GLOB_BRACE, NULL, &globbuf);
                             ^
/home/bernd/buildroot/br4/output/build/openswan-2.6.48/lib/libipsecconf/parser.l:200:29: note: each undeclared identifier is reported only once for each function it appears in
/home/bernd/buildroot/br4/output/build/openswan-2.6.48/lib/libipsecconf/parser.l:207:41: error: ‘GLOB_NOMAGIC’ undeclared (first use in this function)
       globresult = glob(try, GLOB_BRACE|GLOB_NOMAGIC, NULL, &globbuf);

https://github.com/xelerance/Openswan/blob/master/lib/libipsecconf/parser.l#L200
https://github.com/xelerance/Openswan/blob/master/lib/libipsecconf/parser.l#L207
[...]
                                         ^
which musl does not support:
http://www.openwall.com/lists/musl/2013/09/16/1

Fixes
http://autobuild.buildroot.net/results/70d/70da34c1528b51e92e9e60788f83733d70b33353/

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libiio: Disable IIOD on Blackfin

IIOD uses accept4, which is not yet available on Blackfin.

Fixes:
 http://autobuild.buildroot.net/results/b75/b7552c9325ffcc7f1cb6e114b82648f3280fded6/

[Peter: hide other toolchain comment on bfin]
Signed-off-by: Paul Cercueil <paul.cercueil@analog.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

tinydtls: fix issue on u_intXX_t being undefined when building with musl

Fixes:
  http://autobuild.buildroot.net/results/928be69f90476e6b04be3a1afd3b74112bcac0a0

As mentioned in sha2/README, by default, tinydtls uses u_intXX_t data types
for 8 bit, 32 bit, and 64 bit unsigned integer type definitions.  To use
uintXX_t data types as defined by recent ANSI C standards and as included in
the inttypes.h header file, SHA2_USE_INTTYPES_H has to be define at compile
time.

[Peter: reword/simplify]
Signed-off-by: Fabrice Fontaine <fabrice.fontaine@orange.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux-headers: bump 3.10.x series

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libserialport: Fix detection of OS when compiling for ucLinux

When compiling with a toolchain targetted at ucLinux, libserialport's
configure.ac would fail at detecting the target operating system.
As a result, the Linux-specific files were not compiled in that
particular case.

While this commit does not fix any autobuider failure for this package,
it fixes autobuilder failures for other packages, for instance:

http://autobuild.buildroot.net/results/69e/69e43bc171e554cf10f2ad526cebf5e0e524538a/

Signed-off-by: Paul Cercueil <paul.cercueil@analog.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/neon: properly enable expat

Currently, we forcibly tell neon's configure where to look for libexpat.
This means that neon does not search for it, and consequently forgets to
add -lexpat as its Libs.Private in its .pc file. As thus, a static link
against neon forgets to pull in -lexpat (it works for a dynamic link
because libneon has a NEEDED on libexpat).

The reason for specifying the explicit path in this case is lost to
history; neon's configure can correctly find libexpat nowadays.

Fixes:
    http://autobuild.buildroot.org/results/884/884746351fc47f4ad19fb3fda7c5e0ee85fcff15/
    http://autobuild.buildroot.org/results/0c5/0c584522c26640e29980150a87a5b5d8c02e5f8b/

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

package/libfreeimage: fix C++11 compliance with gcc-6

Like in cbe01f6, 38249d5, 97fe953, fix C++11 compliance regarding
narrowing types.

Fixes:
    http://autobuild.buildroot.org/results/081/0811531872f69f9febbdc482dfbdd7fb5c35d1c8/
    http://autobuild.buildroot.org/results/90c/90c9cb4ffe705b8c91f4fb602e33e2ba9bcdee77/
    and a few others...

[Peter: use signed char for src[], fix dos/unix new lines]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/sconeserver: needs shared libs

sconeserver wants to use dlopen(), unconditionally: it does not try to
detect it, and it can't work without it (the code is not conditional).

Fixes:
    http://autobuild.buildroot.org/results/e92/e929799366e6f574c7b08d8635b18623afd3e3e1/

[Peter: drop uClibc dependency, adjust commit message]
Reported-by: Matthew Weber <matt@thewebers.ws>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/sconeserver: unconditionally needs host-pkgconf

sconeserver uses pkg-config macros and uses autoreconf, so it
unconditionally needs host-pkgconf.

[Peter: drop host-pkgconf from sub options, update description]
Reported-by: Matthew Weber <matt@thewebers.ws>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Revert "Added local directory as source of kernel code"

This reverts commit 73da2ff6f718f2889e3c5024d899f8d58f502863.

The reason for adding support for a local location was to be able to do
development on the Linux kernel source tree on a local directory rather
than have to clone it for every build.

We already have a mechanism for that, it's called override-srcdir. It's
been available since September 2011, more than a year before this patch
was committed.

Otherwise, we're going to be adding support for local sources in other
packages. First was U-Boot as submitted by Adam. But what next? We can't
have such support for all packages, especially since override-srcdir
does the job.

Besides, using a local source tree makes the build non-reproducible, so
we don't really want to have this in a .config (or defconfig).

We only handle the boolean option in legacy, as there is nothing we can
do with the directory path.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Rafal Fabich <rafal.fabich@gmail.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

android-tools: disable on some architecture with old kernel headers

The android-tools code is somewhat ugly, and defines its own u64 typedef
becore including kernel headers. Unfortunately, there are specific cases
where that doesn't work properly.

The android-tools code defines u64 as "unsigned long long", which is now
correct in the kernel. However, it used to be a time where u64 was
defined as "unsigned long" on a few 64 bits architecture (at least
PowerPC64 and MIPS64). The kernel headers have introduced a
__SANE_USERSPACE_TYPES__ macro that userspace can define in order to get
the "sane" definition, i.e "unsigned long long" for u64.

Unfortunately, this __SANE_USERSPACE_TYPES__ mechanism only appeared in
3.10 on PowerPC64, and in 3.16 on MIPS64.

Since android-tools is not using the autotools, and there's no easy way
to test types with the C pre-processor, we simply add some more
Config.in dependencies. They are a bit convoluted, but that's what the
dependency really is.

In our autobuilders, this issue was only showing up with an old MIPS64
toolchain that uses 3.9 kernel headers.

Also, since the problem is limited to the "fastboot" tool, the
dependency is only added for fastboot. Both adb and adbd build fine with
this toolchain.

Fixes:

  http://autobuild.buildroot.net/results/ce45c995bd6abda6487ae3a11b4f45a7b9b3f8eb/

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

libamcodec: fix build with musl

This commit adds a patch to the libamcodec in order to fix the build
with musl, which was failing due to missing <stdint.h> includes.

Fixes:

  http://autobuild.buildroot.net/results/1f3b0d483603fb69765c15c7189ff975b72d09c3/

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

libiio: do not install systemd service file, no longer available

Since the bump to libiio 0.7, the systemd service file that used to be
in debian/iiod.service no longer exists. The entire debian/ directory
has been removed from the upstream project:

  https://github.com/analogdevicesinc/libiio/commit/5d49f58982b0c15545bc70724304fbc580e6523c

Due to this, the installation of this service file fails, and causes
build failures. To address this, we simply remove the installation of
the systemd service file.

Fixes:

  http://autobuild.buildroot.net/results/ce2fc81466abd9619aa104c96234d1455de3480d/

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

wine: Bump to version 1.8.4

Bumping to stable version 1.8.4

Signed-off-by: André Hentschel <nerv@dawncrow.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

webkitgtk: security bump to version 2.12.4

Fixes:
CVE-2016-4590 - mishandles about: URLs, which allows remote attackers to
bypass the Same Origin Policy via a crafted web site.

CVE-2016-4591 - mishandles the location variable, which allows remote
attackers to access the local filesystem via unspecified vectors.

CVE-2016-4622 - allows remote attackers to execute arbitrary code or
cause a denial of service (memory corruption) via a crafted web site, a
different vulnerability than CVE-2016-4589, CVE-2016-4623, and
CVE-2016-4624.

CVE-2016-4624 - allows remote attackers to execute arbitrary code or
cause a denial of service (memory corruption) via a crafted web site, a
different vulnerability than CVE-2016-4589, CVE-2016-4622, and
CVE-2016-4623.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/libxmlrpc: Fix build with gcc6

Fixes
http://autobuild.buildroot.net/results/c45/c45530f34bf2074bd204c485615124df928f246c/

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

bfin: change default to bf532, internal toolchain !bf6xx

The default Blackfin processor in Buildroot isn't supported by
gcc 6.1.0, so use bf532 as default. Disable any bf6xx processors
for internal toolchain users.

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

docs/manual: prepare-kconfig can be used as a dependency of documents

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Cc: Thomas De Schampheleire <patrickdepinguin@gmail.com>
Reviewed-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

doc/asciidoc: add possibility to define document dependencies

Currently, a document can not have dependencies, except for the purely
internal ones (like checking asciidoc version, and presence of dblatex).

For our own manual, this will come in handy when we introduce a
generated kconfig snippet, so we can actually make the manual depend on
that snippet being generated first.

For external documents, it can be used to depend on host-packages if
need be (e.g. a custom host packages that generates specific media files
included in the manual).

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Cc: Thomas De Schampheleire <patrickdepinguin@gmail.com>
Reviewed-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

core: introduce a generated kconfig snippet

Add the infrastructure for adding generated kconfig snippet in the
menuconfig.

For now, the kconfig snippet is generated empty, the recipe for filling
it in will be introduced in sub-sequent patches.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Thomas De Schampheleire <patrickdepinguin@gmail.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Reviewed-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

core: move rule to create basic directories

Some of those directories will be needed even during configuration, like
BUILD_DIR, where we'll store the generated kconfig snippet.

So, move the rule to create them outside the BR2_HAVE_DOT_CONFIG block.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Reviewed-by: Romain Naour <romain.naour@gmail.com>
[Thomas: fixup commit log, as noticed by Romain Naour.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

core: introduce an intermediate rule before the configurators

Currently, all configurators depend on generating the out-of-tree
Makefile wrapper.

In an upcoming patch, we'll need to also generate a kconfig fragment,
so it will have to kick in before we run the configurators.

Introduce a new intermediate "prepare-kconfig" rule, so we can
commonalise the dependencies of the configurators. Move the dependency
on the Makefile wrapper to that new intermediate rule.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Thomas De Schampheleire <patrickdepinguin@gmail.com>
Reviewed-by: Romain Naour <romain.naour@gmail.com>
[Thomas: mark prepare-kconfig as a phony target.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

core: commonalise the bundled and br2-external %_defconfig rules

The code for both cases is exactly the same, and only differs in the
location where defconfig files are looked for.

We use an intermediate macro to generate the corresponding rules,
because directly generating the rules is ugly and needs lots of escaping
and double-dollar-ing for the $(eval ...) and $(foreach ...) calls to
play nicely together.

Furthermore, that will be tremendously useful when we support multiple
br2-external trees.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Cc: Thomas De Schampheleire <patrickdepinguin@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Reviewed-by: Romain Naour <romain.naour@gmail.com>
[Thomas: move comment outside of the make target, so that it isn't
displayed on stdout when loading a defconfig.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

core: move pkg-utils.mk to support/

pkg-utils.mk contains various definitions that are used in the package
infrastructures and packages themselves.

However, those definitions can be useful in other parts of Buildroot,
and are already used in a few places that are not related to the package
infrastructure. Also, $(sep) will be needed early in the Makefile when
we eventually support multiple br2-external trees.

Since this file only contains definitions, we can include it anytime.

So, consider that file to no longer be specific to the package infras:
  - move it to support and rename it,
  - move a few similar definitions from the main Makefile to that file.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Reviewed-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

jack2: fix build issue with glibc

With glibc 2.16, we get following build error when building jack2:

  [193/247] cxx: tests/iodelay.cpp -> build/tests/iodelay.cpp.4.o
  ../tests/iodelay.cpp:171:43: error: 'UINT32_MAX' was not declared in this scope
  ../tests/iodelay.cpp:171:55: error: 'UINT32_MAX' was not declared in this scope
  ../tests/iodelay.cpp:172:44: error: 'UINT32_MAX' was not declared in this scope
  ../tests/iodelay.cpp:172:56: error: 'UINT32_MAX' was not declared in this scope

In glibc 2.17 or older version, Header <stdint.h> defines these macros
for C++ only if explicitly requested by defining __STDC_LIMIT_MACROS.

We can't use <cstdint> since it requires C++11 standard.

Fixes:

  http://autobuild.buildroot.net/results/369ce208ffea43dad75ba0a13469159b341e3bf5/

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

trousers: fix build with musl

The trousers code uses getpwent_r, which is not available in musl.
Detect the availability of getpwent_r in the trousers build system, and
use it conditionally.

This broke the build of tpm-tools because linking with libtspi.so
failed.

Fixes:

  http://autobuild.buildroot.net/results/830fc20c68a0653afa5567edffc2ededc4e45cc6

Runtime-tested by running tpm_version in a chroot and verifying that it
creates a "user.data" file with both the Buildroot-provided CodeSourcery
and Musl toolchains on x86_64.

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

package/protobuf: uses fork(), not available on no-MMU platforms

This was not noticed until now because:

 1/ The older Blackfin toolchain doesn't have libatomic, so it didn't
    provide the atomic operations that protobuf needs, so protobuf was
    never built.

 2/ The ARM Cortex-M toolchain is static-only, and protobuf requires
    dynamic library support.

So it's only with the new Blackfin toolchain, which is based on gcc
6.x (and therefore provides libatomic) and is FDPIC-based (and therefore
has dynamic library support) that this problem appeared.

Since protobuf already has a BR2_PACKAGE_PROTOBUF_ARCH_SUPPORTS option,
we use it to add the BR2_USE_MMU dependency (which is architecture
related), which avoids the need to propagate the dependency.

Fixes:

  http://autobuild.buildroot.net/results/2c1/2c151e84d7854a810465dc16869023e0ada2d586/

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
[Thomas:
 - move the BR2_USE_MMU dependency under
   BR2_PACKAGE_PROTOBUF_ARCH_SUPPORTS and remove the propagation to
   reverse dependencies of protobuf, since they already depend on
   BR2_PACKAGE_PROTOBUF_ARCH_SUPPORTS.
 - improve commit log.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/lshw: add patches for musl build

Fixing one issue exposed a second one.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

vlc: fix compile error with uClibc-ng Linuxthreads

The fallback code path didn't work correctly, so fix it.

Fixes:
  http://autobuild.buildroot.net/results/3122287ddea1e316a64ccf0d0dc9415bfefebb49/

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

package/kmsxx: add patch to fix LTO support

The LTO support in the kmsxx package uses the host gcc-ar and gcc-ranlib
instead of the ones from the cross-toolchain. Add a patch that tries to
find the right one based on CMAKE_C_COMPILER.

Fixes:
http://autobuild.buildroot.net/results/16a/16a38a4277dd1152a5955d62cb92f85447791ef3

Possibly also fixes:
http://autobuild.buildroot.net/results/f3c/f3c48da3a9706cd366c0e0a96c3cd0ff959f2a78
(it fails later, possibly because an incompatible host ar)

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Samuel Martin <s.martin49@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/libraw: fix gcc-6 build

Grab the patch from Debian.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

norm: add patch to fix musl build

Fixes:

  http://autobuild.buildroot.net/results/9c7130d5b5aad218c5c576761d495a3f403fdc5b/

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

pixman: properly enable/disable ARM optimizations

The pixman ARM optimized code assumes that ARM instructions are
available. Unfortunately, the configure.ac checks do not detect that the
platform is Thumb-only for Cortex-M builds, so it enables the ARM
optimizations, leading to failures like:

error: /home/test/autobuild/run/instance-3/output/host/usr/arm-buildroot-uclinux-uclibcgnueabi/sysroot/usr/lib/libpixman-1.a(pixman-arm-simd-asm-scaled.o): Conflicting CPU architectures 13/1

When building programs linked with pixman on Thumb-only
architectures. This is due to the fact that some object files in
libpixman-1.a are built for the ARM instruction set.

To resolve this, we give better hints to the pixman configure script
about which ARM optimizations to use: the ARM SIMD optimizations need at
least a CPU that supports ARM instructions, and obviously the ARM NEON
optimizations need NEON support.

Fixes:

  http://autobuild.buildroot.net/results/54bee2ce382fcd067965d30f758f9d15514478d9/

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
[Thomas: add a comment above the --enable-arm-simd option, as suggested
by Arnout.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>