qt: make installation of translation files optional

Commit 93917b6980f7f2b51302e1a3fa451b07cf7d674e introduced the
installation of the binary .qm translation files, unconditionally.
This patch introduces an option to disable this behavior, saving
almost 8MB of space.

[Thomas: rename option to BR2_PACKAGE_QT_TRANSLATION_FILES instead of
BR2_PACKAGE_QT_INSTALL_TRANSLATION_FILES, and move its definition
around the installation of examples/demos rather than in the middle of
the options for the different modules.]

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

package/ffmpeg: Bump to version 2.3.1

- remove obsolete option --disable-vis
- add patch to fix compile error with uClibc
- add patch to fix compile error in freerdp

Compile-tested against all packages depending on ffmpeg using this defconfig:

BR2_TOOLCHAIN_EXTERNAL=y
BR2_PACKAGE_GSTREAMER=y
BR2_PACKAGE_GST_FFMPEG=y
BR2_PACKAGE_MPD=y
BR2_PACKAGE_MPD_FFMPEG=y
BR2_PACKAGE_XORG7=y
BR2_PACKAGE_FREERDP=y
BR2_PACKAGE_OPENCV=y
BR2_PACKAGE_OPENCV_WITH_FFMPEG=y
BR2_PACKAGE_MEDIASTREAMER=y
BR2_PACKAGE_MINIDLNA=y

VLC also compiled when the build of vlc-static is disabled in Makefiles,
otherwise this known error occurs:
http://autobuild.buildroot.net/results/022/022252176c051ff3365b2cb820f32aafc802803a/

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

configs: bump kernel used for the RPi

The rpi-3.12.y branch has now been merged with the 3.12.25 upstream.

Bump kernel headers used for the toolchain at the same time.

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

package/rpi-firmware: Bump version to kernel 3.12.25

This version bump, amongst other things, introduces
OMX_ImageFilterDeInterlaceFast, which is needed by
the upcoming xbmc/kodi version bump.

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

package/rpi-userland: Bump version

This version bump introduces OMX_ImageFilterDeInterlaceFast,
which is needed by the upcoming xbmc/kodi version bump.

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

package/vlc: Bump version to 2.1.5

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

libxcb: Bump version to 1.11

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

xcb-proto: Bump version to 1.11

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

xapp_xrandr: Bump version to 1.4.3

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

musl: bump to version 1.1.4

The gcc 4.9.x breakage is now handled by musl itself.

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

samba4: security update to 4.1.11

Fixes CVE-2014-3560 (Remote code execution in nmbd).

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

linux: bump default to version 3.15.8

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

linux-headers: bump 3.{4, 10, 12, 14, 15}.x series

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

bluez5_utils: new package

This patch adds Bluez 5.x. Its API is not backwards compatible with
BlueZ 4 so it is packaged separately from BlueZ 4.

BlueZ utils will use systemd and/or udev if enabled. It contains a
hook to install the GATT support tool, which is always built but not
installed by default.

[Thomas:
 - Add dependency on 3.4 kernel headers
 - Add options for OBEX support and client support, to make the
   libical and readline dependencies optional.
 - Rewrap Config.in help text.
 - Adapt indentation in .mk file.
 - Bump to version 5.21.
 - Use BR2_PACKAGE_HAS_UDEV instead of BR2_PACKAGE_UDEV.]

Signed-off-by: Marcin Bis <marcin@bis.org.pl>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libdrm/intel: Explain dependency on xorg in Kconfig only for BR2_i386/BR2_x86_64

Suggested by Yann: http://article.gmane.org/gmane.comp.lib.uclibc.buildroot/82731

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

qemu/x86: update defconfig to Linux 3.15.x

Update the default configuration to use Linux 3.15.x.
Enable ext4, so that ext2/ext3/ext4 can be used.
Tested with Qemu 2.0.0.

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

package/patchelf: new host package

In some situations, users may want to tweak the dynamic section of the
binaries (for example to add/set the RPATH to $ORIGIN/../lib).

Because it is not trivial to do it properly from the Buildroot
infrastructure, allow those users to use patchelf (e.g. from a
post-build script) to tweak binaries.

patchelf is able to:
  - modify an existing DT_RUNPATH tags
  - add a DT_RUNPATH tag if not already present
  - do the above to the DT_RPATH tag, too
  - set the path to the interpreter
  - remove DT_NEEDED tags
  - query a binary for the DT_RUNPATH/DT_RPATH tag, or for the
    interpreter path

Does not really fix #7172, but this is an appropriate workaround.

[Thomas: change license to GPLv3+, as mentionned in the project's
README file.]

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Mike Zick <minimod@morethan.org>
Reviewed-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

toolchain: linker options with a $ sign are not supported

As reported in bug #7172 [0], setting BR2_TARGET_LDFLAGS to a value
containing a $ sign can lead to unexpected results.

This is because it is very hard to know when the $ sign gets evaluated:
  - in the Buildroot-level make
  - in the shell called by the Buildroot-level make
  - in the package's own build-system, either at configure time, in the
    Makefile, in a shell in the Makefile...

So, it is very difficult to know how much escaping that would need.

A proposal is to use a shell variable to pass such values unmolested.
But it is not that simple either, since it still contains a $ sign, and
there is not much certainty as to when it would be evaluated.

Instead, just document this limitation, both in the help text for
BR2_TARGET_LDFLAGS, and in the known-issues section in the manual.

Does not really fix #7172, but at least the limitation is documented.

[0] https://bugs.buildroot.org/show_bug.cgi?id=7172

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Mike Zick <minimod@morethan.org>
Reviewed-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

bandwidthd: new package

[Thomas: minor reformatting like removing trailing whitespaces,
wrapping to a shorter length Config.in help texts, etc.]

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

package/dnsmasq: Fix link error with gettext

-lintl needs to be added at the end of the link command, Makefile
has variable LIBS for this purpose, so use it instead of LDFLAGS.

Fixes
http://autobuild.buildroot.net/results/276/276581f6dbbe330799c3a7eaa26b453e38a22907/

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

package: remove the trailing slash sign from <PKG>_SITE variable

Since the trailing slash is stripped from $($(PKG)_SITE) by pkg-generic.mk:

$(call DOWNLOAD,$($(PKG)_SITE:/=)/$($(PKG)_SOURCE))

so it is redundant.
This patch removes it from $(PKG)_SITE variable for BR consistency.

Signed-off-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

manual/adding-packages-perl.txt: remove the trailing slash sign from description of PERL_SITE variable

Since the trailing slash is stripped from $($(PKG)_SITE) by pkg-generic.mk:

$(call DOWNLOAD,$($(PKG)_SITE:/=)/$($(PKG)_SOURCE))

so it is redundant.

Signed-off-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libevdev: use tarball instead of git repository

Tarballs are now provided for libevdev, so use that instead of the Git
repository, as it allows to avoid autoreconf for the package.

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

libevdev: remove host python dependency

In commit 79c31e3297da7f0aec4d4675c154736e4cffe3ed ("package/libevdev:
add dependency on host-python"), Yann added a dependency of libevdev
on host-python, because the Python script in evdev was using argparse
which is only available since Python 2.6. At the time, this was
causing failures on machines that were using Python versions older
than 2.6.

However, since Yann's patch, the libevdev version has been bumped. And
one of the upstream commit,
http://cgit.freedesktop.org/libevdev/commit/libevdev/?id=cea8f0b8cc86332b0643acd62f24f9fef2ecc153,
was precisely made to remove the argparse dependency to avoid the
need to have a >= 2.6 Python version.

Thanks to this, we can avoid the host-python dependency and rely on
the system-installed Python version instead.

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

libevdev: bump to 1.2.2

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

google-breakpad: integration into Makefile and Config.in

Add a new config option that allows to specify a list of files
to dump symbols from, and generate a database of symbols, in
stagins/usr/share/google-breakpad/.

Signed-off-by: Pascal Huerst <pascal.huerst@gmail.com>
[yann.morin.1998@free.fr: add dependency on the host variant; fix
 script; add commit log]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
[Thomas: remove leftover debugging, rewrap Config.in help text,
propagate dependencies correctly.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libarchive: bump to version 3.1.2

Switch to proper upstream source.
And account for many auto dependencies that were missing.
Also fixes link issues:
http://autobuild.buildroot.net/results/774/7747a043123e958952a16e0c0db823296c4e0594/

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

linux-pam: unavailable for static builds

Fixes:
http://autobuild.buildroot.net/results/3bb/3bb8d4662995a11e8e6cc5fdaaba9e25d4c0e0ec/

[Thomas: add more details about the non availability for static builds
as a comment in the Config.in file.]

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

libqmi: bump to 1.10.2

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

libndp: bump to 1.4

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

Remove unsupported libcue from package mpd

Support for libcue has been removed for mpd since version 0.17.

Signed-off-by: Jörg Krause <jkrause@posteo.de>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

udev: remove left-over S10udev script from udev virtual package

The S10udev script is not installed or referenced, and eudev has its own
S10udev.  So we can simply remove it.

Signed-off-by: Danomi Manchego <danomimanchego123@gmail.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ptpd2: fix static linking

Do the same as in commit b64a06d7b70bd7b160826bf9b97e60e18171913d
to avoid missing libraries needed by libpcap. Fixes:
http://autobuild.buildroot.net/results/1a1/1a1dadfa2b22074d401b013606f0cede66efe826/

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

imagemagick: bump to version 6.8.9-6

Signed-off-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

samba4: bump to version 4.1.10

Lots of bugfixes, enhancements to provisioning and printing support via
cups.

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

Revert "packages: autoreconf non-vanilla libtool packages"

Now that we've got a cleaner/fuzzier libtool 1.5 static patch we can
discard the temporary workaround.

This reverts commit e573f5d3267e5916c580e499ecd127c60107b8e1.

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

support/libtool: improve version 1.5 static patching

Alter the libtool 1.5.x support patch to accomodate for wildly different
versions of ltmain.sh
Just make it alter incoming args from -static to -all-static which seems
to apply to all the different variants out there since argument parsing
is unlikely to change much.

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

packages: autoreconf non-vanilla libtool packages

Some packages no longer apply the libtool patch since commit
97703978ac870ce2b14ad144f8e082de82aa2c64 because they use a non-vanilla
version of libtool 1.5.x

Fixes many failures like:
http://autobuild.buildroot.net/results/34e/34e4898e2bdc08e5d34e16e556384b3086b76467/
http://autobuild.buildroot.net/results/ecf/ecf4e7d6812f972d05c95203fb665235856c0817/
http://autobuild.buildroot.net/results/5d9/5d9a05fb70e8a65f2399c4f38375aeafb9686ea4/

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

bind: fix static openssl linking

Apparently LIBS (environment) isn't pushed to LIBS (autoconf) for the
tests, hence when static linking openssl it misses libz and assumes some
basic digests required for ecdsa support are missing, which is false. Fixes:
http://autobuild.buildroot.net/results/204/20411901d1bc6811e1ef8fa39457257d6bf1a146/

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

polarssl: disable microblaze inline assembly

Fixes:
http://autobuild.buildroot.net/results/4d5/4d54958ded61a0d929d992e4ca0bb31c996953cb/

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

support/libtool: make -static behave like -all-static

After switching TARGET_LDFLAGS from --static to -static, one issue
appears: from the point of view of libtool, -static only means to link
statically against the 'uninstalled libtool libraries' (i.e the
libraries that libtool has built in the current package), but
otherwise links dynamically with the other libraries. To really get a
completely static build, you need to pass -all-static to
libtool. Unfortunately, -all-static is only a valid option for
libtool, not as a general LDFLAGS, so we cannot to TARGET_LDFLAGS =
-all-static without breaking virtually all packages.

As pointed out 10 years ago on the libtool mailing list, the current
naming of the options is very confusing and the source of issues, and
there was a proposal to change -static to have the behavior of
-all-static, and instead introduce a separate -lt-static to have the
current behavior of -static. But that never got merged, because it was
breaking the current behavior. See:

  http://lists.gnu.org/archive/html/libtool/2004-11/msg00017.html

However, in Buildroot, when we pass -static, we really mean it, and we
want a completely static build. Therefore, this patch adapts our
ltmain.sh patches so that they alter the behavior of -static to make
it work like -all-static. The changes are small and quite easy to
understand, and have been tested to work fine with a small selection
of packages.

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

package infra: add -static to C/CXXFLAGS

Generic infra packages might not use LDFLAGS at all so add -static for
static builds to CFLAGS and CXXFLAGS too.

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

package infra: revert 0a4bd19f

Revert commit 0a4bd19f4a136930c611027942ce43124a81c5cb
Using --static is not documented and actually breaks static linking
according to my tests.
Fixes:
http://autobuild.buildroot.net/results/327/327c18db4e5d0ddc2c72a4684e103c19a1405e50/
...for external toolchains

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

gcc-final: disable shared build for static

Disable shared build for host-gcc-final when building for static targets.
We really want static or shared, there's no such thing as "preferring static"
since we can't choose with any degree of granularity for which packages.
And it confuses linking scripts having both available at the same time. Fixes:
http://autobuild.buildroot.net/results/c54/c54bdf88eff6d60c7001cb0e2cb6792cc75178db/

[Thomas: slightly amend the commit to factorize the installation of
static libraries.]

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

libiconv: disable building the preloadable library

We were already removing the preloadable iconv library (that can be
used through LD_PRELOAD to override the C library iconv
implementation) from staging/target, but it was still built. And this
causes issues in static only scenarios, so this patch changes that to
not even build/install the preloadable library.

[Thomas: changed Gustavo's patch to take into the fact that we never
need the preloadable library.]

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

binutils, gdb: support unified binutils-gdb git repository

If Binutils and/or GDB are fetched from the unified binutils-gdb
repository, then the tarball will contain both Binutils and GDB
sources, unlike the "normal" tarballs that contain only the titular
package. To keep packages separated in Buildroot we need to disable
undesired components when configuring.

Binutils and GDB migrated to a common Git repository in the October
2013 [1]. Previous Git repositories were incomplete copies of CVS
repository which copied only the relevant files (no binutils files in
GDB, and vice versa). In the new binutils-gdb repository there is no
such separation and a result all files exist in directory after
checkout. So if "configure" and "make" are used without explicit
targets, all projects will be built: binutils, ld, gas, bfd, opcodes,
gdb, etc. In case of Buildroot this would mean that selecting Binutils
only, still will build both Binutils and GDB.  And if GDB is selected
as well, then both packages will be built two times, and Binutils from
GDB directory will overwrite initial build of Binutils (or vice versa
if Binutils will be built after the GDB). This is a serious problem,
because binutils and GDB use separate branches in this common
repository. In case of Buildroot this means that separate Git commits
(or tags) should be used when downloading source from Git.

This affects only Git repositories, because GNU release tarballs still
contain only relevant packages.

This change is backward compatible, because if "normal" tarball is
used (without extra directories), than --disable-* configure options
are just ignored by configure.

[1] https://sourceware.org/ml/gdb/2013-10/msg00071.html

[Thomas: use variables to factorize options, and add comments in the
relevant .mk files to explain what's going on.]

Signed-off-by: Anton Kolesov <Anton.Kolesov@synopsys.com>
Cc: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/libdrm: Bump version to 2.4.56

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

mpd: bump to version 0.18.12

Small bugfixes and improvements, and patches upstreamed.

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

openssh: disable PIE when building for ARC

Fixes build failure reported here:
http://autobuild.buildroot.net/results/262/26218e028f3d2c77c5192b45154627f08384b688/

uClibc toolchain for ARC doesn't support PIE

Attempt to build anything with "-pie" option lead to linker failure:

arc-buildroot-linux-uclibc-gcc -pie test.c
ld: ../4.8-r3/bin/../arc-buildroot-linux-uclibc/sysroot/usr/lib/crt1.o: warning: unresolvable relocation against symbol `__uClibc_main' from .text section
ld: ../4.8-r3/bin/../lib/gcc/arc-buildroot-linux-uclibc/4.8.0/crtbegin.o: warning: unresolvable relocation against symbol `__deregister_frame_info@@GCC_3.0' from .text section
ld: ../4.8-r3/bin/../lib/gcc/arc-buildroot-linux-uclibc/4.8.0/crtbegin.o: warning: unresolvable relocation against symbol `__deregister_frame_info@@GCC_3.0' from .text section
ld: ../4.8-r3/bin/../lib/gcc/arc-buildroot-linux-uclibc/4.8.0/crtbegin.o: warning: unresolvable relocation against symbol `__register_frame_info@@GCC_3.0' from .text section
ld: ../4.8-r3/bin/../lib/gcc/arc-buildroot-linux-uclibc/4.8.0/crtbegin.o: warning: unresolvable relocation against symbol `__register_frame_info@@GCC_3.0' from .text section

In its turn this behavior confuses configure script of openssh so some options
get set improperly. In particular "strnvis" gets determined as existing which
causes failure during compilation:

log.c:67:25: error: 'VIS_SAFE' undeclared (first use in this function)
 #define LOG_STDERR_VIS (VIS_SAFE|VIS_OCTAL)

With disabled PIE ("--without-pie") openssh gets built without issues.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Gustavo Zacarias <gustavo@zacarias.com.ar>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Anton Kolesov <akolesov@synopsys.com>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

flex: delete broken flex++ symlink from target

When flex is built for the target without installing the
flex binary, a flex++ symlink installed by flex's Makefile
points to the missing flex executable.  This mod adds
a post target install hook to remove the broken symlink.

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

linux: remove support of linux26-* targets

The linux-* mirror targets of linux26-* have been added a very long time ago
(2010) and linux 2.6 is now considered 'old' anyway. It no longer makes
sense to support these linux26-* targets, so this patch removes them.

This is a simplification introduced in preparation of the kconfig-package
infrastructure.

Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

uclibc: use $(MAKE) iso $(MAKE1) for menuconfig target

There is no real reason to run uclibc-menuconfig in non-parallel mode,
even though one can neither expect performance benefits from a parallel
menuconfig.
Nevertheless, $(MAKE) is the default, so this patch removes the unnecessary
non-default $(MAKE1) usage for uclibc-menuconfig.

This is a simplification introduced in preparation of the kconfig-package
infrastructure.

Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

uclibc: update-config: preserve freshly configured settings

In the sequence:

make uclibc-menuconfig
make uclibc-update-config

the freshly configured settings from the menuconfig are lost during the
update-config step. This is because update-config depends on the configure
step, which starts by copying the config file to the build directory.

Instead, stop depending on the configure step from update-config, and
introduce a new stamp file .stamp_config_fixup_done, which applies any
fixups on the .config file.
An alternative solution would be to add a call to UCLIBC_FIXUP_DOT_CONFIG to
the relevant targets instead of depending on a new stamp file. The advantage
of the stamp file, though, is that we avoid redoing the fixup unnecessarily.
Moreover, in the light of the plan to extract the kconfig-specific bits into
a separate kconfig-package infrastructure, the stamp file rules are more
easily moved into such an infrastructure, while the alternative solution
requires the package .mk file to explicitly call the FIXUP rules which may
more easily be forgotten.

No longer depending on the configure step has the added bonus that
'uclibc-update-config' no longer needs the toolchain to be available, which
makes:
    make clean uclibc-menuconfig uclibc-update-config
much faster and user-friendly.

Additionally, make sure that 'make clean uclibc-update-config' works
properly, by depending on .stamp_config_fixup_done so that the config file
is present and fixed.

Fixes bug #7154 https://bugs.busybox.net/show_bug.cgi?id=7154

Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

uclibc: menuconfig: take into account initial settings from config file

When executing the sequence 'make clean uclibc-menuconfig', the configured
config file is not taken into account and one starts from the default
settings.

This patch adds an explicit target for the config file and lets the
configure and menuconfig steps depend on it, fixing the problem.

Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

uclibc: rename SETUP_DOT_CONFIG to FIXUP_DOT_CONFIG

In a subsequent patch, the behavior of UCLIBC_SETUP_DOT_CONFIG will change
so that it only applies fixups to the (already copied) configuration file.
This patch renames this function to UCLIBC_FIXUP_DOT_CONFIG to better match
the future behavior.

Suggested-by: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

uclibc: build tests after uClibc libs are installed to staging location

If libuClibc is not installed to staging on attempt to build tests following
errors happen:

============
  TEST_LINK termios/ termios
.../host/usr/arc-buildroot-linux-uclibc/sysroot/usr/lib/crt1.o: In function `__start':
.../build/uclibc/libc/sysdeps/linux/arc/crt1.S:53: undefined reference to `__uClibc_main'
termios.o: In function `main':
.../build/uclibc/test/termios/termios.c:13: undefined reference to `printf'
.../build/uclibc/test/termios/termios.c:14: undefined reference to `printf'
.../build/uclibc/test/termios/termios.c:16: undefined reference to `stdout'
.../build/uclibc/test/termios/termios.c:16: undefined reference to `fileno'
.../build/uclibc/test/termios/termios.c:16: undefined reference to `ioctl'
.../build/uclibc/test/termios/termios.c:19: undefined reference to `perror'
.../build/uclibc/test/termios/termios.c:21: undefined reference to `printf'
.../host/usr/lib/gcc/arc-buildroot-linux-uclibc/4.8.0/../../../../arc-buildroot-linux-uclibc/lib/libgcc_s.so.1: undefined reference to `memcpy'
.../host/usr/lib/gcc/arc-buildroot-linux-uclibc/4.8.0/../../../../arc-buildroot-linux-uclibc/lib/libgcc_s.so.1: undefined reference to `malloc'
.../host/usr/lib/gcc/arc-buildroot-linux-uclibc/4.8.0/../../../../arc-buildroot-linux-uclibc/lib/libgcc_s.so.1: undefined reference to `abort'
.../host/usr/lib/gcc/arc-buildroot-linux-uclibc/4.8.0/../../../../arc-buildroot-linux-uclibc/lib/libgcc_s.so.1: undefined reference to `calloc'
.../host/usr/lib/gcc/arc-buildroot-linux-uclibc/4.8.0/../../../../arc-buildroot-linux-uclibc/lib/libgcc_s.so.1: undefined reference to `dl_iterate_phdr'
.../host/usr/lib/gcc/arc-buildroot-linux-uclibc/4.8.0/../../../../arc-buildroot-linux-uclibc/lib/libgcc_s.so.1: undefined reference to `realloc'
.../host/usr/lib/gcc/arc-buildroot-linux-uclibc/4.8.0/../../../../arc-buildroot-linux-uclibc/lib/libgcc_s.so.1: undefined reference to `memset'
.../host/usr/lib/gcc/arc-buildroot-linux-uclibc/4.8.0/../../../../arc-buildroot-linux-uclibc/lib/libgcc_s.so.1: undefined reference to `strlen'
.../host/usr/lib/gcc/arc-buildroot-linux-uclibc/4.8.0/../../../../arc-buildroot-linux-uclibc/lib/libgcc_s.so.1: undefined reference to `free'
collect2: error: ld returned 1 exit status
make[3]: *** [termios] Error 1
============

And the simplest solution is to build tests right before installation in
UCLIBC_INSTALL_TARGET_CMDS which follows UCLIBC_INSTALL_STAGING_CMDS so
all required libs are in place.

Interesting note - if one enables uClibc tests after initial full buildroot
build mentioned errors don't happen. This is because uClibc libc libs are
already installed to staging. That's why I didn't noticed this problem before.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Anton Kolesov <akolesov@synopsys.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Gustavo Zacarias <gustavo@zacarias.com.ar>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

xdriver_xf86-video-intel: Bump version to 2.99.914

- enable sna module by default
- optionally enable uxa module if DRI2 support is enabled
- add support for DRI3
- dependencies for DRI1/2/3 modules are provided by xserver_xorg-server,
  therefore remove mesa3d dependency
- clean-up ums/kms options, configure picks sane defaults
- replace dependencies xproto_randrproto & xproto_renderproto with xlib_libXrandr

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

xserver_xorg-server: Add support for DRI3

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

popt: now needs gettextize

Because of df9244ff3738face50605eacf4d4f15d963de915 we now need to
gettextize popt. Fixes:
http://autobuild.buildroot.net/results/6e2/6e27c98407f114df07eb5cd16a843420fdb72bf4/

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

python: new module cffi

Foreign Function Interface for Python calling C code.

[Thomas: add dependency on host-pkgconf and libffi, enable on Python 3
since the module builds fine and loads fine with Python 3 as well.]

Signed-off-by: oli vogt <oli.vogt.pub01@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

httping: fix math library linking

Add the math library directive '-lm' to linker options at the end.
The order of the math library directive '-lm' matters.

Fixes:

  http://autobuild.buildroot.net/results/843/84382290696e72c23995f5ed020ad5c157817012/

Signed-off-by: Yuvaraj Patil <yuvaraj.patil@wipro.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

docs/manual: add sob line for patch series

The -s is missing in the documentation when teaching how to use 'git
format-patch' to automatically add the sob line.

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

luarocks: refactor with TARGET_FINALIZE_HOOKS

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

perl: refactor with TARGET_FINALIZE_HOOKS

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

uclibc: fix networking support on sparc

While updating the default config for Sparc, I noticed that networking
is broken. The reason is a uClibc backport patch from uClibc
master. This commit adds a fix on top of it, which was submitted to
upstream uClibc.

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

qemu/sparc-ss10: update defconfig to Linux 3.15.x

Update defconfig to 3.15, tested with Qemu 2.0.0.
Removed hint about framebuffer, as -display none is used.

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

linux: bump default to version 3.15.7

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

linux-headers: bump 3.{3, 10, 14, 15}.x series

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

popt: solve the glob_pattern_p() problem

The current popt build system tests the existence of <glob.h>, and
then assumes that if __GLIBC__ is defined, then glob_pattern_p() must
be available. Unfortunately, that's not true with uClibc: <glob.h> may
be installed, but not necessarily the GNU glob extensions... and
uClibc defines __GLIBC__. This is causing build issues with certain
uClibc toolchains that do not have GNU glob extensions enabled.

To fix this, we introduce a patch called popt-03-glob-detection.patch
which adds a new AC_CHECK_FUNCS() test for glob_pattern_p() and uses
the result to know if this function is available, instead of testing
__GLIBC__.

In order for this patch to work, the popt package must be
autoreconfigured, so another patch, popt-02-fix-autoreconf.patch, is
needed to make the package autoreconfigure properly. And also
POPT_AUTORECONF = YES is added to popt.mk.

Finally, this change avoids the need for the Blackfin toolchain
specific testing, which we forgot to update when introducing the
2014R1 version of the Blackfin toolchain. With this new solution,
there will be no need to update the popt.mk file when new Blackfin
toolchains are added.

Fixes:

  http://autobuild.buildroot.org/results/e09/e09e24fec240382a3197fef3e98eb9a22f76420a/

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Yuvaraj Patil <yuvaraj.patil@wipro.com>
Cc: Sonic Zhang <sonic.zhang@analog.com>

popt: rename patch to use the correct naming convention

This patch renames the popt patch popt-1.14_no-wchar.patch to
popt-01-no-wchar.patch in order to follow the naming convention used
in all packages.

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

PPC: fix ltrace build on PPC32

Add a patch to fix the build on PPC32 platform. That patch has been
submitted upstream.

Fixes:

  http://autobuild.buildroot.org/results/446/4460aac38225289ef58a8693da9ae520ae07a226/

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

icu: fix build after custom data library patch

Commit abded6bfead49d61c24a45d811a83ca082adf677 ("icu: add an option
to add a custom data library file") broke the build because of an
improper test on a non-qstripped variable. This commit fixes that.

Fixes:

  http://autobuild.buildroot.org/results/02a/02a0800b01a4bf8734d601f79dc12663fe8f7542/
  and many other similar occurences of the same issue.

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

python-flup: re-add MIT license

The MIT license was in the original submission, but I had removed it
since I couldn't find where it was used. It is indeed used in
flup/server/paste_factory.py as noted by the original
submitter. Therefore, this commit adds the MIT license as one of the
licenses of python-flup.

Reported-by: Oli Vogt <oli.vogt.pub01@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

flickcurl: adjust openssl dependency

flickcurl does not specifically need openssl: it needs libcurl to have
crypto support to access https:// URLs. So, instead of depending on
openssl in FLICKCURL_DEPENDENCIES, this commit simply leaves it to the
Config.in file to select either OpenSSL, gnutls or libnss to make sure
libcurl will use one of these libraries.

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

flickcurl: remove ca-certificates as a build dependency

ca-certificates is a runtime dependency only, so selecting it in the
Config.in file is sufficient, adding it to FLICKCURL_DEPENDENCIES is
not useful.

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

flickcurl: new package

We need two patches from upstream git as flickr now requires https access.

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
[yann.morin.1998@free.fr: fix the kconfig recursive loop when selecting
 a SSL implementation]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/lftp: inverse the openssl/gnutls selection

openssl is more widely used than gnutls. Other packages prefer openssl
over gnutls and would like to do the select in the opposite direction.

So switch lftp to use openssl by default, and only revert to using
gnutls if explicitly selected by the user.

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

icu: add an option to add a custom data library file

[Thomas:
 - use one single BR2_PACKAGE_ICU_CUSTOM_DATA_PATH string option
   instead of a boolean option + a string option, which always causes
   a lot of issues with random configurations that enable the boolean
   option, but do not provide a valid value for the string option.
 - enclose the definition of ICU_COPY_CUSTOM_DATA into the condition.]

Signed-off-by: Johan Derycke <johan.derycke@barco.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

linux: support multiple custom DTS files

[Thomas: fix minor typo in help text.]

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

perl: handle BR2_PREFER_STATIC_LIB

[Thomas: fix commit title, use one line for both CONF_OPT options.]

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

xapp_xconsole: Needs MMU

Fixes
http://autobuild.buildroot.net/results/be8/be89847832a2588dbda8dd921edb09b2af130e03/

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

postgresql: enable staging installation

Certain packages need the headers/libraries from Postgresql to build
properly.

[Thomas: slightly reword the commit message.]

Signed-off-by: Nathaniel Roach <nroach44@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ltris: needs MMU

This package needs MMU. Hence added dependency on BR2_USE_MMU

Fixes:
http://autobuild.buildroot.net/results/5cf/5cf2262460a72ce51b479a42fca61cea5f6e23fd//

Signed-off-by: Yuvaraj Patil <yuvaraj.patil@wipro.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

perl-module-build: bump to version 0.4206

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

perl-net-ssleay: bump to version 1.65

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

perl-mojolicious: bump to version 5.21

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

toolchain: add option to copy the gconv libraries

The gconv libraries are used to translate between different character sets
('charsets', even 'csets' sometimes). Some packages need them to present
text to the user (eg. XBMC Gotham).

In (e)glibc they are implemented by the internal implemenation of iconv,
called gconv, and are provided as dlopen-able libraries.

Note that some gconv modules need extra libraries (shared by more than
one gconv module), so we must, when adding a subset of modules, scan the
installed modules in search of the missing libraries.

[Thomas: add general explanation in expunge-gconv-modules and fix
coding style.]

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Cc: Eric Limpens <limpens@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

qt5: Need pcre with UTF support

Enabling PCRE16 is not enough for Qt5. PCREUTF is also necessary, else
Qt5 may complain with this kind of (non fatal) message:
  QRegularExpressionPrivate::doMatch(): called on an invalid QRegularExpression object

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

system: move tz setup outside of default skeleton clause

Allow time zone setup and installation for configurations
using custom skeletons as well as default skeletons.

Signed-off-by: Danomi Manchego <danomimanchego123@gmail.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

dropbear: bump to version 2014.64

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

package/fetchmail: gettextize to match the new gettext version

The gettext macros in the package are from a gettext version older than
the one we have in Buildroot, so autoreconf fails.

Run gettextize prior to running autoreconf, like we do in wget.

Fixes:
http://autobuild.buildroot.net/results/8dd/8dd6651ce99c0b81497fd285909b2b614009e273/
http://autobuild.buildroot.net/results/0bb/0bbfddda94ac7f75f74a54157c27de5b4fb3e559/
http://autobuild.buildroot.net/results/92b/92bcb493619ce4508bc4571cbfc3d5c3bdf49ff4/
http://autobuild.buildroot.net/results/768/768efd71745c72cec159edaa92e266167629ec5d/

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

ltrace: use current master

ltrace 0.7.3 is the latest release but it is actually broken on ARM
since PTRACE_SINGLESTEP emulation has been removed, see:

http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=425fc47adb5bb69f76285be77a09a3341a30799e

It fails with:
PTRACE_SINGLESTEP: Input/output error

Using master solves that until a new release is made.

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

python-daemon: new package

[Thomas: adjust license information to use SPDX code for Python
license, add GPLv2+ which is used for the tests, and fix the license
file variable.]

Signed-off-by: David Bachelart <david.bachelart@bbright.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

python-ipython: new package

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

python-flup: new package

This module is useful to implement a fastCGI webserver
Tested with ARM and Python 2.7.6

[Thomas: fix license informations.]

Signed-off-by: oli vogt <oli.vogt.pub01@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

docs/manual: enhance doc for BR2_TOOLCHAIN_HEADERS_AT_LEAST_X_Y

The BR2_TOOLCHAIN_HEADERS_AT_LEAST_X_Y symbols can be used to denote the
lowest required kernel headers version, but also the highest supported
version. Document that.

Use the `...` construct instead of +...+ used eveywhere else, otherwise
asciidoc would render +<=+ to the ⇐ arrow, while `...` does no rendering
at all.

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

python-numpy: add patch to fix build on uClibc

Fixes:

  http://autobuild.buildroot.org/results/01d/01da44bc4b644a46326d8fbd87708b7a98971487/

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

python3: define license and license files

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

python3: optional decimal module support

This commit makes the decimal module support optional: since Python
3.3 it relies on an external library. The source code of this library
is available built-in as part of the Python sources, but in Buildroot
we generally prefer to use the external library when possible.

To achieve this, this commit adds a patch to Python that is similar to
the one we use for expat support, but this time for the
libmpdec/mpdecimal library.

As a consequence, since mpdecimal now builds properly even when
<fenv.h> is not available (on i386), this commit fixes:

  http://autobuild.buildroot.org/results/b64/b64d5c941a7cac00619da3a0696939f86a8eafc2/

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

mpdecimal: new package

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