python-backports-abc: bump to version 0.5

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

fontconfig: needs host-gperf

Commit 701380c53b4d("fontconfig: add build fix for glibc 2.25") added
a patch that touches src/fcobjs.h. Package build system detects change
in timestamp and tries to regenerate fcobjshash.gperf which requires
gperf tool. So this commit adds dependency on host-gperf.

Since the patch is in upstream fontconfig, we can get rid of this
dependency once version is bumped next time.

Fixes:
  http://autobuild.buildroot.net/results/174/17471088b26ebc840d1922d7f2aa9e69b4b49ced

Cc: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Rahul Bedarkar <rahul.bedarkar@imgtec.com>
[Thomas: improve comment in the code.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

iptables: fix build with musl

Add a patch that defines _GNU_SOURCE to expose some struct tcphdr fields.

Fixes:
http://autobuild.buildroot.net/results/bee/bee20b689345b593378cfd91e0abc48bf381d3f9/

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

execline: new package

This package provides execline, a (non-interactive) scripting language,
like sh, used in the s6 supervision system.

The host variant is provided as it is required to build and run the host
variants of s6 and s6-rc.

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
[Thomas: add entry to DEVELOPERS file.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/x11r7/xdriver_xf86-video-vmware: fix compilation without udev

https://cgit.freedesktop.org/xorg/driver/xf86-video-vmware/commit/configure.ac?id=755e38f373a511bd774a61c9420a67f630f71037
added a check for libudev which fails if the check is not disabled.

Fixes
http://autobuild.buildroot.net/results/7b3/7b38105dd1115ac622964cf243ac137b7624fb43/

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

toolchain-external: fix broken handling of 'usr/lib/locale'

Function copy_toolchain_sysroot, which is in charge of copying the relevant
bits from the external toolchain to the staging directory, performs an rsync
loop of various directories and excludes the pattern 'usr/lib/locale' with
the intention of skipping the directory <toolchain>/usr/lib/locale.

However, while this worked in the original commit, commit
5628776c4a4d29d0715633ea463b64cc19e19c5a broke it inadvertently. The
relevant part of the diff:

-    rsync -au --chmod=Du+w --exclude 'usr/lib/locale' \
-          $${ARCH_SYSROOT_DIR}/$$i $(STAGING_DIR)/ ; \
+    rsync -au --chmod=Du+w --exclude 'usr/lib/locale' \
+          --exclude lib --exclude lib32 --exclude lib64 \
+          $${ARCH_SYSROOT_DIR}/$$i/ $(STAGING_DIR)/$$i/ ; \

Notice how the source directory now contains a trailing slash, which impacts
the way the exclude rules are interpreted. Previously, when 'i' was 'usr',
the exclude of 'usr/lib/locale' would find a match. With the trailing slash,
there will never be a match, unless for a directory 'usr/usr/lib/locale'.
The right rule would have been '--exclude lib/locale'.

However, just that fix does not solve the problem in all cases, in
particular in the (common) case where ARCH_LIB_DIR is 'lib'. This is due
another change in that commit, changing the iterated values of the above
rsync:

- for i in etc $${ARCH_LIB_DIR} sbin usr ; do \
+ for i in etc $${ARCH_LIB_DIR} sbin usr usr/$${ARCH_LIB_DIR}; do \

Due to the fact that we rsync both 'usr' as 'usr/lib' (assuming ARCH_LIB_DIR
is 'lib') we need to add the correct exclude in both cases. But the exclude
is different for both. When i == 'usr', the correct exclude rule would be
'--exclude lib/locale' while when i == 'usr/lib' the correct rule would be
'--exclude locale'.

Since we would like to avoid separate cases for this, use the following
exclude: '--exclude locale/'. The trailing slash will make sure only
directories called 'locale' will match. The targeted directories are then
usr/lib/locale and usr/share/locale. The latter directory was not matched
originally, but it should not hurt changing that.

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

toolchain-external: reduce nesting in copy_toolchain_sysroot

As discussed with Thomas Petazzoni, we can reduce the nesting level by early
returning on an invalid iteration.

I did not move the 'else' case (the common case) outside the if-else because
it would make the code less symmetrical and IMO makes it _less_ clear.

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

mesa3d: Add support for the Etnaviv gallium driver

Add support for the Etnaviv gallium driver.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Tested-by: Gary Bisson <gary.bisson@boundarydevices.com>
Reviewed-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/{mesa3d, mesa3d-headers}: bump version to 17.0.0

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

glmark2: add imx-drm driver support

Upstream status: pending
https://github.com/glmark2/glmark2/pull/29

Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

perl-gd: fix build of native part

EUMM don't find .xs file in subdirectory (only .pm files are handled)
So, let move lib/GD.xs in the root directory.

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

libsidplay2: fix build on ppc64le by not autoreconfiguring

The libsidplay2 package build system is completely broken. It is made
of a top-level configure script, which calls into sub-configure
scripts in sub-directories. However, since it doesn't use the autoconf
provided AC_CONFIG_SUBDIRS() mechanism, an "autoreconf" doesn't
recurse into the subdirectories.

Due to this, the aclocal.m4 in the libsidplay/ subdirectory doesn't
get re-generated when Buildroot autoreconfs the package. However,
since we patch one of the .m4 files in this subdirectory, when build
time comes, the package notices its aclocal.m4 is older than one of
the .m4 file, and triggers an automatic autoreconf.

Since <pkg>_AUTORECONF = YES is enabled, this automatic autoreconf
works fine: host-autoconf and host-automake are available.

Expect that on powerpc64le, we patch the configure script itself to
make it recognize powerpc64le. But this patching of the configure
script itself gets overwritten by the automatic autoreconf at the
beginning of the build step, causing the build to fail on powerpc64le.

Switching to AC_CONFIG_SUBDIRS() would allow to fix this, but
libsidplay2 needs to pass custom configure options to each of the
sub-configure scripts, something that AC_CONFIG_SUBDIRS() doesn't
support. And since libsidplay2 upstream looks completely dead, the
incentive to fix the whole thing is very limited.

Since what's broken is the autoreconfiguration of the package, what we
do is modify patch 0001-sidplay2-libs-2.1.1.patch to directly tweak the
configure script (instead of the relevant .m4 file). Thanks to this,
<pkg>_AUTORECONF = YES is no longer needed, the .m4 file is no longer
newer than the sub-configure script, and no automatic autoreconf
triggers at build time. This allows the package to build properly on
powerpc64le.

While we normally don't like patching 'configure' scripts directly, in
this case the size of the change in the configure script is very small,
and as explained above, the incentive to fix the package properly is
very limited.

In detail, the changes:

 * Patch 0001-sidplay2-libs-2.1.1.patch is turned into a Git-formatted
   patch

 * The irrelevant changes to Makefile.in files, aclocal.m4, config.h.in,
   sidint.h are removed.

 * The change to my_macros.m4 is applied directly to the corresponding
   configure script.

 * The change to the configure.ac script regarding libdir is applied
   directly to the corresponding configure script.

 * The change to the configure.ac script regarding "*-k*bsd*-gnu" is
   dropped, since we don't care about kFreeBSD support.

 * LIBSIDPLAY2_AUTORECONF = YES is dropped from the .mk file.

Fixes:

  http://autobuild.buildroot.net/results/1f6a42bfece24e09c9c7f4078d549ec5c099c89d/

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

Revert "package/cwiid: add optional dependency to bluez5_utils"

This reverts commit d4c4c259d25a8f2c5c2e2a5736609ba285fdd018. It was
mistakenly committed and pushed: it causes Kconfig circular
dependencies, so it cannot be applied as-is.

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

fontconfig: add build fix for glibc 2.25

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

glibc: drop version 2.22

We don't want a dozen glibc versions and there's no particular reason to
keep this old version around so drop it.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
[Thomas: add entry to Config.in.legacy.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

glibc: add version 2.25

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

glibc: bump default to version 2.24

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

package/cwiid: add optional dependency to bluez5_utils

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

package/cwiid: bump version

kodi needs CWIID_MESG_BALANCE
https://github.com/xbmc/xbmc/blob/Krypton/tools/EventClients/Clients/WiiRemote/CWIID_WiiRemote.cpp#L106

It was added after the last cwiid release, which took place 2009:
https://github.com/abstrakraft/cwiid/commit/2174214bc2caca51aa49a47025ccb80b3f75e53f

Instead of adding another 20k-sized patch we switch to the upstream
github repo and bump to its HEAD commit, dating back to 2010.

By doing this we can remove two patches which were applied upstream:

0001-fix-link-options-for-as-needed-90.patch
https://github.com/abstrakraft/cwiid/commit/6af678616531eb1f3d3d0a052313ef9d8125bac7

0002-Update-for-BlueZ-changes.patch
https://github.com/abstrakraft/cwiid/commit/c5dd7d4a9af5a7d8ead8ad26d9e5e0f8f8292d29

The remaining patches were renumbered.

This patch is needed for the upcoming Kodi version bump which also adds
optional cwiid support.

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

docs/website: improve website speed

Currently the buildroot website is not leveraging common
techniques to have a better loading speed (browser caching,
gzip compression, deflating).
This commit provides an .htaccess files with all the needed to
enable mentioned features. This .htaccess only works if the
webserver has deflate, expires and headers modules enabled.

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

memtool: new package

memtool allows one to read and write memory mapped registers via /dev/mem.
The commands are inspired by the respective commands of the barebox
bootloader.  This is handy during driver development to inspect and modify
register settings.  It can also be used to modify regular files and
character devices (e.g. to paint to /dev/fb0).

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
[Thomas: add entry to DEVELOPERS file.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

putty: bump to version 0.68

Drop upstream patches.

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

Revert "dependencies/cmake: blacklist cmake 3.7"

This reverts commit 4422eca2d418e2b817b419ff6c4c62f7f4cb7871.

We now have a workaround for the RPATH issue introduced in 3.7, so we
can use it again.

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

mpd: fix build for ARC

mpd package fails for both internal and external ARC toolchain as check
for pthread support fails. Such checks fails because _REENTRANT flag is
not defined in gcc even when -pthread is passed.

So we add patch to gcc that defines _REENTRANT  on ARC when -pthread is
passed.

Also it disables mpd package for external ARC toolchain as it fails due
to the same issue.

This patch should be reverted as soon as the patch for GCC becomes a
part of ARC toolchain.

Fixes:
  http://autobuild.buildroot.net/results/7d7/7d70b62ad996830fbeca46dffcc7a1dc030e575d//

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

cmake: bump version to 3.7.2

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

cmake: adjust dependencies added for libuv

Commit 1ffcf364b6e9894a876dc581a090f87685945412 updated cmake to 3.7.0,
which requires selecting the libuv package. At the time, the libuv
package only depended on BR2_TOOLCHAIN_HAS_THREADS. However, later on,
it was changed in master to depend on BR2_TOOLCHAIN_HAS_THREADS_NPTL, a
change which was not taken into account in the cmake 3.7.0 bump that was
merged in the next branch.

Due to this, builds of cmake is attempted on architectures that don't
provide NPTL thread support, causing a build failure. This commit fixes
that by adjusting the dependency.

Fixes:

  http://autobuild.buildroot.net/results/16a5e1cbb57c0124537c4f3dc0807ba1eaa975ec/

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

cmake: bump version to 3.7.1

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

cmake: bump version to 3.7.0

libuv is now a required dependency.

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Tested-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

core/pkg-cmake: provide our own platform description

The handling of RPATH in cmake-3.7 has changed drastically, causing a
slew of build failures dues to libraries from the host being pulled in:

  - domoticz : http://autobuild.buildroot.org/results/fd0/fd0ba54c7abf973691b39a0ca1bb4e07d749593a/
  - freerdp  : http://autobuild.buildroot.org/results/5d4/5d429d0e288754a541ee5d8be515454c5fccd28b/
  - libcec   : http://autobuild.buildroot.org/results/3f3/3f3593bab7734dd274faf5b5690895e9424cbb89/
  - and so on...

The bug was reported upstream [0], which dismissed it altogether [1] as
being expected behaviour, quoting:

    I don't think there is anything wrong with that change on its own.
    It merely exposed some existing behavior in a new case.

Instead, upstream suggested in that same message that a platform
definition be used instead, quoting:

    If a toolchain file specifies CMAKE_SYSTEM_NAME such that a custom
    `Platform/MySystem.cmake` file is loaded then the latter can set
    them as needed for the target platform.

So here we are doing so:

  - we add a new platfom definitions that inherits from the Linux one,
    then overrides the problematic settings;

  - we change our toolchain file to use that platform instead;

  - we tell cmake where to find additional modules, so that it can find
    our custom platform file.

This has been tested to work in the following conditions:

 - pre-installed host cmake, versions 3.5.1 (Ubuntu 16.04) and 3.7.2
   (manually built)

  - internal cmake, versions 3.6.3 (the current version as of this
    patch) and 3.7.2 (with the followup patches).

Thanks to Jörg, Ben and Baruch for the help investigating the issue.
Special thanks to Jörg for handling the discussion with upstream and
pointing to the relevant messages! :-)

[0] http://public.kitware.com/pipermail/cmake/2017-February/064970.html
[1] http://public.kitware.com/pipermail/cmake/2017-February/065063.html

To be noted: Thomas suggested we set these directly in the toolchain
file. Unfortunately, wherever we put those settings in the toolchain
file, this does not work.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Jörg Krause <joerg.krause@embedded.rocks>
Cc: Ben Boeckel <mathstuf@gmail.com>
Cc: Baruch Siach <baruch@tkos.co.il>
Cc: Samuel Martin <s.martin49@gmail.com>
Cc: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

classpath: enable on sparc and sparc64

GNU classpath builds fine on sparc and sparc64 architectures. So
we allow building it there to be consistent with the naming of
BR2_PACKAGE_CLASSPATH_ARCH_SUPPORTS.

Signed-off-by: Marcus Hoffmann <m.hoffmann@cartelsol.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/sdl2: disable altivec built-in function for PowerPC

As reported in this bug report [1], altivec support in SDL break
arbitrary C++ code.

Issue reported by test-pkg script while testing supertux package:
    error: could not convert 'true' from 'bool' to '__vector(4) __bool int'

[1] https://bugs.debian.org/cgi-bin/bugreport.cgi/?bug=770670

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Sam Bobroff <sam.bobroff@au1.ibm.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

fs/iso9660: doesn't support (grub2) EFI

The iso9660 generation for grub2 assumes that grub-eltorito.img is
available. However, this image is only available for the i386-pc target
(i.e. legacy BIOS). An EFI-bootable iso9660 requires a different layout.

Since we currently can't generate the EFI-bootable iso9660 layout,
require the i386-pc target to be selected in grub.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reported-by: arnaud.miche@orange.com
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

cjson: bump to version 1.3.2

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

cjson: fix build for toolchains without ssp

Set ENABLE_CUSTOM_COMPILER_FLAGS to OFF to disable custom flags, in
particular -fstack-protector-strong which depends on
BR2_TOOLCHAIN_HAS_SSP

Fixes:

  http://autobuild.buildroot.net/results/cfcf3bc8066159dfddd1786954d78e8527858f2f/

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ncftp: fix host/target confusion

The ncftp build process tries to build and run a small program called
ccdv to beautify the build process output. If it manages to build and
run it, then it uses it.

Unfortunately, this doesn't work well when the target architecture is
close to the host architecture, but not exactly the same. Because both
architectures are close to each other, the test run of ccdv succeeds,
but real use of ccdv during ncftp build process causes an Illegal
instruction issue.

This for example happens with the CodeSourcery AMD64 toolchain, on a
build machine running an i7-4600U, and has been detected in the
autobuilders since the CodeSourcery AMD64 toolchain was upgraded at
the end of January:

  http://autobuild.buildroot.net/?reason=ncftp-3.2.6

The issue was also reported by Christopher Arguin back in July 2016:

  http://lists.busybox.net/pipermail/buildroot/2016-July/168026.html

and at the time, we identified that simply disabling the ccdv tool, by
passing --disable-ccdv, was enough to solve the issue. But Christopher
never submitted the patch, so the problem remained unfixed.

Therefore, we pass --disable-ccdv to the configure script, which
fixes:

  http://autobuild.buildroot.net/results/6eadad0e879ca70bb07b13b4196d42c64b11699f/

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

Merge branch 'next'

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

Kickoff 2017.05 cycle

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

Update for 2017.02

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

qt5base: fix VNC platform plugin build on big-endian machines

Fixes [1], [2]:

  In file included from qvncintegration.cpp:38:0:
  qvncscreen.h:76:22: error: expected ';' at end of member declaration
       bool swapBytes() const

[1] http://autobuild.buildroot.net/results/718/7180ad7d7e78f1c3ce6a1f83a5a8fcc5e0ac03a2
[2] http://autobuild.buildroot.net/results/14f/14f8c436b184192681fb79c7f0c0d0daa35f867b

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

mpv: disable with ARC external toolchain

This package fails when building with ARC external toolchain

Marking this with special comment "# ARC toolchain issue" as the package
is to be enabled as soon as the issue with the ARC toolchain is
resolved.

Fixes:
  http://autobuild.buildroot.net/results/fb6/fb677a917545adee321bdcd2c2519c81326448c4//

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

xterm: disable with ARC external toolchain

This package fails when building with ARC external toolchain

Marking this with special comment "# ARC toolchain issue" as the package
is to be enabled as soon as the issue with the ARC toolchain is
resolved.

Fixes:
  http://autobuild.buildroot.net/results/28a/28a92049a6ceef005787c5779f77ecf3fe8ad642//

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

qt5quickcontrols: fix installation of PrivateWidgets

Commit e482ebf51d9e8e00c3e58eb65af0dfb70d05d8cc attempted to fix the
installation of qt5quickcontrols by only installing the PrivateWidgets
directory for the 5.8.0 version.

However, the availability of PrivateWidgets has nothing to do with the
version; in both 5.6.2 and 5.8.0, the installation is gated by the
following statement in src/src.pro:

qtHaveModule(quick):qtHaveModule(widgets): SUBDIRS += widgets

i.e. it is installed when both the Quick and the Widgets module are
available. The Widgets module is controlled by Buildroot's
BR2_PACKAGE_QT5BASE_WIDGETS symbol, the Quick module is controlled by
Buildroot's BR2_PACKAGE_QT5DECLARATIVE_QUICK. The qt5quickcontrols
package selects BR2_PACKAGE_QT5DECLARATIVE_QUICK so it is not really
needed to include it in the condition. However, it is theoretically
possible to build this package without QtQuick. Also, adding this
condition makes it consistent with src.pro.

Note that commit e482ebf51d9e8e00c3e58eb65af0dfb70d05d8cc introduces a
second fix (not mentioned in the commit message): for version 5.6.2, the
Layouts directory is installed, but in 5.8.0 this directory doesn't
exist any more. Therefore, a separate condition on the version is still
needed.

Fixes:
http://autobuild.buildroot.net/results/1ff3e9ad4ba518d0a37f9fc12038bf9020f28094

Cc: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

genimage: Fix typo in description

Signed-off-by: Marcus Hoffmann <m.hoffmann@cartelsol.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

classpath: only allow on supported architectures

In commit a75ab1fc1dcc7eee8668b1c3c19b703f1390b2cd ("package/classpath:
Don't depend on jamvm"), we removed the dependency of classpath on
jamvm. Since jamvm is only available for a reduced set of architectures,
classpath could until this commit until be built on those architectures.

However, now that this dependency has been removed, classpath can
potentially be built for all architectures supported by Buildroot, even
though it doesn't support all of them.

Since adding support for additional architectures in classpath doesn't
make much sense, because classpath is in Buildroot only usable with
JamVM anyway, and JamVM is only available for a small set of
architectures, this commit simply makes classpath available on the
architectures that it supports.

By doing so, it also removes the or1k support patch which was added by
commit f12a146f817c8ef07a7d41a31a5336b5ef6a96e8, since anyway or1k is
not supported by JamVM.

Fixes:

  http://autobuild.buildroot.net/results/55eb89f89e96b94a821778bc18ed844af08b7460/
  (classpath on microblaze)

  http://autobuild.buildroot.net/results/279dd731bd9ecf5f9d54bda3715caeaa7cbcdbb3/
  (classpath on nios2)

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

openocd: avoid documentation rebuild to fix build issues

On PowerPC64(le), we patch all configure scripts. Due to this, the
version.texi in OpenOCD files gets regenerated, and then since it has
a newer date than openocd.info, openocd build system rebuilds the
documentation. Unfortunately, this documentation rebuild fails on old
machines.

We work around this by faking the date of the generated version.texi
file, to make the build system believe the documentation doesn't need
to be regenerated.

Fixes:

  http://autobuild.buildroot.net/results/3cbe65a46e75b8e67846d593884c96df97dec7a4

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Sam Bobroff <sam.bobroff@au1.ibm.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

dependencies/cmake: blacklist cmake 3.7

cmake-3.7 has a bug in how it handles rpath, linking with libraries from
the host.

Until we completely understand the issue, just blacklist cmake-3.7.

The issue has been reported upstream:
    http://public.kitware.com/pipermail/cmake/2017-February/064970.html

Reported-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Jörg Krause <joerg.krause@embedded.rocks>
Cc: Ben Boeckel <mathstuf@gmail.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

assimp: fix gcc version condition

In commit 68cebedeb977396f64f5f591f5e93f34fab8e0a8 ("assimp: work around
gcc bug on SuperH"), a work around was added to make the package build
with gcc on SuperH. The condition included a test on the gcc version,
which was mistakenly done on the host gcc version, while a test on the
target gcc version was intended.

Thanks to Peter Korsgaard for spotting the issue.

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

systemd: enable optional PAM support

This commit enables PAM support in systemd if BR2_PACKAGE_LINUX_PAM is
set. Some essential config files are not installed without the
--enable-pam option.

Signed-off-by: James Balean <james@balean.com.au>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

assimp: work around gcc bug on SuperH

gcc versions earlier than gcc 6.x fail to build assimp on SuperH when
static linking:

  AssxmlExporter.cpp:623:1: error: unable to find a register to spill in class 'GENERAL_REGS'

It's the combination of -Os and *not* having -fPIC that makes gcc
fail, which explains why configurations with dynamic linking work
fine.

 -Os -fPIC -> works
 -Os       -> fails
 -O2 -fPIC -> works
 -O2       -> works

Therefore, as a workaround, we are forcing the use of -O2 on SuperH
when the gcc version is older than gcc 6.x and we're statically
linking.

Fixes:

  http://autobuild.buildroot.net/results/ec88aa8118179e30e24603cc45292047dca19216/

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

assimp: remove dead "depends on" dependency

BR2_TOOLCHAIN_EXTERNAL_BLACKFIN_UCLINUX has been removed in commit
311bc137da2952445683413794f3ebc8e477c9ef ("toolchain: kill ADI
Blackfin toolchain"), so this "depends on" is useless.

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

vlc: workaround microblaze gcc internal compiler error

Override gcc optimization flags with -O0. The workaround is not required
for gcc6 anymore, so we take this into account into our condition.

Fixes:

  http://autobuild.buildroot.net/results/a318f0838a6a602046e719103ac81965c0084d52

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
[Thomas: add gcc 6.x condition.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

usb_modeswitch_data: bump to version 20170205

Drop upstream patch.

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

usb_modeswitch: bump to version 2.5.0

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

modem-manager: bump to version 1.6.4

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

mplayer: not supported on or1k

MPlayer doesn't support OpenRISC:

The architecture of your CPU (or1k) is not supported by this configure script
It seems nobody has ported MPlayer to your OS or CPU type yet.

so disable it on this architecture.

Fixes:

  http://autobuild.buildroot.net/results/47207dfd10ff6e5ec4ccdcf8454aaf5f408ad1e3/

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

python-libconfig: disable on musl

The python-libconfig package fails to build with musl with very weird
errors coming all the way from Boost Python, which nobody ever
bothered to fix. It's time to disable this package on musl to avoid
the repetitive build failures.

Fixes:

  http://autobuild.buildroot.net/results/f0f6cdc8c38c024772615d5e677b0f4ad63ef7ec

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

Update for 2017.02-rc3

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

libuv: add dependency on BR2_TOOLCHAIN_HAS_SYNC_4

The libuv library uses __sync atomic built-ins:

$ readelf -a -W output/target/usr/lib/libuv.so.1.0.0 | grep __sync
   122: 00000000     0 NOTYPE  GLOBAL DEFAULT  UND __sync_val_compare_and_swap_4

but this is not currently taken into account in the libuv package,
causing some build failures in packages using libuv as an optional
dependency, such as janus-gateway and mosquitto.

Therefore, this commit updates the libuv package with this additional
dependency.  The reverse dependencies of libuv are also updated: luv, luvi
and moarvm.

Fixes:

 - http://autobuild.buildroot.net/results/bdaa67d763c0d8d377a04529c74f73bee7d4ccef/
   (janus-gateway)

 - http://autobuild.buildroot.net/results/2bc84ba2d1167018e2d48e5183ead22b6425dcf5/
   (mosquitto)

[Peter: drop cmake changes after cmake revert]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

htop: work-around old uClibc issue when static linking

htop wants to use backtrace() support from the C
library. Unfortunately, with old uClibc versions such as the one we
use for the ARC architecture, the backtrace() implementation is in
libubacktrace. In addition, this implementation needs dladdr()
support, which is in libdl, not available when static linking.

Since this problem no longer exists in more recent versions of uClibc,
we simply special case the ARC+static linking case, and make the
configure script believe that <execinfo.h> is not available.

Fixes:

  http://autobuild.buildroot.net/results/cdea351fad7a0f61ddec3e6a141da8da0523a902/

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

pkg-cmake: pass <pkg>_CONF_ENV for the target configure step

The Buildroot manual documents that <pkg>_CONF_ENV is passed in the
environment when calling cmake during the configure step. However, the
actual implementation in pkg-cmake passes HOST_<pkg>_CONF_ENV when
configuring the host variant of a cmake package, but does not pass
<pkg>_CONF_ENV when configuring the target variant of a cmake package.

This commit fixes that by passing <pkg>_CONF_ENV in the environment as
expected. It should not cause any behavior change, because this
feature is in fact not used by any package in upstream Buildroot:

$ grep CONF_ENV $(git grep -l cmake-package package/)
package/pkg-cmake.mk:$(2)_CONF_ENV ?=
package/pkg-cmake.mk: $$($$(PKG)_CONF_ENV) $$(BR2_CMAKE) $$($$(PKG)_SRCDIR) \
package/pkg-cmake.mk: $$($$(PKG)_CONF_ENV) $$(BR2_CMAKE) $$($$(PKG)_SRCDIR) \

This issue was reported by Olivier <ovalentin@awox.com> as bug #9616.

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

Revert "cmake: bump version to 3.7.0"

cmake 3.7 causes serious regressions in some cmake-based packages,
related to how RPATH is handled.

This reverts commit 1ffcf364b6e9894a876dc581a090f87685945412.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Jörg Krause <joerg.krause@embedded.rocks>
Cc: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Revert "cmake: bump version to 3.7.1"

cmake 3.7 causes serious regressions in some cmake-based packages,
related to how RPATH is handled.

This reverts commit d96fafc3d346af8790df45557b16069d6ce81e79.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Jörg Krause <joerg.krause@embedded.rocks>
Cc: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Revert "cmake: adjust dependencies added for libuv"

cmake 3.7 causes serious regressions in some cmake-based packages,
related to how RPATH is handled.

This reverts commit b754237520a4795eadc0edec768b267dd598ab0e.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Jörg Krause <joerg.krause@embedded.rocks>
Cc: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Revert "cmake: bump version to 3.7.2"

cmake 3.7 causes serious regressions in some cmake-based packages,
related to how RPATH is handled.

This reverts commit f8a6990c92451dab2ff9e14f3e94fdecbd79b604.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Jörg Krause <joerg.krause@embedded.rocks>
Cc: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/directfb: disable for static builds

DirectFB is written in C++, but can be linked to by a C library or
program.

When doing shared link, this is fine because the linker gets help from
the DT_NEEDED flags and knows what libraries to pull in during the link.

However, during a static link, the linker does not get such help.

Properly fixing this would require that there is support in pkg-config
and autotools to specify that the C++ runtime must be linked. Alas there
is no sush support. The only option is to add -lstdc++ to the
Libs.Private field in directfb.pc. But this is not upstreamable, because
there are other C++ runtimes in the wild (e.g. -lc++ from llvm/clang).

However, DirectFB in a static scenario is probably not a very common
scenario.

Disable DirectFB for static builds.

Fixes:
    http://autobuild.buildroot.org/results/3d3/3d3036d40ddad71d872d910aae7a24975706d2e9/
    http://autobuild.buildroot.org/results/d1c/d1c35a6003396942b584f2f2a5e8bf4ac2fbe370/
    http://autobuild.buildroot.org/results/d45/d4504871bd47930e8363032d380cdfcc5bb8aee7/
    [...]

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

nano: bump to version 2.7.5

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

libsigsegv: bump to version 2.11

Drop all patches, they're upstream.
And match that by dropping autoreconf as well.

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

kmod: bump to version 24

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

ti-cgt-pru: bump version to 2.1.4

Signed-off-by: Cam Mannett <camden.mannett@protonmail.ch>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

pru-software-support: bump version to 5.1.0

Signed-off-by: Cam Mannett <camden.mannett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

support/misc/Vagrantfile: reorganize provisioning

* Remove distribution upgrade cause it slows down the first boot and
  presents a bug when executed non interactively.

* Reorganize provision scripts to be in privileged and non privileged
  sections

* Add Ubuntu mirror automatic handling for apt packages sources

Fixes bug #9581

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

package/lugaru: new package

Lugaru needs SDL2 library with OpenGL support which require X11 support.
Otherwise Lugaru error out with "No dynamic GL support in video driver"
message.

https://osslugaru.gitlab.io/

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

package/sdl2: enable OpenGL ES support

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

package/sdl2: enable opengl support

OpenGl support mean GLX so it require X11 support.
Select automatically BR2_PACKAGE_SDL2_X11 to enable minimal X11
libraries support (libx11 and libxext).

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

classpath: add patch to fix or1k compile error

Fixes:
  http://autobuild.buildroot.net/results/a0bb5535fd197b9dd1607a5f0bca477c9575c99f

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

ncurses: remove BR2_PACKAGE_NCURSES_TARGET_{FORM, MENU, PANEL} options

The ncurses sub-options BR2_PACKAGE_NCURSES_TARGET_{FORM,MENU,PANEL}
are currently very badly broken: they only control whether the
libform, libmenu and libpanel libraries are installed in
$(TARGET_DIR), but do absolutely nothing about their installation in
$(STAGING_DIR).

This means that when one of those options is disabled, the
corresponding library is indeed not installed in the target, but is
available in staging. It can therefore be detected by the configure
script of another package and used... even though the library will not
be in the target, causing a runtime failure.

Internally, ncurses.mk uses the "make install" logic of ncurses for
the staging installation, but uses a completely hand-written logic for
the target installation, which is the reason for this
desynchronization between what's installed in staging and target.

When BR2_PACKAGE_NCURSES_WCHAR=y, this also causes some build
failures. Indeed, when BR2_PACKAGE_NCURSES_WCHAR=y, Buildroot creates
some symbolic links lib<foo>.so -> lib<foo>w.so in staging and target,
but only for the lib<foo> that have been enabled by
BR2_PACKAGE_NCURSES_TARGET_{FORM,MENU,PANEL}. Due to this, a package
that for example needed the libmenu library but forgot to select
BR2_PACKAGE_NCURSES_TARGET_MENU was:

 - Building fine with BR2_PACKAGE_NCURSES_WCHAR disabled (because
   libmenu.so exists in staging), but would fail to run at runtime
   because libmenu.so is not in the target.

 - Fail to build with BR2_PACKAGE_NCURSES_WCHAR=y because only
   libmenuw.so exists, and not the libmenu.so symbolic link.

Since those libraries are small (43K for libform, 21K for libmenu and
8.2K for libpanel), this commit takes the very simple approach of
removing those options, and installing the libraries
unconditionally. It therefore uses the "make install" logic for both
the staging *and* target installation.

In detail, this commit:

 - Removes the NCURSES_PROGS variable, not needed since
   --without-progs already allows to disable the build and
   installation of programs.

 - Removes the NCURSES_LIBS-y variable, and replaces it with a single
   unconditional assignement to NCURSES_LIBS, only used to create the
   lib<foo>w.so -> lib<foo>.so symbolic links when wchar support is
   enabled.

 - Removes NCURSES_INSTALL_TARGET_CMDS and the functions it was
   calling: NCURSES_INSTALL_TARGET_LIBS and
   NCURSES_INSTALL_TARGET_PROGS.

 - Adds a NCURSES_TARGET_SYMLINK_RESET hook to create the reset ->
   tset symbolic link, as was done before.

 - Adds a NCURSES_TARGET_CLEANUP_TERMINFO to cleanup the terminfo
   files in the target, so that we stay in the same situation in terms
   of installed terminfo files.

 - Removes the BR2_PACKAGE_NCURSES_TARGET_{FORM,MENU,PANEL} options
   from the Config.in files: both their definition and usage.

 - Simplifies all the symlink dance for lib<foo> -> lib<foo>w, because
   as Yann E. Morin suggested, this dance is only needed in staging, not
   in the target. Once binaries have been built, they refer to the
   SONAME of the library, which is the lib<foo>w variant (for shared
   linking). For static linking and .pc files, it's obvious that we
   don't care about them on the target. Therefore the
   NCURSES_LINK_LIBS_STATIC, NCURSES_LINK_LIBS_SHARED and
   NCURSES_LINK_PC functions no longer take any argument: they always
   apply to STAGING_DIR only. NCURSES_LINK_TARGET_LIBS is removed.

It is worth mentioning that adding Config.in.legacy support is *NOT*
necessary. Indeed:

 - If they were disabled before this patch, having them in
   Config.in.legacy would not trigger the legacy warning.

 - If they were enabled before this patch, then the behavior is
   unchanged: all libraries are now unconditionally installed. So
   there is no point in warning the user.

We double-checked the installed size of a filesystem containing just
ncurses before and after this patch, and the only folder that has its
size changed is /usr/lib, growing from 852 KB to 932 KB in the wchar
enabled case. That's a 80 KB system size increase.

This commit fixes the sngrep build failure and potentially numerous
runtime issues with ncurses.

Fixes:

  http://autobuild.buildroot.net/results/7b5db21a6c568e6c6c8fe2b5d5a2f5ca24df510c/

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

linux-headers: bump 4.{4, 9}.x series

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

linux: bump default to version 4.9.13

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

docker-engine: bump to version 1.13.1

- Remove patch as it has been applied upstream since 1.13
 - go packages are now in vendor instead of vendor/src so update
slightly the configure and build commands

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

docker-containerd: bump to version aa8187dbd3b7ad67d8e5e3a15115d3eef43a7ed1

Version bump as a dependency of docker-engine v1.13.1

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

runc: bump to version 9df8b306d01f59d3a8029be411de015b7304dd8f

Version bump as a dependency of docker-engine v1.13.1

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ccache: bump to version 3.3.4

Switch download URL to avoid a redirect.

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

openssh: Move key generation to the start function of init script.

Since there's not much point in generating missing host keys when the
init script is called with "stop", the call to ssh-keygen should not
be done inconditionally, but in the start function instead.

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

linux-headers: bump 4.{4, 9}.x series

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

linux: bump default to version 4.9.12

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

package/xfsprogs: fix static build

xfsprogs links to some libraries from util-linux. Those in turn may use
i18n functions when locales are enabled.

When doing a uClibc build, those i18n functions are provided by the
libintl library from the gettext package.

xfsprogs does not use pkgconfig to find the util-linux libraries, so it
misses the Requires.private field from libblkid, and thus misses the
Libs.private field from libuid.

And so a static link fails because of undefined references to i18n
symbols.

Furthermore, xfsprogs does not use automake; the hand-made Makefiles do
not make use of the LIBS variable set by configure.

We fix that in two ways:

  - pass LIBS=-lintl to configure, so that the configure step has a
    chance to succeed  (as suggested by Waldemar)

  - patch a Makefile to link with EXTRALIBS, and pass EXTRALIBS via
    XFSPROGS_MAKE_OPTS

Fixes:
    http://autobuild.buildroot.org/results/503/50314716a9f72e8bb238e52e9dc342c68f7e0b8b/
    http://autobuild.buildroot.org/results/253/253a024698b55dece992b2631e30edf4ff9092f7/
    http://autobuild.buildroot.org/results/3fe/3fe0d96cf13af12b0051a95f7865d4e38f4a78af/
    http://autobuild.buildroot.org/results/d48/d48e61785d25d33106b7dab1b5cb200cf27d4044/
    ...

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

libpcap: sparc needs -fPIC as sparc64/microblaze

Fixes nmap build for sparc.

Fixes:
 http://autobuild.buildroot.net/results/8e476939f1daaa20c316dbd0508fa343034f2397

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

synergy: Fix build with gcc 6

Fixes:
http://autobuild.buildroot.net/results/05b/05bbc22dd6cd5564462226e612ab1e95778fe935/
http://autobuild.buildroot.net/results/edf/edfe58749d1b7a1bb2e0184a6824a74b9d38ddb9/
http://autobuild.buildroot.net/results/139/1395eca13ca537bde928ddd68a5bc6e130e82ba3/
http://autobuild.buildroot.net/results/94a/94ad6e8bbbeb926ea834d9d6e3ba87d0398acb86/

Gcc 6.x defaults to C++14, and the iostream operator bool behaviour changed
in C++11.  In previous versions, a somewhat odd operator void* was used to
return the status of the stream as a pointer.  Since C++11 a more sensible
operator bool is used to return the stream staus.

For details, see:

http://en.cppreference.com/w/cpp/io/basic_ios/operator_bool

The code in CConfigReadContext assumes the pre-C++11 behaviour and provides
its own operator void overload to return the status of the embedded
iostream.  With C++11, iostream no longer provides this overload, breaking
the build:

CConfig.cpp: In member function 'CConfigReadContext::operator void*() const':
CConfig.cpp:1851:9: error: cannot convert 'std::istream {aka std::basic_istream<char>}' to 'void*' in return
  return m_stream;

To fix it, backport part of upstream commit 3d963bfbe7897d0a33ad (possible
fix for mavericks) which changes the code to simply provide a getStream()
method which returns a reference to the embedded stream and the calling code
is changed to use operator bool on the returned stream, making the code
compatible with both old and new compilers.

This upstream commit is part of the 1.6.0 release, so can be dropped when
the version is bumped.

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

binutils: fix for ld segfault for microblaze when --gc-sections is used

The workarounds for kmod/libnss can be removed when the patch is
applied and the autobuilder toolchains got rebuild.

kmod: 0d81107f021bf2a663f102bace1e3f9590641170
libnss: fceb1afd5dda45cf180f22877a6ab0e51d1b3dac

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

package/oracle-mysql: fix path to systemd service file

The path is a remnant from before the rename of the package.

Rather than hard-code the path, use the available variable. Also use
that variable when copying the Sys-V init script.

Fixes:
    http://autobuild.buildroot.org/results/767/767f0be3894c45c1e6d18bc4528d0da7b50e3757/
    http://autobuild.buildroot.org/results/e93/e934683ebcb33c9cc8fe9de7b1d5dcd7866f3bce/
    http://autobuild.buildroot.org/results/5d0/5d04163de80f7900d4aed14cff75b2e0761f101d/

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Ryan Coe <bluemrp9@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

toolchain-external: install libanl.so for glibc

libanl.so is needed for asynchronous network address and service
translation, declared in netdb.h

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

glibc: install libanl.so to target

libanl.so is needed for asynchronous network address and service
translation, declared in netdb.h

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

util-linux: re-add <pkg>_AUTORECONF = YES

Commit a5015f1025a1aee173c2c2ab39ccac5ce38f30c3 ("util-linux: security
bump to version 2.29.2") incorrectly removed <pkg>_AUTORECONF = YES.

While the patches touching configure.ac have indeed been removed, there
is still a patch touching a Makemodule.am file, which triggers an
autoreconf a build time, which itself fails because autoconf/automake
are not available.

So re-add <pkg>_AUTORECONF, with a comment pointing specifically at the
patch that makes it necessary.

Fixes:

  http://autobuild.buildroot.net/results/309127a532eed00e406bbaf0b1a51b7241a10505/

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

board/atmel/readme.txt: fix typos

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

package/libepoxy: needs host-pkgconf

configure.ac uses PKG_CHECK_MODULES() so it needs host-pkgconf.

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

gflags: bump to version v2.2.0

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

package/python: apply unicode config to host build

Currently the host build of Python 2 defaults to narrow unicode (UCS2),
ignoring the BR2_PACKAGE_PYTHON_UCS4 configuration option which may be
set to wide (UCS4).

This results in host and target Python packages which are incompatible
in subtle ways.

For example, installing wheels into the target fails when they are made
with the host python, citing incompatibility (as can be seen by the
package tags which may be "cp27u-manylinux1" instead of
"cp27mu-manylinux1").

Compiling the host Python 2 with the same UCS configuration as the
target ensures that the packages are compatible (and the tags match).

This does not affect Python 3 as support for narrow unicode was
deprecated in version 3.3, see https://www.python.org/dev/peps/pep-0393/

Thanks to Tony Breeds <tony@bakeyournoodle.com> for reporting this.

Signed-off-by: Chris Smart <mail@csmart.io>
[Thomas: add comment in the code.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gst-omx: bump to version 1.10.4

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

gst1-validate: bump to version 1.10.4

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

gst1-rtsp-server: bump to version 1.10.4

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