package/android-tools: drop musl-compatibility cdefs patching out

We now always have a sys/cdefs.h, so we no longer need to patch it out.

Simplify the patch by removing any hunk removing cdefs.h or the use of
__BEGIN_DECLS/__END_DECLS. However, it must be included when macros it
defines are being used.

Also, renumber patches to guarantee ordering (static patch was added
before big-endian one).

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

package/aircrack-ng: drop a musl-compatibility patch

We now always have a sys/cdefs.h, so we can drop this patch.

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

package/rpcbind: fix musl build

rpcbind uses the __P() macro defined in sys/cdefs.h but does not
includes it explicitly, relying on the glibc behaviour to include it
from its own headers.

But this macro does nothing but expands as its argument. In the past,
it did provide support for pre-ANSI compilers that did not support
function prototypes. Those are now gone, and glibc provides this macro
for legacy code.

Patch out the use of __P() altogether.

Fixes:
    http://autobuild.buildroot.org/results/346/346c5b562c244ddb039db2f86f71d670a01ed9ce/
    http://autobuild.buildroot.org/results/d77/d77454265c45539178c008c5d92509c441447675/
    and numerous such autobuild failures

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>

package/musl-compat-headers: provide compatibility headers not in musl

musl provides neither sys/queue.h nor sys/cdefs.h. Those two headers are
however quite widely used in a lot of packages (though they should at
least not use cdefs.h which is only full of mostly-legacy macros, and
which is mostly an internal header of glibc and was never really meant to
be exposed to, and used by packages).

But we don't live in an ideal world, so a lot of packages break when
those two headers are missing.

We already took care of sys/queue.h with the netbsd-queue package. But
the need for cdefs.h is getting more and more pressing.

We rename the netbsd-queue package into musl-compat-headers, and we
make it install sys/queue.h (from NetBSD) and sys/cdefs.h (a minimalist
one we bundle in Buildroot). We can't use the cdefs.h from NetBSD
because it includes machine-dependent headers; instead we bundle a very
minimalistic one, that covers only what we need.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Baruch Siach <baruch@tkos.co.il>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

linux: cpupower needs gettext

The cpupower linux tool needs gettext, always (even without locales).

We need to disable NLS, otherwise it tries to compile the .po files.
We also need to pass -lintl, otherwise it forgets to link with it
(because, the world is glibc-only, you did not know? And glibc does not
need we link with -lintl, so why would we? Oh, yes, we also reinvented
our super intelligent one-off Makefile rather than use one of the
standard buildsystems).

Fixes #9181:

      CC       utils/helpers/sysfs.o
    In file included from utils/helpers/amd.c:9:0: ./utils/helpers/helpers.h:13:21: fatal error: libintl.h: No such file or directory
     #include <libintl.h>
                         ^

Without NLS=false (yes, we could depend on host-gettext):

      MSGFMT   po/de.gmo
    make[3]: msgfmt: Command not found

Without LDFLAGS=-lintl:

      CC       cpupower
    ./utils/cpupower.o: In function `main':
    cpupower.c:(.text.startup+0x1a4): undefined reference to `libintl_textdomain'
    ./utils/idle_monitor/cpupower-monitor.o: In function `list_monitors':
    cpupower-monitor.c:(.text+0x5dc): undefined reference to `libintl_gettext'
    ./utils/cpupower-set.o: In function `cmd_set':
    cpupower-set.c:(.text+0x38): undefined reference to `libintl_textdomain'
    ./utils/cpupower-info.o: In function `cmd_info':
    cpupower-info.c:(.text+0x20): undefined reference to `libintl_textdomain'
    collect2: error: ld returned 1 exit status

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

manual: fix typo

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

libiio: Add a patch to fix compilation with old glibc

Cherry-picked from upstream.

Fixes:

  http://autobuild.buildroot.net/results/fa7f7db861c342f4d189f74b983ae9d2751cf29b/

Signed-off-by: Paul Cercueil <paul.cercueil@analog.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/rt-tests: add syscall-number for sched_(gs)etattr() for SH4

The dependency on kernel headers >= 3.14 was added by
b9fa6c00b1213fd038f2aed52057f22c7e18fd4b since
librttest requires __NR_sched_setattr and
__NR_sched_getattr syscalls (rt-sched).

But these syscalls have just been added to kernel 4.8 for SH4.

Fixes:
http://autobuild.buildroot.net/results/67fc16366717f34df492683a14ecd23376e23110

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

uclibc: ARC: Support syscall ABI v4

When used with GCC 6 ABIv4 is used.
Missing this patch leads to numerous runtime errors.

The patch has already been accepted in uclibc-ng:
http://cgit.uclibc-ng.org/cgi/cgit/uclibc-ng.git/commit/?id=3e40f9669279f005f7154892539166f5081fbcb2

So the patch should be removed after update to a new version of uclibc-ng.

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

core/pkg-autotools: don't use APPLY_PATCHES for libtool patches

Since [1] <package>-reconfigure target is broken.

$ make elementary-reconfigure
Applying buildroot-libtool-v2.4.4.patch using patch:
Error: duplicate filename 'buildroot-libtool-v2.4.4.patch'
Conflicting files are:
  already applied: buildroot/support/libtool/buildroot-libtool-v2.4.4.patch
  to be applied  : buildroot/support/libtool/buildroot-libtool-v2.4.4.patch

When a package uses AUTORECONF, the libtool patch can be applied a
second time on the same ltmain.sh in the same directory as the
<package>-reconfigure target is called. This is not a problem since
autoreconf will overwrite the previously patched files. But with the
check introduced by [1], apply-patches.sh throw an error when the
package is reconfigured since the libtool patch is already present in
.applied_patches_list.

In addition to this, the .applied_patches_list file generated by
apply-patches.sh while patching ltmain.sh is not in the top-level
package source directory. Instead a duplicated .applied_patches_list
is generated beside the patched ltmain.sh which can be in a
subdirectory.

As noticed by Arnout Vandecappelle [2], the apply-patches.sh script
doesn't bring anything for the libtool patches.

"apply_patches.sh does the following:

* It handles directories -> not needed here.
* It handles compressed patches and tarballs -> not needed.
* It handles series files -> not needed.
* It handles errors in case of multiple patches -> not needed since
it's only one patch.
* It detects errors based on *.rej files -> not needed since it's only
a single patch so patch exit code is OK.
* It writes the patch list -> for libtool, this is quite silly because
it will be written in the directory where ltmain.sh is found, not in the
top-level directory, so you have these patch lists spread over the
source tree."

So use patch directly rather than apply-patches.

[1] 19241598147e7555dce40b6dd44b28ef22b67ed9
[2] http://lists.busybox.net/pipermail/buildroot/2016-August/169810.html

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Tested-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

squeezelite: needs NPTL

It uses PTHREAD_PRIO_INHERIT, which isn't available in Linuxthreads.

Fixes:
  http://autobuild.buildroot.net/results/455465c332d291929c1612d18a9e7448d9c76e73/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
[Thomas: drop comment about the NPTL dependency being related to
alsa-lib, as it's not the case. Suggested by Baruch.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libconfuse: add optional dependency on gettext

When available, libconfuse will use the libintl provided by the gettext
package.

Signed-off-by: Rahul Bedarkar <rahul.bedarkar@imgtec.com>
[Thomas: make it really an optional dependency.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

nettle: add security patch

Fixes:
CVE-2016-6489 - RSA code is vulnerable to cache sharing related attacks.

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

package/lldpd: explicit optional dependencies

check and valgrind are optional depenencies of lldpd, so we must depend
on them when either is enabled.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

pkg-cmake.mk: set CMAKE_SYSROOT variable in toolchainfile.cmake

This change enforces the CMAKE_SYSROOT value set in the toolchainfile.cmake.

This fix overrides the CMake heuristics used to guess it, and turns off some
non-desirable behavior adding "-isystem ..." flags to the compiler command
line, misleading the compiler and making the build failed due to some
unfound standard headers.

Fixes:
  http://autobuild.buildroot.net/results/f7e/f7e92678e91a6cb15ccf32d4a7d75b39f49d6000/defconfig
  (and others)

Cc: Ben Boeckel <mathstuf@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ulogd: fix build with musl

Add upstream patches to fix build issue with musl.

Fixes:

  http://autobuild.buildroot.net/results/c42/c42db6f8c89df87e9b55d6c7aaf6098d64d8ffdb/

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

gnupg: security bump to version 1.4.21

Fixes CVE-2016-6313: An attacker who obtains 580 bytes from the standard RNG
can trivially predict the next 20 bytes of output.

Add cryptographically secure sha256 hash.

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

libgcrypt: security bump to version to version 1.7.3

Fixes CVE-2016-6316: Bug in the mixing functions of Libgcrypt's random number
generator. An attacker who obtains 4640 bits from the RNG can trivially
predict the next 160 bits of output.

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

stella: fix compiler search in configure script

In configure script there are some tests that are looking for
a suitable compiler. For this purpose script tries to compile
and link some C++ code that is hardcoded in the body of script.

The problem is that only linker flags ($LDFLAGS) are passed to the
compiler when the script is compiling test code. Therefore some
necessary command line options can be lost and this would lead to
errors.

This exactly happens to ARC:
  1. -matomic option is lost as it compiler option,
  2. test code compilation fails,
  3. the script decides not to use arc-buildroot-linux-uclibc-g++,
  4. scipt checks host compiler (x86_64-linux-gnu-g++ in your case),
  5. package is being built for x86_64, not for ARC.

About last 2 steps - I don't know why configure script should try
to build stella for host machine. As for me an error should be
raised and a build should be stopped. But this is up to stella
developers.

Current patch adds compiler options ($CXXFLAGS) when compiling
test code in configure script.

Fixes stella build for ARC, e. g.:
http://autobuild.buildroot.net/results/c23/c23d655137b1c1825b1da69b18307c6a1d4b23b8//

[Peter: s/CPPFLAGS/CXXFLAGS/]
Signed-off-by: Vlad Zakharov <vzakhar@synopsys.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

busybox: add upstream fixes for 1.25.0

Add upstream patches to fix gzip (bug #9131 - incorrect compression
levels), ip (route add bug) and ntpd (CVE-2016-6301 - NTP server denial
of service).

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

Update for 2016.08-rc2

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

website/news.html: fix 2016.08-rc1 release date

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

website/news.html: add 2016.08-rc1 announcement link

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

jasper: add security patches

Fixes:
CVE-2016-2116 - Memory leak in jas_iccprof_createfrombuf causing
memory consumption.
CVE-2016-1577 - Double free vulnerability in jas_iccattrval_destroy.
CVE-2016-1867 - out-of-bounds read in the jpc_pi_nextcprl() function.
CVE-2015-5221 - Use-after-free and double-free flaws in Jasper
JPEG-2000 library.
CVE-2015-5203 - double free in jasper_image_stop_load()

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

wayland: fix musl build issue

With musl C library, we see following build failure.

  src/scanner.c: In function 'find_enumeration':
  src/scanner.c:811:2: error: unknown type name 'uint'
    uint idx = 0, j;
    ^

uint is defined in <sys/types.h> header. It is usually included by <stdlib.h>
but in musl stdlib.h doesn't include it.

This patch adds upstream patch where uint32_t is used instead of uint.

Fixes:

  http://autobuild.buildroot.net/results/fcf/fcf6dd3bd31f92eadbcae17338e5887a23e43ff9/

Cc: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Rahul Bedarkar <rahul.bedarkar@imgtec.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

mtd: fix musl build issue

With musl C library, we get following build error

  integck.c:37:28: fatal error: bits/stdio_lim.h: No such file or directory
   #include <bits/stdio_lim.h>
                              ^
  compilation terminated.
  make[2]: *** [integck] Error 1

Header <bits/stdio_lim.h> is not available in musl C library. However
<stdio.h> has all definition that <bits/stdio_lim.h> supposed to be
providing. Moreover <bits/stdio_lim.h> shouldn't be included directly
instead we should be using <stdio.h>.

Since we already include <stdio.h> and in case of uClibc or glibc
<bits/stdio_lim.h> gets included internally, we can safely remove it.

Fixes:

  http://autobuild.buildroot.net/results/175/1754861457af520480cc34d7d2d0edff2868ff66/

Signed-off-by: Rahul Bedarkar <rahul.bedarkar@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/qt: webkit needs gst-plugins-base

QtWebkit needs gstreamer-0.10 and the 'app' plugins, so we depend on
gstreamer and select the required plugins.

We can't select gstreamer as that would cause circular dependencies in
kconfig:

    package/kodi/Config.in:37:error: recursive dependency detected!
    package/kodi/Config.in:37:      symbol BR2_PACKAGE_KODI depends on BR2_PACKAGE_KODI_EGL_GLES
    package/kodi/Config.in:16:      symbol BR2_PACKAGE_KODI_EGL_GLES depends on BR2_PACKAGE_HAS_LIBEGL
    package/opengl/libegl/Config.in:1:      symbol BR2_PACKAGE_HAS_LIBEGL is selected by BR2_PACKAGE_MESA3D_OPENGL_EGL
    package/mesa3d/Config.in:150:   symbol BR2_PACKAGE_MESA3D_OPENGL_EGL is selected by BR2_PACKAGE_MESA3D_GALLIUM_DRIVER_VIRGL
    package/mesa3d/Config.in:92:    symbol BR2_PACKAGE_MESA3D_GALLIUM_DRIVER_VIRGL depends on BR2_PACKAGE_MESA3D
    package/mesa3d/Config.in:1:     symbol BR2_PACKAGE_MESA3D is selected by BR2_PACKAGE_XDRIVER_XF86_VIDEO_IMX_VIV
    package/x11r7/xdriver_xf86-video-imx-viv/Config.in:1:   symbol BR2_PACKAGE_XDRIVER_XF86_VIDEO_IMX_VIV depends on BR2_PACKAGE_IMX_GPU_VIV_OUTPUT_X11
    package/freescale-imx/imx-gpu-viv/Config.in:34: symbol BR2_PACKAGE_IMX_GPU_VIV_OUTPUT_X11 is part of choice <choice>
    package/freescale-imx/imx-gpu-viv/Config.in:27: choice <choice> contains symbol <choice>
    package/freescale-imx/imx-gpu-viv/Config.in:27: choice <choice> contains symbol BR2_PACKAGE_FREESCALE_IMX
    package/freescale-imx/Config.in:1:      symbol BR2_PACKAGE_FREESCALE_IMX is selected by BR2_PACKAGE_GST_FSL_PLUGINS
    package/gstreamer/gst-fsl-plugins/Config.in:8:  symbol BR2_PACKAGE_GST_FSL_PLUGINS depends on BR2_PACKAGE_GSTREAMER
    package/gstreamer/gstreamer/Config.in:5:        symbol BR2_PACKAGE_GSTREAMER is selected by BR2_PACKAGE_QT_WEBKIT
    package/qt/Config.in:464:       symbol BR2_PACKAGE_QT_WEBKIT depends on BR2_PACKAGE_QT
    package/qt/Config.in:5: symbol BR2_PACKAGE_QT is selected by BR2_PACKAGE_PYTHON_PYQT
    package/python-pyqt/Config.in:5:        symbol BR2_PACKAGE_PYTHON_PYQT depends on BR2_PACKAGE_PYTHON
    package/python/Config.in:5:     symbol BR2_PACKAGE_PYTHON is selected by BR2_PACKAGE_KODI

This one is fairly complicated to fix properly, so it is easier to just
depend on gstreamer.

Also rework the comment when QtWebkit is not available.

Fixes:
    http://autobuild.buildroot.org/results/84d/84d338b70c8ce3f21dd202de387326f18526aa63/
    http://autobuild.buildroot.org/results/379/379ad4eadf5af7e1d5b977bc229d3fcaba25b4fe/
    [...]

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

linux: bump defaut to version 4.7.1

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

linux-headers: bump 3.14.x and 4.{4, 6, 7}.x series

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

package/libsidplay2: Fix build with gcc6

Fixes
http://autobuild.buildroot.net/results/787/787c715814b09d380ca521712f54cbcf5ae121d6/

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

ffmpeg: don't pass --enable-mipsfpu on non-MIPS platform

The current logic to pass the --{enable,disable}-mipsfpu option is:

ifeq ($(BR2_MIPS_SOFT_FLOAT),y)
FFMPEG_CONF_OPTS += --disable-mipsfpu
else
FFMPEG_CONF_OPTS += --enable-mipsfpu
endif

In practice, this means that on MIPS soft-float, --disable-mipsfpu is
passed, and that in *all* other cases, --enable-mipsfpu is passed,
including if we are *not* targetting the MIPS architecture.

Even though this doesn't seem to cause any problem, it is a bit weird to
see --enable-mipsfpu when you're building ffmpeg for a non-MIPS
architecture, so we better fix this by enclosing the MIPS-related
options in a MIPS condition.

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

ffmpeg: explicitly disable NEON support

When BR2_ARM_CPU_HAS_NEON=y, we pass --enable-neon. However, when
BR2_ARM_CPU_HAS_NEON is disabled, we don't pass anything. This generally
works fine, but turned out to integrate NEON code in ARMv7-M
builds (since it's ARMv7, ffmpeg assumed it should enable NEON code).

Even though ffmpeg is now disabled for ARMv7-M, it still makes sense to
be explicit, and disable NEON support when the CPU doesn't have it.

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

spidev_test: fix musl build

spidev.h needs _IOC_SIZEBITS. Add linux/ioctl.h for that to fix build under
musl.

The patch suggested upstream (https://patchwork.kernel.org/patch/9276909/)
does not apply to the spidev_test.c version we currently use. A backported
patch would break once we bump spidev_test.c. A bump+patch solution would
break if/when upstream accepts this patch. That's why this commit uses the
unusual sed patching to keep compatibility with current and future
spidev_test.c versions.

Fixes:
http://autobuild.buildroot.net/results/128/1284a36c70ff8a1412e2128266c6654b2aa5b2da/
http://autobuild.buildroot.net/results/0ad/0adea6121b238adf0e64d455a7573b1a31c38264/
http://autobuild.buildroot.net/results/21b/21b92651ce456f3819eef315cecd92ff70b9e98f/

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

package/sg3_utils: fix uclinux build

sg3_utils grep for '^linux' as the host OS to decide if we are
targetting Linux.

However, on noMMU targets, the host OS can be 'uclinux' instead of just
plain 'linux', which make sg3_util confused and thus it forgets to
include some source files in the build, resulting in link failures.

Fix that by ammending the regexp that recognises Linux. We now need to
run autoreconf.

Fixes:
    http://autobuild.buildroot.org/results/79a/79a572f67bafa6a007256a5f22529c2daabd5733/
    http://autobuild.buildroot.org/results/4f7/4f7ea4bb4c4c12ceb9cd08bfb6720386f5bdea0b/

[Peter: fix typos as pointed out by Arnout]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/quota: fix static build with e2fsprogs

quota does not use pkg-config to find e2fsprogs libs, so it misses out
on the fact that it may require -pthread when doing static links.

Fix that by manually adding -pthread in case of the static build (of
course, only if the toolchain has threads).

Fixes:
    http://autobuild.buildroot.org/results/733/733145e94d7c4f3b11e867c60a1e8f8e049a3ce0/
    http://autobuild.buildroot.org/results/666/6666f7c07339830fb45a5decb8825560ee432ce5/
    http://autobuild.buildroot.org/results/f5b/f5b1d57bd26adaf125555b97ff1eb9faf223a293/
    [...]

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

package/axel: fix link order

libintl may use thread-related functions, so -lpthread must come after
-lintl.

Fixes:
    http://autobuild.buildroot.org/results/e9d/e9d89fd5b24a76a3c5ab9dc8f01f4643f7c13396/

[Peter: use -lpthread in comment as suggested by Arnout]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux: fix comment about version choice

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

barebox: fix ARCH value for arm64

barebox 2016.08 added arm64 support using ARCH=arm.

[Peter: extend commit message to clarify]
Signed-off-by: Raphael Poggi <poggi.raph@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

uclibc: add upstream patch for CVE-2016-4429

Add security related bugfix to fix CVE-2016-4429 from
GNU libc project.

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

poco: fix m68k compile

Add a patch from double-conversion upstream project which is bundled
with poco to fix compile for m68k.

Fixes:
  http://autobuild.buildroot.net/results/791917c1343f53e12f27b06cfd5e1d2e88078014/
  http://autobuild.buildroot.net/results/349d4b49d65a652551c4de20233684bb4ffa5ae3/

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

gmp: disable assembly for arc

It has different inline assembly constraints that
are not supported after update to gcc-6.

Fixes:
http://autobuild.buildroot.net/results/c91/c9143eec0834017d57d5352d6fc95b5fae38da00//
and a lot of other failures of gmp-6.1.1 for ARC.

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

xen: Ensure host Python2 is used

To avoid Xen build issues ensure that host Python 2 is used.

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

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

gcc-6 has enabled more warnings by default, and turned some existing
warnings into errors.

Fixes (all on arc or arm):
    http://autobuild.buildroot.org/results/8c4/8c4a6cf492bccf4e1b99897bd9658e21b89b31ab/
    http://autobuild.buildroot.org/results/875/875a9deb15d47d348bb53dafdbb7ed2f2bbd3da8/
    http://autobuild.buildroot.org/results/af4/af4935f0ad2054b7dc2443e2d126fe8e50996f66/
    [...]

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

system: sysvinit needs MMU

It uses spawn() and thus fork(), so it needs an MMU.

Fixes a build issue reported on IRC for a cortex-m4 build:
    http://pastebin.com/dGCsy0sr

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

package/weston: fix patch

The newly iontroduced patch was created against weston's master branch,
but was not rebased correctly on top of the release we are using, hence
it does not apply because files have been moved since the release.

Fixes:
    http://autobuild.buildroot.org/results/5fb/5fbfd26d05a0ccaf4a70a100a7499cedea83ce6f/
    http://autobuild.buildroot.org/results/769/76976fbfc2162c2c671a52eafdad82df6d25a660/

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

package/systemd-bootchart: don't build manpages

Building the man pages requires xsltproc, which is not always available
on the build machine, so the build fails at configure:
    configure: error: *** xsltproc is required for man pages

But then, even when xsltproc is present, it can't actually build the man
pages, because it can't "download external entities":
    I/O error : Attempt to load network entity http://docbook.sourceforge.net/release/xsl/current/manpages/docbook.xsl
    warning: failed to load external entity "http://docbook.sourceforge.net/release/xsl/current/manpages/docbook.xsl"
    compilation error: file ./man/custom-man.xsl line 27 element import

Add a patch (sent upstream) to allow not building the man pages.

Fixes:
    http://autobuild.buildroot.org/results/e2a/e2ae38775b6acff3c3a8ea233ec70dd68b017319/
    http://autobuild.buildroot.org/results/66d/66d85622db9f7ff5a4e8d6eed3fd3b32959685cf/

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

package/systemd-bootchart: needs headers >= 3.17

It uses GRND_NONBLOCK, introduced in 3.17.

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

toolchain: Bump ARC tools to arc-2016.09-eng008

As described at:
4520524ba055706236db9f00dd79f1b2e2e87fde
this commit continues a series of updates of ARC tools.
This time we're updating tools to arc-2016.09-eng008.

Main updates were made for gcc. It was switched to GCC 6 and to
OSABI v4.

Besides this patch fixes buildroot ARC failures connected to
"crtbeginT.o" object file missing. This issue lead to two main errors:

1) "crtbeginT.o: No such file or directory", e. g. bootutils-1.0.0.
No comments are required here I hope.

2) Errors like "compiler cannot create executables", e.g.:
  a) host-gcc-final-arc-2016.09-eng007 static build,
  b) aespipe-2.4c.
That was caused because the test to determine if compiler is able to
create executables was failing due to missing "crtbeginT.o" file.

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

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

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

uboot-tools: add missing !BR2_STATIC_LIBS dependency for FIT signature

BR2_PACKAGE_UBOOT_TOOLS_FIT_SUPPORT selects BR2_PACKAGE_DTC, but forgets
to depends on !BR2_STATIC_LIBS like BR2_PACKAGE_DTC does.

Fixes the following kconfig warning:

  arning: (BR2_PACKAGE_UBOOT_TOOLS_FIT_SUPPORT && BR2_PACKAGE_KVMTOOL && BR2_PACKAGE_UBOOT_TOOLS_MKIMAGE_FIT_SIGNATURE_SUPPORT) selects BR2_PACKAGE_DTC which has unmet direct dependencies (!BR2_STATIC_LIBS)

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

Revert "blktrace: disable on PowerPC"

This reverts commit 0d8158fc619ce849b531d07b27da4ba9748d4ea4. Thanks to
a fix in libaio, blktrace now builds fine on PowerPC, so there's no
reason to disable it anymore.

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

libaio: work-around for PowerPC issue

Both the blktrace and gadgetfs-test packages were failing to build on
PowerPC due to the mysterious:

  hidden symbol `_rest32gpr_30_x' in libgcc.a(e500crtresx32gpr.o) is referenced by DSO

Due to this error, we disabled blktrace on PowerPC in commit
0d8158fc619ce849b531d07b27da4ba9748d4ea4. However, gadgetfs-test
continued to fail with the same error. As Romain Naour pointed out, the
problem seems in fact to come from a common dependency of blktrace and
gadgetfs-test: libaio. As Romain investigated, the problem started
appearing after the last bump of libaio, from version 0.3.109 to
0.3.110.

A quick bisect through the libaio changes between 0.3.109 and 0.3.110
has revealed that the problematic change is one in the libaio build
system, which now obeys to the CFLAGS provided in the environment,
rather than overriding them. So the CFLAGS provided by Buildroot cause
this problem. It turns out that the problematic CFLAGS is -Os, which is
indeed known to cause issues on PowerPC in some corner cases.

Even though it would probably be a better long-term solution to switch
to -O2 by default, and mark -Os as not available on PowerPC, it is a too
radical change so close to 2016.08. So we simply adjust the libaio
package so that it uses -O2 instead of -Os.

Fixes:

  http://autobuild.buildroot.net/results/5e6cc4c432ce6c964ac285026978ad14d9eae97c/

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

ffmpeg: disable for ARMv7-M

In the ffmpeg code, libavutil/arm/asm.S makes some assumption about the
ARM architecture for which the code is built. Only ARMv4, ARMv5, ARMv6
and ARMv7-A is supported. Due to this, object files built out of
ARM-optimized assembly code have the wrong architecture information,
causing a failure at link time.

Adding ARMv7-M support would be possible, but it doesn't exist yet in
ffmpeg, and it's pretty unlikely that ffmpeg will ever be needed on an
ARMv7-M platform, so this commit takes the simple approach of disabling
ffmpeg for ARMv7-M.

Fixes:

  http://autobuild.buildroot.net/results/ca4c67b093afd6f14349fcdc87b02e0480172e8c/

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

mtd: integck needs MMU support

After fixing build issue related to MS_DIRSYNC and MS_RELATIME, we get
following build error with blackfin external toolchain.

fstest_monitor.c: In function 'main':
fstest_monitor.c:237: warning: implicit declaration of function 'fork'
/tmp/ccCzZIuW.o: In function 'main':
/home/rahul.bedarkar/buildroot/output/build/mtd-1.5.2/tests/fs-tests/utils/fstest_monitor.c:237: undefined reference to '_fork'
collect2: ld returned 1 exit status
make[2]: *** [fstest_monitor] Error 1

Since it uses fork(), add dependency on MMU support for integck.

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

libical: use system time zone info

libical requires time zone information in order to make use of time
zone functions (otherwise, segfaults occur when applications call
the time zone functions). This patch:
1. forces the system to build time zone info
2. ensures libical does not build its own time zone info (this is
   the default behaviour, but the flag is added in case libical's
   default changes in the future).

Signed-off-by: Marc Khouri <marc@khouri.ca>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libxml2: fix m68k uclinux compile

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

Fixes:
  http://autobuild.buildroot.net/results/8a4e26bb1c0c64759e3914eced0a36baa0967f0b/

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

package/kmsxx: bump version for portability fix with musl

Bump the version to get two portability patches (as well as a small,
innocuously-looking minor feature).

Fixes:
    http://autobuild.buildroot.net/results/682/68266cb5b26a62387dc99aef31fc9654c8fcd505/
    [and quite a few others...]

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

armadillo: propagate clapack dependency

Commit 7b75d8c3fa7b234c948ff37c4f14467900083036 added a new dependency
to the clapack package, but forgot to propagate it to the reverse
dependencies, which in this case are just the armadillo package. This
commit fixes that.

Fixes:

  http://autobuild.buildroot.net/results/f34ad4f0cf094fa31b7dc18522e9eedfc5c2c940/

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

xserver_xorg-server: move existing patch, and add to 1.17.4

As reported by Manuel in bug #9161, the commit
19b503b0d26fa16fffc980c287ca844077f3f675 that bumped the X.org server
from 1.18.3 to 1.18.4 forgot to also rename the patch directory, making
the existing patch orphan, while it is still needed. This commit
therefore renames the 1.18.3 directory to 1.18.4.

Moreover, this patch is also needed for the 1.17.x X.org server series
which has been added recently, so this commit also adds the patch to a
new 1.17.4 directory.

Fixes bug #9161.

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

gcc: disable C++ support for Blackfin

As discussed with Waldemar, the C++ support for Blackfin is currently
broken, and we don't have a fix in sight for the 2016.08
release. Therefore, this commit disables C++ support entirely on the
Blackfin architecture in the internal toolchain backend.

This will avoid a significant number of Blackfin build failures, that
occur when building C++ packages.

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

lftp: fix static link with readline

When readline is static library, we need to link against ncurses,
it is because, dependent library's symbols are not resolved when
static library is built. Those symbols are resolved when binary tries
to link with static library.

We can't pass linker flags for ncurses by setting LIBS environment
variable via <PKG>_CONF_ENV because it looks like build system is not
taking that into account and even though it would have been, order of
linking is important.

We can't also pass linker flags for ncurses via --with-readline-libs
conf options because it causes lftp_LIB_READLINE macro to take readline
headers from host machine if available. To use --with-readline-libs
we need to set --with-readline=yes and --with-readline-inc to include
dir. But when --with-readline=yes, readline_prefix is computed based
on if headers can be found in /usr/local or /usr. If readline is
installed on host machine, then configure fails since we are using
headers for host machine. If headers are not found in /usr/local or /usr
then only path specified --with-readline-inc is taken into account.
So specifying linker flags for ncurses via --with-readline-libs will
not work in all cases.

So it looks like, updating linker flags directly is only option to fix
static link issue against readline.

Fixes:
http://autobuild.buildroot.net/results/ae9/ae93bd1b4f6e1ed8f386830d4a51f00d9466f267/

Signed-off-by: Rahul Bedarkar <rahul.bedarkar@imgtec.com>
[Thomas: update autoreconf comment in .mk file.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

lftp: remove unneeded config options

When installation prefix is specified in "--with-readline" config
option, lftp's build systems figures out include dir and linker flags.
So don't specify --with-readline-{inc,libs} options which are anyways
get ignored by build system.

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

openipmi: needs NPTL support

openipmi uses the pthread_condattr_setclock() function, which is only
available with the NPTL thread implementation.

Fixes:

  http://autobuild.buildroot.net/results/fc4084f8d1ff3fe98066c7cc051c2127bbe77ec5/

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

php: add patch to fix static library build

This commit adds a patch to PHP which fixes the build for
BR2_STATIC_LIBS=y configurations.

Fixes:

  http://autobuild.buildroot.net/results/885f418cc26c6832a8a9abf703aecbeea8fbe550/

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

php: rework patches as Git formatted patches

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

flex: add security patch

Fixes:
CVE-2016-6354 - buffer overflow in generated code (yy_get_next_buffer).

Patch status: upstream backport.

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

linux-headers: bump 3.{14, 18}.x and 4.{1, 4, 6}.x series

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

fontconfig: security bump to version 2.12.1

Fixes:
CVE-2016-5384 - possible double free due to insufficiently validated
cache files.

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

system: zoneinfo usable by musl, too

Currently, we limit installing zoneinfo to non-musl toolchain, by lack
of knowledge on how it would work on musl.

Turns out that musl uses the same zoneinfo format as glibc does.

Make it possible to install the TZ info whatever the C library; for
musl, use tzdata as for glibc.

Thanks Rich! ;-)

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Rich Felker <dalias@aerifal.cx>
Cc: Marc Khouri <marc@khouri.ca>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/weston: no-break space U+A0 (U8+C2A0)

Fixes:
    http://autobuild.buildroot.org/results/f49/f49a9cbb7bdc5d9e05dcf0a20bd83f059e234e74/
    http://autobuild.buildroot.org/results/1d3/1d3e7f07388ae11d70103e04d8a14c20d50fc212/
    [...]

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

openblas: fix build issues on ARM

This commit fixes several build issues of OpenBLAS on ARM:

 - The first one occured on ARMv5 platforms, when the ARMV5 OpenBLAS
   architecture is used. In this case, OpenBLAS build system forces
   -march=armv5, which may not be correct for certain toolchains. As an
   example, the Sourcery CodeBench toolchain has an ARMv4 and an ARMv5
   sysroot. The ARMv5 sysroot is actually an armv5te sysroot, so when
   OpenBLAS forces -march=armv5, gcc thinks it should use the ARMv4
   sysroot, causing build failures.

   To address this, a patch to completely remove the -march ARM CFLAGS
   is added to OpenBLAS.

   Fixes:

     http://autobuild.buildroot.net/results/991497b12b70f948169e5ad99eebd0fe7f6209a2/

 - The second one occured on ARMv7 platforms, when the ARMV7 OpenBLAS
   architecture is used. The OpenBLAS code expects an EABIhf build, so a
   dependency is added for EABIhf for both ARMv6 and ARMv7.

   Fixes:

     http://autobuild.buildroot.net/results/0ba0bee48a83367fcefab827e8eaa72f0c8fe90b/

 - Once the previous ARMv7 problem has been fixed, it turns out that the
   ARMv7 specific code in OpenBLAS contains VFPv3 specific
   code. Therefore, the user *must* have choosen either VFPv3 or VFPv4,
   or the code will not build. VFPv3-D16/VFPv4-D16 are not sufficient,
   as more than 16 registers are used by the OpenBLAS code.

   To address this, the ARMV7 platform of OpenBLAS is restricted to the
   proper VFPv3/VFPv4 selection, and the ARMV6 platform is restricted to
   the proper VFPv2 selection.

   This problem was not visible in the autobuilders, as it was hidden by
   the previous one.

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

mpv: fix build on sparc

This commit adds a new patch for mpv that fixes the stdatomic detection
logic so that it realizes linking against libatomic is needed on some
architectures. This fixes the build of mpv on SPARC. The patch has been
submitted upstream.

Fixes:

  http://autobuild.buildroot.net/results/f607eed9a48c4ca00db43f9d2652e615b4bd7c93/

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

libsigsegv: make the architecture dependencies more robust

While musl and glibc provide the ucontext structure definition on all
architectures they support, it is not the case of uClibc. We have been
excluding more and more architectures over time, but this isn't really
nice as we continuously need to add more. Latest example: the fact that
ucontext on ARM is only available on ARM platforms that support the ARM
instruction set (i.e not on the Thumb-2 only Cortex-M platforms).

Therefore, we introduce a BR2_PACKAGE_LIBSIGSEGV_ARCH_SUPPORTS variable,
which:

 - is set to 'y' for musl and glibc

 - is set to 'y' for uClibc for the only architectures that are known to
   provide ucontext (those are the ones that select ARCH_HAS_UCONTEXT in
   uClibc.

Note that we remove the comment, as it would become a way too
complicated comment, and we usually don't add comment for such
complicated situations (see libunwind for a similar example).

Fixes:

  http://autobuild.buildroot.net/results/b1e40b3ec64cd98b535d83e89e5780c46680e095/

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

linux-zigbee: fix build on musl

This commit adds a patch to the linux-zigbee package that fixes the
build on musl by adding a missing <time.h> include.

Fixes:

  http://autobuild.buildroot.net/results/5d56f998b5c5ab9e06cf048e6ec95b8671989cef/

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

dante: fix build on some PowerPC platforms

The dante configure script has some logic to remove the -g flag from
CFLAGS, but this logic is flawed and also removed -g in
-mfloat-gprs=double, turning it into the invalid -mfloatprs=double,
causing a build failure.

This commit adds a patch that gets rid of this -g removal logic, since
it is in fact unnecessary.

Fixes:

  http://autobuild.buildroot.net/results/313370bf05efe7fd87c281a97ecb6e06531a87ed/

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

procps-ng: fix BR2_STATIC_LIBS=y build

The procps-ng package currently fails to build on BR2_STATIC_LIBS=y
configurations. Indeed, by default NUMA support is enabled, and it
requires dlopen(). As hinted by the configure script, passing
--disable-numa allows to disable NUMA support.

However, once this is done, another issue pops up: dlopen() is also used
by the SELinux support. But even when SELinux support is disabled, the
procps-ng code incorrectly includes <dlfcn.h>. This is addressed by the
addition of a patch.

Fixes:

  http://autobuild.buildroot.net/results/b385bf435085728aece6323a5006ba9fa6631744/

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

webrtc-audio-processing: fix build with musl

webrtc-audio-processing assumes that execinfo.h is available when
__UCLIBC__ is not defined, which is an incorrect assumption, and
specifically not valid for musl.

This commit adds a patch that moves to autoconf based checks for
cxxabi.h and execinfo.h, which solves the problem.

Fixes:

  http://autobuild.buildroot.net/results/480f529a53b321344e499f958be8e5d1d0f58646/

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

webrtc-audio-processing: fix build on ARM Cortex-M

The webrtc-audio-processing configure script assumes that if the
host_cpu part of the tuple is "arm", then ARM instructions are
available. This is obviously incorrect for ARM Cortex-M platforms, which
only support the Thumb-2 instruction set.

In order to address this, we add a patch,
0001-configure.ac-fix-architecture-detection.patch, which changes how
webrtc-audio-processing detects the architecture: instead of relying on
the host_cpu part of the tuple, it relies on the built-in definitions of
the compiler.

Not only it fixes the Cortex-M detection, but it also enables ARMv7
optimizations on ARMv7-A: until now they were only enabled when the
host_cpu part of the tuple was armv7, which is never the case in
Buildroot.

However, once this issue is fixed for Cortex-M, the build nonetheless
fails later due to the usage of NPTL-only functions. So we change the
thread dependency to a NPTL dependency.

Fixes:

  http://autobuild.buildroot.net/results/4933082cd7cc5781404c77ccef5c2b9333c5f714/

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

docs/manual: fix patch naming convention

Otherwise conflicts with the naming convention from previous section.

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

gnuradio: backport patch to fix gr-fec compilation

Fix:
http://autobuild.buildroot.net/results/f14/f14f4f62b5f23c652fa9411d9c569c2ab3394fb5

Signed-off-by: Gwenhael Goavec-Merou <gwenhael.goavec-merou@trabucayre.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gst1-imx: fix V4L2 plugin build with kernel headers < 3.18

This patch fix the following build issue:
http://autobuild.buildroot.net/results/b46/b460a770c8f4e992d29dde8fe37fc23a949937f2

It has been submitted to the package maintainers:
https://github.com/Freescale/gstreamer-imx/pull/106

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

size-stats: fix bug when parsing file names with comma

The size-stats script fails when the usb_modeswitch_data is enabled,
because this package installs files that contain commas in their
name. However, the size-stats script also uses comma as a separator for
its CSV files, causing a "ValueError: too many values to unpack" in:

  pkg, fpath = l.split(",")

Fix this by splitting only the two fields that need to be split.

The bug was reported by Matthias <porto.rio@gmx.net>, who also suggested
a fix.

Fixes bug #9136.

Reported-by: Matthias <porto.rio@gmx.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

qt: Fix missing runtime Qt3Support dependency

Qt3Support module has a runtime dependency on QtSql, see
https://wiki.qt.io/Qt_Library_Cross_Dependencies. This patch fix this
issue.

Signed-off-by: Alexander Shiyan <shc_work@mail.ru>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

qt: fix build with ALSA >= 1.1.x

ALSA > 1.1.x are not determined correctly when configuring the library.
A patch, identical to the one used for Qt5, is added to the qt package
to solve this problem.

Signed-off-by: Alexander Shiyan <shc_work@mail.ru>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

owfs: disable PHP support

Since we upgraded PHP to PHP 7.x, the PHP support of OWFS started
failing to build. Since we can hardly fix that on our own, the issue was
reported upstream (https://sourceforge.net/p/owfs/support-requests/32/)
and for now, we will disable PHP support in OWFS.

Fixes:

  http://autobuild.buildroot.net/results/fe6843ec53fbed452d2e38a8577a235f73abb8db/

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>

netplug: add two patches to fix musl build issues

This commit adds two patches to netplug that are needed for this package
to build with musl:

 - One patch to add a missing header inclusion
 - One patch to fix a conflicting prototype on __assert_fail().

Fixes:

  http://autobuild.buildroot.net/results/c2a5dfedba46cc8eb3d0c5c43f1f76fe0bb1805f/

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>

thrift: fix m68k uclinux compile

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

The failures happens after gnuplot compiles.

Fixes:
  http://autobuild.buildroot.net/results/d1e030c42f43f49575f36ecc82e9dd9018ca4f73/

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

gnuplot: fix m68k uclinux compile

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

Fixes:
  http://autobuild.buildroot.net/results/d1e030c42f43f49575f36ecc82e9dd9018ca4f73/

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

cairo: fix m68k uclinux compile

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

Fixes:
  http://autobuild.buildroot.net/results/31a488316fe3c14729a6dd9259205f8530684cab/
  http://autobuild.buildroot.net/results/dc6b899e3e502d65b2cf2f6d922a4f63857edf82/

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

package/uclibc: add upstream patch to fix Thumb2 builds

When uClibc-ng 1.0.17 was released, there was a regression when building
Thumb2-only for a CPU that is capable of running in arm mode (e.g. an
armv7a cpu).

We hastily added a patch to revert the upstream commit, as a stop-gap
measure, waiting for the actual fix.

That actual fix is there, now. :-)

Drop our revert-patch, and add the upstream patch.

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

package/bdwgc: fix pthread detection on uclinux platforms

bdwgc does not recognize "uclinux" as a valid OS part of the target
tuple which is used by some arm cortex-M toolchains.

Fixes:
http://autobuild.buildroot.net/results/94f/94fbc1e5afe183e5b071d1e869b2d780025389e2

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

iproute2: use @D instead of IPROUTE2_DIR

We more generally use $(@D) instead of $(pkg_DIR) to refer to the
package source directory, so this commit updates iproute2 to also use
$(@D) everywhere.

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

iproute2: fix no-thread build

iproute2 believes that it needs to link with libpthread for its arpd
binary, because "some db implementations require thread". Therefore, our
iproute2.mk explicitly disables the build of arpd when thread support is
not available.

However, the sed expression it uses no longer works. The Makefile used
to look like:

TARGETS = foo baz baz arpd foobar

so replacing " arpd " with a space was working fine. However, the
Makefile got changed in iproute2 to:

ifeq (... berkeleydb available ...)
     TARGETS += arpd
endif

i.e, with no space at the end of the line. This made our sed expression
ineffective, causing build issues with no-thread configurations since
arpd was no longer disabled.

To address this, instead of sed-ing the Makefile, we overwrite the
berkeleydb detection of iproute2, by writing to the "Config" file, like
we're doing for other aspects of the package.

Fixes:

  http://autobuild.buildroot.net/results/03a37a2372a4c2e438a073e015c49d9e554b86b7/

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>

uclibc: proper thread default on m68k and Microblaze

Traditionally, Buildroot has a default of enabling thread
support. However, with the current construct of the thread choice in the
uclibc package, the m68k and microblaze architecture end up with no
thread support as the default.

In order to avoid having to explicit a more complicated "default" value
for the choice, we take a simple approach: we order the 3 possible
choices by order of "preference", since Kconfig selects the first
selectable option in a choice by default.

So, NPTL is first and is the default when available. Then comes
linuxthreads which only gets selected as the default when NPTL is
available. None is offered as a last choice (in the current
implementation, it is never the default, since all architectures can
have thread support, either through NPTL or linuxthreads).

[Thomas: reworked according to Yann's comment that we could rely on the
Kconfig behavior that selects the first available choice option.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

am33x-cm3: available only for ELF builds

The am33x-cm3 build system does some objcopy based binary file
manipulation that assumes the compiler for the platform generates ELF
file. While it's true for most ARM platforms, and especially the AM335x
based ones, it is no longer always true since we added support for
Cortex-M. Therefore, this package needs to depend on BR2_BINFMT_ELF.

Fixes:

  http://autobuild.buildroot.net/results/05f46ab28cbfc587e2e031bf38fb975447f8bab6/

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

libunwind: fix architecture dependencies

Due to a misplaced parenthesis, the libunwind dependency were
incorrect. Indeed, they were of the form:

 default y if LIBC_TYPE && (arm dependency) || (list of other architectures)

Which meant that that the LIBC_TYPE dependency was not taken into
account for the "other architectures". This commit changes that to:

 default y if LIBC_TYPE && ((arm dependency) || list of other architectures)

Without this fix, libunwind can be selected for example on
PowerPC/uClibc configurations, while only PowerPC/glibc should be
allowed. This fixes:

  http://autobuild.buildroot.net/results/96c53ff240dcd68fcc9e3d32c4026c9886edcbe7/

Similarly, it also fixes cases where SuperH/uClibc was allowed, while
only SuperH/glibc should be allowed. This fixes:

  http://autobuild.buildroot.net/results/95a6cca21aea9914ba8bb3f571760fc054c53ecd/

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>

libcofi: disable for static-only builds

The libcofi package unconditionally builds a shared library, so we need
to make it unavailable on BR2_STATIC_LIBS=y configurations.

Fixes:

  http://autobuild.buildroot.net/results/329c09bf123cd5b6dbcfff6251ba7a2c7cce50b8/

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>

blktrace: disable on PowerPC

This package has been failing to build on PowerPC since March 2016, with
nobody taking care of the problem, so let's disable this package on
PowerPC for the time being. PowerPC is the only architecture on which
blktrace fails to build currently.

Fixes:

  http://autobuild.buildroot.net/results/5a673c692aeb957cd8beb2c2c10ce5fd882ba4b5/

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>

blktrace: add missing dependency on Config.in comment

The BR2_PACKAGE_BLKTRACE option "depends on
BR2_PACKAGE_LIBAIO_ARCH_SUPPORTS", but this architecture dependency was
not replicated in the Config.in comment. This commit fixes this
inconsistency.

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

ntp: disable debugging

When running ntp it randomly aborts at ntp-4.2.8p8/libntp/recvbuff.c:326
which seems to be a debugging feature. This patch just disables
debugging, it does not fix the root cause of the problem.

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

configs/odroidc2: enable console display

Redirect the console output to the screen and not only to the serial
port.

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