As indicated by the uClibc-ng source code, the NPTL thread
implementation is only available on MMU platforms, so we replicate
this dependency in Buildroot so that the appropriate thread
implementation is chosen by default on ARM noMMU.
Signed-off-by: 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>
As explained by Waldemar, enabling DOPIC in uClibc will lead to the
creation of a Position Independent library. In turn, this will cause
elf2flt to generate a "Has-PIC-GOT" flat binary, which doesn't work on
ARM. In fact, elf2flt on ARM really expect to have non-PIC code as
input, so we must disable DOPIC in the uClibc configuration.
Signed-off-by: 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>
Older gcc were not capable of building a uClibc library, with threads
enabled, in Thumb1. However, the issues have been fixed since gcc 4.9,
so this commit narrows down the condition to just gcc 4.7 and 4.8.
Signed-off-by: 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>
Now that uClibc is capable of figuring out by itself whether 'bx' can
be used or not, we can simplify the logic in Buildroot.
Signed-off-by: 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>
This commit adds a number of patches to uClibc that radically
simplifies the Thumb handling. uClibc currently has three options that
you need to toggle on Thumb configurations depending on the specific
ARM CPU being targeted.
However, it turns out that none of those options are necessary:
- USE_BX can simply be guessed by looking at the ARM core being
used. The bx instruction is available for all ARM cores >=
ARMv4T. This is exactly what glibc is doing.
- USE_LDREXSTREX can also be guessed by looking at the ARM core being
used: whenever you have Thumb2, ldrex/strex is available.
- COMPILE_IN_THUMB becomes useless, since all it does is passing
-mthumb. But just like the uClibc config options to set
--march=<foo> have been removed a long time ago, there's no need to
-have an option to pass -mthumb.
Signed-off-by: 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>
mesa3d can in fact build the DRI infrastructure/drivers just fine
without lingering x11 libs around, it just needs libdrm & friends which
are already accounted for.
So make the libGL (full OpenGL) providing abilities dependant on x11
being present.
It serves it's purpose for EGL+GLES hardware acceleration, and can in
fact be built without them, but it's probably not very useful (still,
leave the option for people inclined to do so).
Simple test defconfig:
BR2_arm=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_PACKAGE_MESA3D=y
BR2_PACKAGE_MESA3D_DRI_DRIVER_SWRAST=y
BR2_PACKAGE_MESA3D_DRI_DRIVER_NOUVEAU=y
BR2_PACKAGE_MESA3D_DRI_DRIVER_RADEON=y
BR2_PACKAGE_MESA3D_OPENGL_EGL=y
BR2_PACKAGE_MESA3D_OPENGL_ES=y
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Add more upstream commits to fix build errors with perl 5.10
introduced by the previous patch to fix build errors with perl 5.22.
Fixes
http://autobuild.buildroot.net/results/325/32519c6d4084f334b7fed9edfb8a8c68a1f840a8/
and many others.
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
[Thomas: reformat patches as proper Git formatted patches.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Ensure that each command line in the .config rule is prefixed by $$(Q)
Signed-off-by: Romain Izard <romain.izard.pro@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
New release fixes a number of bugs on AMI BMC firmware.
Signed-off-by: Vasant Hegde <hegdevasant@linux.vnet.ibm.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Removed patches applied upstream. This version bump is needed to
provide support for ffmpeg 3.0:
278747ba45
Use BR2_PACKAGE_FFMPEG_AVRESAMPLE by default because OpenCV can use it
after
2b2e3ee93c
Before this patch:
-- FFMPEG: YES
-- codec: YES (ver 57.24.103)
-- format: YES (ver 57.25.100)
-- util: YES (ver 55.17.103)
-- swscale: YES (ver 4.0.100)
-- resample: NO
After this patch:
-- FFMPEG: YES
-- codec: YES (ver 57.24.103)
-- format: YES (ver 57.25.100)
-- util: YES (ver 55.17.103)
-- swscale: YES (ver 4.0.100)
-- resample: YES (ver 3.0.0)
Disabled new options:
WITH_MATLAB, WITH_VA, WITH_VA_INTEL, DOWNLOAD_EXTERNAL_TEST_DATA
Removed obsolete option BUILD_opencv_hal
Removed options WITH_OPENGL & WITH_OPENMP from the "Disabled features"
block because they have their own ifeq-block later on.
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Linphone 3.6 is incompatible with bumped mediastreamer and ortp
packages. Like Debian we use the bundled versions instead and add the
Debian patch series to support ffmpeg 3.0
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The version bump is needed to provide support for ffmpeg 3.0.
Removed 0001-misc-fixes.patch which previously provided patches to
support newer ffmpeg versions. This is not needed anymore after
this upstream commit:
5e653bfe4c
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
This version bump is needed by mediastreamer 2.12.1, the mediastreamer
bump is needed to provide support for ffmpeg 3.0.
Removed srtp option after srtp support was removed from ortp:
5e37fa55f1
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Instead of installing login.pam (/etc/pam.d/login) from util-linux
conditionally do so unconditionally from linux-pam.
If busybox login is pam-enabled (linux-pam package enabled) it will be
required, otherwise it will lead to a system where login won't work, and
if util-linux is enabled it will supercede busybox login and will be
necesarry regardless.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The installer is a 32-bit executable, so we need the 32-bit libraries on
the host to be able to execute it.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
owserver is started unconditionally, but it needs the device(s) to
be specified at startup. Therefore a check for non-empty OWSERVER_ARGS
is added to the owserver start script.
owfs is started only if is built. It will connect to the owserver.
It will try to load the fuse module; if fuse is built-in or already
loaded, modprobe -q will fail silently. The filesystem will be mounted
on /dev/1wire, because it represents the 1-wire slave devices; also,
the bundled Debian init scripts mount it there. We have to create that
directory in the init script in case of devtmpfs, and we have to create
it in the rootfs for static /dev; for simplicity, just do both.
Users who want a setup without owserver need to remove the owserver
init script post-build and need to give appropriate defaults to owfs.
[Peter: fix mountpoint variable typo in S30owfs]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
If OpenGL is enabled then enable cairo-gl.
However if GLESv2 is also present we must choose which implementation to
use.
Since normally OpenGL is more featured, has more dependencies and
wouldn't be present "by accident" it seems like the logical choice to
win over it.
This feature is useful for webkitgtk's accelerated 2d canvas.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
A seq (midi) source has been added to gst-plugins-base 1.8.x hence we
need to account for it, so select BR2_PACKAGE_ALSA_LIB_SEQ. Fixes:
http://autobuild.buildroot.net/results/474/474e4b9bc3a44224302d0576cf1c0de7ce709f1a/
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Raptor is a free software / Open Source C library that provides a set of parsers
and serializers that generate Resource Description Framework (RDF) triples
by parsing syntaxes or serialize the triples into a syntax.
http://librdf.org/raptor/
Patch added in the package:
0001-configure.ac-remove-conditional-lib-inclusion.patch
Signed-off-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Nitin Mendiratta <nitin.mendiratta@rockwellcollins.com>
[Thomas:
- add else clause for the three optional dependencies, as suggested
by Baruch.
- fix they yajl optional dependency, as noticed by Baruch.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
All Qt5 packages gained the GPLv3 license option and all the documentation
is now under GFDLv1.3 license (except qt5websocket and qt53d)
Signed-off-by: Julien Corjon <corjon.j@ecagroup.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
This reverts commit 7bac31ceea.
This commit is no longer needed, since libdrm no longer needs arm >=
v6 support thanks to the fix to the CAS implementation.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The ARM patch that was added in commit
94841d87fd ("libdrm: fix support for
xserver with Vivante drivers") makes the assumption that ldrex/strex
is available on ARM. While this is true on ARMv6+, it is not true on
ARMv4/ARMv5. Due to this, we had to make follow-up commits like
7bac31ceea
("package/x11r7/xserver_xorg-server: dri needs arm >= v6") and more is
needed in the reverse dependencies to disable DRI usage on
ARMv4/ARMv5.
In fact, it turns out that the CAS code in libdrm is only an
optimization, and it will gracefully fall back to a system call based
lock/unlock mechanism.
So we simply change the patch to only provide the optimized CAS
implementation on ARMv6+. The original reason for this patch was to
fix Vivante drivers, and Vivante GPUs are only used in ARMv6+ cores,
so we should be fine (famous last words).
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Since 834893a965, pkg-config --libs libprocps
doesn't work anymore because libprocps.pc is installed in
STAGING_DIR/lib/pkgconfig/ instead of STAGING_DIR/usr/lib/pkgconfig/.
Add --libdir=/usr/lib in PROCPS_NG_CONF_OPTS to fixes the issue.
Note that libprocps.so is now installed in /usr/lib/ instead of /lib/.
Fixes:
http://autobuild.buildroot.net/results/2e5/2e5100ccf1ed336c26b83930e7149941c97fc4d2/
Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
To build shared libraries on SPARC64, -fPIC must be used instead of
just -fpic. This commit fixes the setools package, which allows to get
rid of the following autobuilder failure:
http://autobuild.buildroot.net/results/ba1e473178bb66ebc43897b20e534606d2577ee3/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The luajit Makefile has the following logic:
$(LDCONFIG) $(INSTALL_LIB) && \
$(SYMLINK) $(INSTALL_SONAME) $(INSTALL_SHORT1) && \
$(SYMLINK) $(INSTALL_SONAME) $(INSTALL_SHORT2) || :
So if ldconfig fails for some reason, the symbolic links to refer to
the luajit library are not created. Since ldconfig for the target
didn't make sense, LDCONFIG=true has been added a long time ago when
building the target variant of luajit.
However, since commit 9ba8d1cce4, the
host variant of luajit is also linked dynamically. While ldconfig for
the host should work, on certain machines ldconfig is not available in
the PATH, and needs to be explicitly called as
/sbin/ldconfig. However, since output/host/usr/lib is in the RPATH of
the binaries we create, there is really no need to bother running
ldconfig, so we simply pass LDCONFIG=true, like we do for the target
variant. This ensures that the symlinks are always created.
Fixes:
http://autobuild.buildroot.net/results/4dc/4dcbcde0046e265317e170a977e699e5c3a06b87/
Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Francois Perrad <francois.perrad@gadz.org>
[Thomas: rewrite commit log with a proper explanation.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The version bump doesn't inherently fix the security issues, however the
added CVE patches do, which fix:
CVE-2016-2147 - out of bounds write (heap) due to integer underflow in
udhcpc.
CVE-2016-2148 - heap-based buffer overflow in OPTION_6RD parsing.
Drop patches that are upstream as well.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The target name is qt5base, not qt5.
[Peter: extend commit message]
Signed-off-by: David Picard <davepiq@yahoo.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
It was accidentaly dropped in the loaders cache rework, shame on me.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
0003-thread-Posix-Mutex-Cond-use-constexpr-only-with-glib.patch is
upstream so remove it.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This version of Qwt is compatible with both Qt and Qt5. This patch
adds the option to compile it for Qt5.
Signed-off-by: David Picard <davepiq@yahoo.fr>
[Thomas: slightly adjust the OpenGL dependencies.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Rework the selection of Qt options in the qwt package to make it
easier to support other versions of Qt than just Qt4. Instead of
depending on Qt options, we select the ones we need, and simply have a
global "depends on BR2_PACKAGE_QT". This also allows to remove a
comment that becomes useless.
Signed-off-by: David Picard <davepiq@yahoo.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
After the commit f67a4f50e2 "gcc: preserve CXXFLAGS_FOR_TARGET"
target CFLAGS are passed to all libraries bundled with gcc. This breaks
libsanitizer on gcc-4.9.x when building with _FILE_OFFSET_BITS=64:
error: size of array ‘assertion_failed__837’ is negative
...
libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.cc:837:1:
note: in expansion of macro ‘CHECK_SIZE_AND_OFFSET’
CHECK_SIZE_AND_OFFSET(dirent, d_ino);
This issue is fixed in the libsanitizer mainline and in gcc-5.x.
There's no issue with gcc-4.8.x and earlier.
Reported-by: Julien CORJON <corjon.j@ecagroup.com>
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Backported from: https://llvm.org/svn/llvm-project/compiler-rt/trunk@220328
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
0006-fix-Error-allocating-memory-for-pkgDataFlags.-compil.patch is
upstream so drop it.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Create the gdk pixbuf loaders cache file at build time rather than run
time, since the root filesytem can be read-only.
We also didn't cater for systemd-based systems, so this solves two
issues at once.
The only side-effect is that we must now pull host-gdk-pixbuf into
target gdk-pixbuf dependencies, but generally users will have gdk-pixbuf
selected if they use libgtk2/3 which already pull in host-gdk-pixbuf.
And remove initscript correspondingly.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
libaio support is now automatic so drop the enable/disable (it will fall
back to pthread aio if libaio is not present).
0002-build-improve-stack-protector-check.patch is upstream so remove it.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
packages shouldn't select libgtk3 directly, just depend on it, like for
libgtk2.
In the past libgtk3 didn't require any *GL backend and the dead-end
solution/last resort was the broadway (networked) gdk backend - though
not very useful it didn't require any funky dependencies.
But now we do. Fixes:
http://autobuild.buildroot.net/results/794/794c7ed221432e46a810fc281732ba417cd4cda3/
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
DES is terribly outdated and a security vulnerability.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The tool helps to create a working SPL to boot Altera SoC FPGA boards.
The code of mkpimage is integrated directly from the Barebox repository
as stated in the 'mkpimage.mk' file.
This tool is *NOT* necessary for any other boards so far.
Signed-off-by: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Tested-by: Jan Viktorin <viktorin@rehivetech.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
vala/valac can use gir and vapi data files installed by other packages,
but since these are normally installed to staging and host-vala looks
for them in the host directory (logically) this leads to failure.
So wrap them to call the real tool and add this information via
command-line parameters to them.
This is required for vala-in-vala bindings (vapi).
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Drop 0003-support-for-non-glibc-libcs.patch since it's upstream.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Bug was already reported upstream, as solution upstream refered to this
commit: https://www.nlnetlabs.nl/bugs-script/show_bug.cgi?id=733#c2
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Switch to system/unbundled pcre since it's the default and recommended
by upstream now.
It's also good security practice since pcre patches won't get updated in
the bundled version inside glib so often.
As stated in glib's NEWS:
Overview of changes in GLib 2.47.5
* the system copy of PCRE is now used by default to implement GRegex.
Configure with --with-pcre=internal if a system PCRE version
is unavailable or undesired.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
They're required for host-libglib2 and using system pcre is the
default/recommended with newer versions.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
lsipc segfault when no option is given.
Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Erlang/OTP 18.3 is a service release on the 18 track with mostly bug
fixes, but is does contain a number of new features and improvements as
well.
Some highlights of the release are:
. New statistics info about runnable and active processes & ports.
Call erlang:statistics with: total_run_queue_lengths |
run_queue_lengths | total_active_tasks | active_tasks.
. Time warp improvements: dbg:p/2 and erlang:trace/3 with
monotonic_timestamp |strict_monotonic_timestamp.
. Introduced a validation callback for heart.
. The module overload in sasl has been deprecated.
. ~90 contributions since 18.2
Signed-off-by: Frank Hunleth <fhunleth@troodon-software.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Required to nicely match the previous libgtk3 major version bump.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Mark the wayland backend as good again since the bump and consequent
protocol version match fixes it.
Drop upstream 0004-Fix-undefined-reference-to-get_xkb.patch
Drop unnecessary 0005-do-not-build-extract-strings.patch
(extract-strings doesn't exist any more).
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
On newer toolchains (glibc >= 2.20) _BSD_SOURCE behaviour was deprecated
in favour if the _DEFAULT_SOURCE macro. See man 7 feature_test_macros.
Add patch from Fedora to also consider _DEFAULT_SOURCE. Fixes:
http://autobuild.buildroot.net/results/9e2/9e2126b0e68d0d59d37616a268adb810efd8281a/
Signed-off-by: Gustavo Zacarias <gustavo.zacarias@free-electrons.com>
Reviewed-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Tested-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
This package consists of scripts that setup cgroups at boot without
doing any cgroup management or classification of tasks into cgroups
Signed-off-by: Niranjan Reddy <niranjan.reddy@rockwellcollins.com>
[Thomas:
- rename to cgroupfs-mount to match upstream
- add proper hash, since hashes should be added for github sourced
packages
- fix minor typos in the init script
- fix the license file information.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
- Use the recently introduced BR2_TOOLCHAIN_HAS_ATOMIC boolean.
- Import an upstream patch to fix error handling when atomic operations
are not detected. Without this patch the build fails due to a syntax
error instead of showing the proper message.
- Add a patch to configure.ac to check if libatomic is needed and force
linking to it (we will attempt to submit this upstream).
- Disable build for SPARC64 because it fails due to a missing definition
of Atomic64.
On PowerPC, the __atomic_*() built-ins for 1-byte, 2-byte and 4-byte
types are available built-in. The corresponding built-ins for 8-byte
types, however, are implemented via libatomic, so requiring gcc >= 4.8.
In Buildroot, to simplify things, it was decided to require gcc 4.8 as
soon as the architectures has at least one __atomic_*() built-in variant
that requires libatomic.
Since protobuf most likely only uses the 1, 2 and 4-byte variants, it
*could* technically build with gcc 4.7. This is probably not a big deal,
and we can live with requiring gcc 4.8 on PowerPC to build protobuf. The
same restriction applies to SPARC.
The build for SPARC64 breaks even using the master branch of protobuf
due to undefined references to some NoBarrier_Atomic*() functions.
Signed-off-by: Henrique Marks <henrique.marks@datacom.ind.br>
Signed-off-by: Carlos Santos <casantos@datacom.ind.br>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
This commit modifies the cairo, icu and webkitgtk24 packages to use
BR2_TOOLCHAIN_HAS_LIBATOMIC when appropriate.
Fixes:
http://autobuild.buildroot.net/results/ec4/ec4e48c0e4b8fa72d8bb7ef4ad67a166699c0b62/
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>
While building libevas.so for the host-efl with luajit support, the link fail
when libluajit is build statically:
/usr/bin/ld: output/host/usr/lib/libluajit-5.1.a(ljamalg.o): relocation R_X86_64_32S against `.rodata' can not be used when making a shared object; recompile with -fPIC
output/host/usr/lib/libluajit-5.1.a: error adding symbols: Bad value
collect2: error: ld returned 1 exit status
Build luajit dynamically in order to build host-efl with luajit support.
Also, passing HOST_LDFLAGS is needed so that a proper rpath is added
to the resulting binaries. Note that it is expected to have
HOST_LDFLAGS assigned to TARGET_LDFLAGS, it's just that luajit has a
somewhat non-conventional naming of variables. This is needed to avoid
the following error from the rpath sanity check:
***
*** ERROR: package host-luajit installs executables without proper RPATH:
*** output/host/usr/bin/luajit
Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Remove files that aren't used for runtime purposes.
Size savings delta +450 KB.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
It was added to the networking category close to connman for the ability
to tweak connman features without going around (i.e. add support for
wifi and so on).
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Remove source schema files from target
(/usr/share/glib-2.0/schemas/*.dtd and *.xml) at target finalization
time.
This is because other packages install schemas as well to avoid
duplicating file removal efforts all around.
Add a schema compiler run on target finalization as well - we previously
didn't do this since not many packages use it, however the upcoming
connman-gtk package does require a valid gschemas.compiled to work
properly.
We also do this at target finalization to avoid each package handling
this on their own.
Purge some extra binaries and files from the target, like the schema
compiler to save additional space since they're unused at run time.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
This reverts commit c44cf2cc97.
Now that xtensa gas don't move literals into .init and .fini this fix is
no longer needed.
See upstream commit
https://sourceware.org/git/?p=binutils-gdb.git;a=commitdiff;h=4111950f363221c4641dc2f33bea61cc94f34906,
which was backported to all supported binutils version, under the name
*-xtensa-fix-.init-.fini-literals-moving.patch.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
[Thomas: add more details in the commit log, as suggested by Yann
E. Morin, and using information provided by Max.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
[yann.morin.1998@free.fr: independently done a patch similar to the one
by Thomas]
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The optinal dependency on celt is not added, because we only have
celt-0.5.1 and jack2 requires celt >= 0.5.2 (even though it does look
for celt >= 0.5.0, it does not build with celt- < 0.5.2). Since we
cannot upgrade celt (for now), we just never make jack2 depend on
celt051, and it won't find it either, as it just searches for celt.pc
and we have celt051.pc.
As well, the dependency on eigen is only useful in conjunction with
gtkiostream, for which we do not have a package. So, we don't need to
depend on eigen.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
jack2 allows to use three modes:
- A legacy jackd service alone.
- A DBUS jackd service alone.
- A mixture between the standalone jackd and the DBUS jackd.
This patch adds the possibility to choose between these three modes,
while the legacy jackd was the only option before.
In addition, the jack_control tool is removed when the DBUS jackd is not
available, as this tool is used to control it.
Signed-off-by: Antoine Tenart <antoine.tenart@free-electrons.com>
[yann.morin.1998@free.fr:
- rebase
- only use two booleans, not a choice
- python is a runtime-only dependency
- use python3 if enabled, fallback to python
- simplify post-install condition
]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
[Thomas: fix minor typo.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
For wayland 1.10+ they're split from wayland core and required by
weston.
Hide it behind the wayland package since it's useless independently even
though it can be "built" and installed.
And it makes no sense to install to target since they are DTDs used at
build-time.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
wayland-scanner host/cross support is now native so drop the funky
trick.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Move it outside x11r7 scope since it can be built without it with a few
patches (sent upstream via github pull request).
Motivation is simple, it's a dependency for newer gtk3 versions which
can work with a wayland or broadway backend, and having a full x11 stack
is pointless for that scenario.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
[Thomas: split the first patch into four separate patches, since
that's how they were submitted upstream.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
We need to match xserver here, since we need glamor support there in
order to enable it here, hence need to match the dependencies.
For xserver it needs dri3proto+libepoxy, however if only libepoxy is
present then glamor would be disabled for the server but enabled for the
driver, leading to build failure.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
This allows to build a m68k toolchain with uClibc.
Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Plugins want to use dlopen even when configured with --disable-shared.
Add --disable-plugins when STATIC_LIBS is selected.
Fixes:
http://autobuild.buildroot.net/results/a59e5858d777dcee3d8f7fb6e82ee4735b520155/
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
some new runtime options, minor fixes, and fixes for issues found by
various code analyze and lintian tools.
Signed-off-by: Alexander Dahl <post@lespocky.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Add a BR2_PACKAGE_WEBKITGTK24_HTTPS to pull in required dependencies for
HTTPS protocol support.
This is almost essentially the same as using midori with HTTPS enabled,
however Midori's HTTPS support is based on glib-networking, and webkit
via libsoup-ssl (for MiniBrowser, it doesn't affect midori's support).
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Add a BR2_PACKAGE_WEBKITGTK24_MULTIMEDIA option that will select/pull in
the basic required components in order for a few multimedia websites
(HTML5 video) to work.
Tested with videojs.com & youtube.com
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Needed by imxmp3audioenc and imxuniaudiodec plugins.
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Changelog:
- Add basic epdc firmware copy to correct output path
- Move legacy epdc firmware to this package out of kernel
- VPU firmware upgrade to v3.1.1_r46070
This patch is based on the Yocto equivalent:
https://github.com/Freescale/meta-fsl-arm/commit/0fdf829
This package has been implicitely tested through gstreamer as the
plugins rely on them for vpu decoding for instance:
# gst-launch-1.0 filesrc location=/root/tears_of_steel_1080p.webm ! \
matroskademux ! imxvpudec ! imxipuvideosink
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Changelog:
- Support AVcVuiInfo for Decoder through DecInitialInfo or DecOutputInfo
- Clear structure iram IOGetIramBase() fails to set them to avoid
using uninitialized variables.
This patch is based on the Yocto equivalent:
https://github.com/Freescale/meta-fsl-arm/commit/be030922
This package has been implicitely tested through gstreamer as the
plugins rely on it for vpu decoding:
# gst-launch-0.10 playbin uri=file:///root/tears_of_steel_1080p.webm
# gst-launch-1.0 playbin uri=file:///root/tears_of_steel_1080p.webm
# gst-launch-1.0 filesrc location=/root/tears_of_steel_1080p.webm ! \
matroskademux ! imxvpudec ! imxipuvideosink
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Changelog:
- add new feature to read bch geometry setting from debugfs, it provides
the feasibility to support large oob NAND devices.
This patch is based on the Yocto equivalent:
https://github.com/Freescale/meta-fsl-arm/commit/9953874c
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
In order to match the imx-gpu-viv graphics libraries version.
This patch is based on the Yocto equivalent:
https://github.com/Freescale/meta-fsl-arm/commit/dcfa6752
This package has been tested with the following commands:
# modprobe galcore
# cd /usr/share/examples/viv_samples/vdk/
# ./tutorial7
Signen-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This version is released with 3.14.52-1.1.0_ga release.
Includes many of the bug fixes and stability improvements.
For more information refer to i.MX Linux Release Notes from NXP website:
L3.14.52_1.1.0_LINUX_DOCS package is under Supporting Information.
This patch is based on the Yocto equivalent:
https://github.com/Freescale/meta-fsl-arm/commit/f1161869
This package has been tested with both X11 and Framebuffer backends:
# cd /usr/share/examples/viv_samples/vdk/
# apitrace trace --api egl ./tutorial7
# gmem_info
... display memory use per PID ...
# apitrace replay tutorial7.trace
# eglretrace tutorial7.trace
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Changelog:
- Add frame rate check and return failure if frame rate is invalid
value (<=0)
This patch is based on the Yocto equivalent:
https://github.com/Freescale/meta-fsl-arm/commit/67b3b998
This package has been implicitely tested through gstreamer as the
plugins rely on it for vpu decoding:
# gst-launch-0.10 playbin uri=file:///root/tears_of_steel_1080p.webm
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Changelog since 4.0.7:
-Support hevc in MPG2 parser.
-Enhance the parsing conditions in SPS nal unit.
Parse system header to get stream id.
For mpeg video, don't call parseh264 to avoid mistakes.
-Fix memory leak, free temp data buffer after parsing header.
This patch is based on the Yocto equivalent:
https://github.com/Freescale/meta-fsl-arm/commit/c3aa06b3
This package has been implicitely tested through gstreamer as the 0.10
plugin relies on it:
# gst-launch-0.10 playbin uri=file:///root/tears_of_steel_1080p.webm
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Changelog since 4.0.7:
- Version alignment with other Multimedia components.
This patch is based on the Yocto equivalent:
https://github.com/Freescale/meta-fsl-arm/commit/6a1f559a
This package has been implicitely tested through gstreamer as the 0.10
plugin relies on it:
# gst-launch-0.10 playbin uri=file:///root/tears_of_steel_1080p.webm
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Since commit 604095fe9b ("libcap: add
patch to fix build issue with old kernel headers"), libcap builds fine
with headers < 3.0, so it is no longer the reason why lxc needs
headers >= 3.0.
However, lxc uses setns(), which is only available since kernel 3.0,
so we simply update the comment next to the dependency so that it is
accurate.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Now that libcap no longer needs kernel headers >= 3.0, we can remove
this dependency from lxc. However, building with headers 2.6.32
exhibits a build issue caused by the redefinition of the setns()
function.
Since setns() is not implemented in the C library, lxc provides its
own version. However, for some reason, while the C library doesn't
implement setns(), it provides a prototype for it, which is not
exactly the same as the one in lxc, causing a build failure. We re-use
a solution implemented in gdb to solve the same problem: define in lxc
a function called do_setns(), which calls setns() when available, or
manually does the system call otherwise.
Of course, with old kernels the system call will not be available, so
things will fail at runtime, but this was anyway already the behavior
of lxc's setns() dummy implementation.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Add dependency on BR2_TOOLCHAIN_HAS_SYNC_2/4 since it uses both
__sync_fetch_and_add_2() and __sync_fetch_and_add_4() atomic builtins.
Fixes:
http://autobuild.buildroot.net/results/8f2/8f2a3571611dc9414c23808e7615f87b677557dd/
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
As part of bumping to version 0.9.3a, two patches are dropped because they
are already upstream.
Signed-off-by: Joshua Henderson <digitalpeer@digitalpeer.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
libpjsip bundles several third party libraries. In Buildroot we prefer
either not to build them or to depend on a proper package for each of
them. The current recipe disables most of them, but not all, so
disable the remaining ones.
Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The -rpath option was being stripped from sdl-config via a post install
staging hook, but the same wasn't being done for sdl.pc. Because of
this, packages that detect SDL via pkg-config ended up passing
'-Wl,-rpath,/usr/lib' to the linker, which caused build failures under
certain circumstances since libraries were being looked for in the wrong
directory.
Fix by passing the --disable-rpath option to the SDL configure script,
which takes care of disabling -rpath everywhere. This also allows the
SDL_FIXUP_SDL_CONFIG hack to be completely removed.
Fixes:
http://autobuild.buildroot.net/results/624/62499217eeaf3228b46652e3f65776d7ece8fce6/http://autobuild.buildroot.net/results/cc1/cc1f78f6c43e3a7bf3ed80d759d9c4d7363d0e48/
Signed-off-by: Rodrigo Rebello <rprebello@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Now that the libcap package has a patch that makes it build with
kernel headers < 3.0 (which was needed for the host variant of
libcap), there is no longer a need to have a dependency on headers >=
3.0 for the target variant of libcap.
All reverse dependencies of libcap are handled in this commit, except
lxc, which will be handled in a separate commit since it needs some
special solution.
The build of all those packages has been tested with a toolchain that
uses kernel headers 2.6.32, which is the oldest that our default glibc
version accepts to use.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Old kernels (before 2.6.36) were defining XATTR_NAME_CAPS in
<linux/capability.h>, but using XATTR_SECURITY_PREFIX and
XATTR_CAPS_SUFFIX which were defined in the kernel-only part of
<linux/xattr.h>.
In kernel 2.6.36 (commit af4f136056c984b0aa67feed7d3170b958370b2f),
the XATTR_NAME_CAPS definition was moved to the kernel-only part of
<linux/xattr.h>. It's only in kernel 3.0 (commit
1dbe39424a43e56a6c9aed12661192af51dcdb9f) that <linux/xattr.h> was
fixed to expose XATTR_NAME_CAPS and the related definitions to
userspace.
This is the reason why the target variant of libcap has a dependency
on headers >= 3.0 for the moment.
However, this doesn't solve the problem for the host variant of
libcap, which doesn't build properly on old systems.
To solve this, we simply add a patch that defines the missing
definitions. Their values haven't changed over time since they are
part of the kernel to userspace ABI.
Fixes:
http://autobuild.buildroot.org/results/856b71bccf14c3334a8c0fc66c1d985b09734313/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
attr is no longer a dependency, not even optional.
Refresh our patches, and drop the backport from upstream.
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>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Even though it's inherited by the python dependency it's more clear this
way for graph-depends, since it's used by the waf buildsystem.
And even though we have a hard dependency on python for the distro this
python could ostensibly be 3.x which isn't compatible with the bundled
waf series (1.5.x) in samba (as of current shipping version and upcoming
4.4.x series).
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Only used/useful with the gtk3 backend though.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This reverts commit 2dcab526a9.
Now that gcc correctly propagates CXXFLAGS_FOR_TARGET for libstdc++
build this is no longer needed.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
gcc-4.7.x, gcc-4.8.x and gcc-4.9.x don't propagate CXXFLAGS_FOR_TARGET to
CXXFLAGS for libstdc++ build. As a result libstdc++ is built without
TARGET_CFLAGS and may fail to link with applications using it, see e.g.
http://autobuild.buildroot.net/results/81a3bca5cbcf789c7ce1aa221a6a4154dd7c3917/
Instead of passing TARGET_ABI or TARGET_CFLAGS for libstdc++ in
--enable-cxx-flags parameter backport the patch that fixes propagation
of CXXFLAGS_FOR_TARGET to CXXFLAGS.
This issue is fixed in gcc-5.x
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
2.39.1 (and all odd-numbered versions for y in x.y.z) are development
releases not intended for general usage, so roll back to the latest
stable version.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
It's been deprecated for a year now so remove it.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Remove for the upcoming 2016.05 release, it's been deprecated for a year
now.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
It's been deprecated for a year now so remove it.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
It's been deprecated for a year now so remove it.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
It's been deprecated for a year now so remove it.
[Peter: drop !samba dependency from samba4]
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
samba(3) has been deprecated for quite some time so switch mplayer to
use samba4 if available.
It needs a little extra tweak to pick up the proper cflags since the odd
configure script doesn't do it.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fixes CVE-2016-2342 (AKA VU#270232): Quagga bgpd with BGP peers enabled for
VPNv4 contains a buffer overflow vulnerability.
Remove the --enable-babeld and --enable-opaque-lsa configure options that were
removed in this release.
See the release announcement at
https://lists.quagga.net/pipermail/quagga-dev/2016-March/014938.html.
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Commit d16b32a401 (package/rpi-firmware: add option to install vcdbg) added
a sub option to install the vcdbg tool, but the commit unfortunately
contained two errors:
- The package sets _INSTALL_TARGET = NO, so the _TARGET_CMDS are never used
- The install command used -d (handle arguments as directories) instead of
-D (create needed directories)
Fix these issues.
Cc: yann.morin.1998@free.fr
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
