mesa3d-demos: select libglew/libglu if needed

Fixes:
http://autobuild.buildroot.net/results/4b1/4b16c9797c653c248725a55d9b8eafc4634b968c/build-end.log

libglew/libglu are added to _DEPENDENCIES for X11+GL, but not selected at
the kconfig level.

While we're at it, correct the dependencies for the X11 selects as well
(only used if GL is available).

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

xapp_xcompmgr: select xlib_libXrender

Fixes:
http://autobuild.buildroot.net/results/969/9690a67083f31070e27c95b053fc0fd389fd373a/

libXrender is added to _DEPENDENCIES, but not selected at the kconfig level.
While we're at it, sort the selects alphabetically and drop the double
libxfixes select.

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

package/x11r7/xserver_xorg-server: move libpciaccess selection

Move libpciaccess selection under BR2_PACKAGE_XSERVER_XORG_SERVER_MODULAR
to simplify.

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/x11r7/xserver_xorg-server: select xlib_libxshmfence required by dri3proto support dependencies

When BR2_PACKAGE_XPROTO_DRI3PROTO and
BR2_PACKAGE_XSERVER_XORG_SERVER_MODULAR are selected, xlib_libxshmfence
package build is triggered at Makefile level.

Since 28f67899e54f15c6e48484ea2976c3e431756bcb, the build
stop if a package build is trigged at Makefile level without
being selected at Config.in level.

Fixes:
http://autobuild.buildroot.net/results/a25/a253d19191dbc8bcc54f9e9d7fddae03d6695b4b/

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/x11r7/xserver_xorg-server: reorder select/depends

>From [1]:
"Even though the ordering has absolutely no consequences in Kconfig, it
is not logical (when reading). It is more logical and far easier to
understand when depends come first, followed by the selects."

Also, the Config.in exemple in the manual suggest to use this coding
style [2].

[1] http://lists.busybox.net/pipermail/buildroot/2015-October/142955.html
[2] http://buildroot.uclibc.org/downloads/manual/manual.html#writing-rules-config-in

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/mesa3d: select packages required by dri3proto support dependencies

When BR2_PACKAGE_MESA3D_DRI_DRIVER and BR2_PACKAGE_XPROTO_DRI3PROTO
are selected, xlib_libxshmfence and xproto_presentproto packages
build are trigged at Makefile level.

Since 28f67899e54f15c6e48484ea2976c3e431756bcb, the build
stop if a package build is trigged at Makefile level without
being selected at Config.in level.

Fixes:
http://autobuild.buildroot.net/results/6a4/6a471d2873ff7e645b0f0962d1719632e290993c

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/mesa3d: reorder select/depends

>From [1]:
"Even though the ordering has absolutely no consequences in Kconfig, it
is not logical (when reading). It is more logical and far easier to
understand when depends come first, followed by the selects."

Also, the Config.in exemple in the manual suggest to use this coding
style [2].

[1] http://lists.busybox.net/pipermail/buildroot/2015-October/142955.html
[2] http://buildroot.uclibc.org/downloads/manual/manual.html#writing-rules-config-in

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

python3: switch to Git formatted patches

In preparation for the bump to Python 3.5.0, let's switch all the
patches to the Git format. This way, a Git repository of the Python
source code can be used to manage those patches, which makes it easier
to bump to newer Python versions.

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

fakeroot: fix build when libcap-dev is installed

Fixes https://bugs.busybox.net/show_bug.cgi?id=8541

Signed-off-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Tested-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

strace: enable supported nios2 arch

Support for Nios-II has been added in strace release 4.11.

Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Tested-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

pciutils: correct dns disabling logic

Fixes #8581

Commit 2c879f95a0 (pciutils: cleanup) refactored the make arguments, but
accidently changed the logic to disable dns based lookups to pass
PCIUTILS_DNS=no instead of DNS=no, causing the autotection to be used.

This autodetection looks at host includes, so it isn't safe for cross
compilation.

Disable it by setting the correct make variable instead.

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

mongodb: disable support for armv4 and armv5

When compiling for armv4 or armv5, we get the following error:

src/mongo/util/signal_handlers_synchronous.cpp:188:9: error: 'current_exception' is not a member of 'std'
     if (std::current_exception()) {
         ^

This is because libstdc++ will only enable std::current_exception if ATOMIC_INT_LOCK_FREE > 1, which is not the case for armv4 and armv5 architectures, so disable them.

Fixes:
http://autobuild.buildroot.net/results/589a02f511939531141135def73785b600113152/
http://autobuild.buildroot.net/results/89b9cd2dae8645b50407409c7f2a8567c551822c/
http://autobuild.buildroot.net/results/b8ffce927cb0563f5c43533ec2f02d3487acc198/

Signed-off-by: Sergio Prado <sergio.prado@e-labworks.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

xapp_xbacklight: select xcb-util package

xapp_xbacklight depends on the xcb-utils package but doesn't select it.
Fixes:

http://autobuild.buildroot.net/results/d96/d96aa77e02954bef6fb6e9d244fbf9d1e939a374/

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

mongodb: fix legal info build error

Remove extra comma to fix legal info build error.

Fixes:
http://autobuild.buildroot.net/results/87faca89bc9ff6ff4699c4b50bbb890ff80b95ee/
http://autobuild.buildroot.net/results/ea54721ef0d1c210b2baa9eef7c6c68863020b50/
http://autobuild.buildroot.net/results/5a81db472d432880f44dfb30bae0d777b6ae6c79/
...

Signed-off-by: Sergio Prado <sergio.prado@e-labworks.com>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

pciutils: bump to version 3.4.1

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

lighttpd: bump to version 1.4.39

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

inittab: disable reboot by ctrl-alt-del

Most embedded systems do not have a keyboard.
If a keyboard is attached afterwards (e.g. to an accessible USB host port)
this allows the user to reboot the system by pressing Ctrl-Alt-Del, which
might not be desired.

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

Add ARM11 MPCore CPU target support

gcc differentiates the mpcore-with-vfp from the mcpore-without-vfp
CPUs. The former is named just 'mpcore', while the latter is named
'mpcorenovfp'.

We only add one entry, 'mpcore' and let the user select whether or
not to use the VFP. We then name the CPU according to the user's
selection.

Signed-off-by: Sergi Granell <xerpi.g.12@gmail.com>
Tested-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

pkg-generic: improve incorrectly used package detection

Currently, the check that packages we build are indeed enabled is done
at the time a package is configured.

This can come quite late in the build process, and does not provide
direct knowledge of the real culprit for the incorrect dependency.

However, we can improve these two issues quite easily, albeit at the
expense of a very slightly more complicated make code.

First, the check can not be done at the time we define the package, i.e.
in the inner-generic-pacakge, because all its dependencies might have
not been parsed yet, so we can't yet know whether it is enabled or not
(because we can't match the package name of the dependency to its
Kconfig variable yet).

But then, we know we have all packages definitions after we scanned the
the bundled packages, kernel, bootloaders and toolchains, as well as the
br2-external tree (if any).

So, at this location, we iterate through the list of enabled packages,
and check that the packages they each depend on are indeed enabled.

This allows us to:
 1- do the check very early, before any build action,
 2- report on the exact offending package very easily.

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

tpm-tools: fix build failures caused by -Werror

This fixes this build failure (and many others):
  http://autobuild.buildroot.org/results/d56/d56bfc69ef0a478219c959d50fb96d526fc2155e/

The build failure is caused by the build system using -Werror; this is
already patched upstream, so this commit just imports the upstream
patch.

Signed-off-by: Noé Rubinstein <noe.rubinstein@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/boost: add patch to fix libquadmath issue

Backport patch from upstream [1] to fix build issues with toolchains not
providing libquadmath, but still supporting __float128.

Fixes:
http://autobuild.buildroot.net/results/fcc/fccb40e7b05b84275cfe36a890cedbbaaaa4f953/
http://autobuild.buildroot.net/results/cfc/cfcc6c44d2225b9baf0fd92d3cfba4922bfb248b/
http://autobuild.buildroot.net/results/ab1/ab1a5aceedb7f9edcd5536b601f1ffb66c5fe1e8/
http://autobuild.buildroot.net/results/1ed/1ed3c1ae1e31325b49dd05c1f616637bb5e6382d/
http://autobuild.buildroot.net/results/9e5/9e53a5aff9966c6f26e705c91d13692107452feb/
and many more.

Backported from: 74ff2db959c5fa75bec770c41ed2951a740fe936

[1]
https://github.com/boostorg/math/commit/74ff2db959c5fa75bec770c41ed2951a740fe936

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

python-urllib3: new package

Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

support/download: generate reproducible Bazaar archives

Similarly to what has previously been done for the Hg download backend,
instruct bzr to generate the archive on stdout, so that we can generate
reproducible archives.

When instructing bzr to generate the output file by itself, it uses a
temporary file that is then fed to gzip, which in turn stores the
timestamp of that file in the generated archive, whereas when the output
is generated on stdout, there is no timestamp, so the archive is then
reproducible.

Bizarely enough, we can tell 'bazaar' not to generate a bazaar in the
archive. Cool, uh? ;-]

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

package/tremor: select libogg package

tremor depends on libogg package but doesn't select it
in Config.in. so libogg build is trigged at Makefile level.

Since 28f67899e54f15c6e48484ea2976c3e431756bcb, the build
stop if a package build is trigged at Makefile level without
being selected at Config.in level.

Fixes:
http://autobuild.buildroot.net/results/0da/0da3265f0f1e499f5d34ed32b0d13a2eda53f976

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

support/download: generate reproducible Hg archives

When hg directly creates the output file, the hash for that file changes
everytime.

However, if we just tell hg to output the archive on stdout and we do
the redirect to the file, then the archive is reproducible.

(The reason is that in the first case, a temporary file is created and
then compressed, and gzip is adding the filename and its timestamp in
the gzip header, while in the second case, there is no temporary file,
and thus no timestamp and thus it is reproducible.)

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Yegor Yefremov <yegorslists@googlemail.com>
Tested-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Makefile: drop ldconfig handling

The ldconfig handling in the main Makefile is utterly broken, as it
calls the build machine ldconfig to generate the ld.so.cache of the
target. Unfortunately, the format of the ld.so.cache is architecture
specific, and therefore the build machine ldconfig cannot be used
as-is.

This patch therefore simply drops using ldconfig entirely, and removes
/etc/ld.so.conf.d/ from the target skeleton. The idea is that all
libraries that should be loaded by the dynamic linker must be
installed in paths where the dynamic linker searches them by default
(typically /lib or /usr/lib).

This might potentially break a few packages, but the only way to know
is to actually stop handling ldconfig.

In order to be notified of such cases, we add a check in
target-finalize to verify that there is no /etc/ld.so.conf file as
well as no /etc/ld.so.conf.d directory.

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

package:sysdig: propagate dependencies from jsoncpp

sysdig select jsoncpp which needs a gcc compiler >= 4.7

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/sysdig: reorder select/depends

>From [1]:
"Even though the ordering has absolutely no consequences in Kconfig, it
is not logical (when reading). It is more logical and far easier to
understand when depends come first, followed by the selects."

Also, the Config.in exemple in the manual suggest to use this coding
style [2].

[1] http://lists.busybox.net/pipermail/buildroot/2015-October/142955.html
[2] http://buildroot.uclibc.org/downloads/manual/manual.html#writing-rules-config-in

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

tinyalsa: fix toolchain comment

Documented syntax is '.. needs A toolchain ..'

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

kodi: fix libva toolchain comment

Documented syntax is '.. needs A toolchain ..'

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

gst1-plugins-good: fix v4l2 toolchain comment

Documented syntax is '.. needs A toolchain ..'

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

cxxtest: new package

[Peter: fix toolchain comment, use INSTALL instead of mkdir/cp]
Reviewed-by: Marcin Nowakowski <marcin.nowakowski@imgtec.com>
Reviewed-by: Rahul Bedarkar <rahul.bedarkar@imgtec.com>
Reviewed-by: Piotr Nakraszewicz <piotr.nakraszewicz@imgtec.com>
Reviewed-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Abhishek Singh <Abhishek.Singh@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

python-pyxml: remove obsolete package

PyXML is bitrotten and obsolete. You'll find the functionality it
previously provided is now included in the Python standard library.

So no need to keep this package in Buildroot anymore.

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

package/openocd: select libusb when selecting libusb-compat

libusb-compat depends on libusb package, so we need to select
libusb package at the same time as libusb-compat to avoid an
unmet dependency.

Since 28f67899e54f15c6e48484ea2976c3e431756bcb, the build
stop if a package build is trigged at Makefile level without
being selected at Config.in level.

This is the case here with BR2_PACKAGE_OPENOCD_USBPROG which
select BR2_PACKAGE_LIBUSB_COMPAT without selecting
BR2_PACKAGE_LIBUSB.

Fixes:
http://autobuild.buildroot.net/results/b4a/b4a8c54eed6bf75ab8ade0653953d19792155727/

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/gstreamer/zbar-plugin: propagate dependencies from zbar

BR2_PACKAGE_GST_PLUGINS_BAD_ZBAR select BR2_PACKAGE_ZBAR which
has several dependencies from libv4l.

Propagate these dependencies to avoid unmet dependencies while
selecting BR2_PACKAGE_GST_PLUGINS_BAD_ZBAR.

[Peter: show comment if toolchain dependencies aren't available]
Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

dillo: fix jpeg dependency

dillo was explicitly pulling in libjpeg instead of the virtual 'jpeg'
package, even though it also works with jpeg-turbo.

Reported-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

mongodb: new package

Compiled and tested on arm (beaglebone black), aarch64 (qemu),
i386 (qemu) and x86_64 (qemu).

[Peter: limit to supported archs, disable -Werror]
Signed-off-by: Sergio Prado <sergio.prado@e-labworks.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

acsccid: propagate dependencies from pcsc-lite

The new package acsccid properly propagated the dependency of libusb,
but not the one of pcsc-lite, causing pcsc-lite to be selected in
situation where it shouldn't be (such as in static-only
configurations).

Fixes:

  http://autobuild.buildroot.org/results/e0b/e0b205dceb84303d1dfaf8b39562b934f6914b8d/

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

configs: orangepipc defconfig needs dtc for u-boot

As can be seen at
https://travis-ci.org/buildroot/buildroot-defconfig-testing/jobs/99518455,
the new Orange Pi PC defconfig fails to build due to DTC being missing
when building U-Boot. This commit adds the appropriate option to add
DTC as a dependency to U-Boot.

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

configs/raspberrypi*: update them to use genimage

Update the raspberrypi and raspberrypi2 configs to use genimage directly
to build the image.
Update the documentation to reflect this, and drop the volatile rootfs
option since it doesn't make much sense and it's not easily integrated
with the genimage configurations.

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

vlc: add optional dependency to libmatroska

libmatroska is an optional dependency to vlc:
http://git.videolan.org/?p=vlc.git;a=blob;f=configure.ac;h=c46120d7f2a7d49d68d488d617e071148b8c5595;hb=HEAD#l2044

Add it to vlc.mk to get reproducable builds.

Signed-off-by: Pieterjan Camerlynck <pieterjan.camerlynck@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

linux-headers: bump 3.2.x series

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

nodejs: version 5.X needs host GCC >= 4.8

Fixes
http://autobuild.buildroot.org/results/4589dd076585d6472ad1e65926ffe68343b94422
http://autobuild.buildroot.org/results/e3147c0d9c9eb9f58773b75c8cb4ea49df483611
and many more.

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

skeleton: fix VCS folders not being excluded from rootfs

There is a typo in skeleton.mk preventing rsync from excluding VCS
folders such as .svn.

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

Config.in: add symbols for BR2_HOST_GCC_AT_LEAST_X_Y

Some host packages need a recent gcc version. Add symbols to Config.in
to specify the HOSTCC version. The values are passed through the
environment, and this environment is generated in a new support script.

Also update the documentation to mention the new symbols.

[Thomas: simplify by using only make logic instead of an external
shell script.]

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

python-twisted: bump to 15.5.0

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

vlc: add support for libdvbpsi

Signed-off-by: Pieterjan Camerlynck <pieterjan.camerlynck@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libdvbpsi: new package

libdvbpsi is a library used by vlc for decoding MPEG TS

[Thomas:
 - add missing Config.in comment about the thread dependency.]

Signed-off-by: Pieterjan Camerlynck <pieterjan.camerlynck@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

bind: add libressl-enabling patch

Besides the version bump we also require this patch, which is upstream.

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

fetchmail: add libressl-enabling patch

Patch status: upstream.

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

acsccid: new package

[Thomas:
 - fix commit title
 - change libusb back to a mandatory dependency, since even though
   there is a --disable-libusb option, it has no effect, and the code
   fails to build due to missing libusb.h
 - add hash file, noticed by Arnout
 - remove host-perl dependency, noticed by Arnout
 - rewrap Config.in help text, noticed by Arnout.]

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

libvncserver: disable on nios2, due to binutils issues

Fixes:

  http://autobuild.buildroot.org/results/469/469d5667f05ca6954805772fcaaa2b33dfdc6167/

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

xserver_xorg-server: add missing select BR2_PACKAGE_MCOOKIE

Fixes:

  http://autobuild.buildroot.org/results/7af/7afbf93492819ce59141f056f72679e4cdec0649/

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

toolchain-external: CodeSourcery PowerPC: Revert the removal of CS PowerPC 2011.03

As reported by Yann E. MORIN [1], the latest CS PowerPC toolchain (2012.03)
requires a PPC CPU with SPE, which is basically two variants, 8540 (e500v1) and
8548 (e500v2) in Buildroot. All other PPC CPU can't use that toolchain.

Keep CS PowerPC 2011.03 as latest available version and add a second Kconfig
symbol for the CS PowerPC 2012.03 since it's verry specific to one CPU type
(e500v2).

Previously it was possible to select the CS 2012.03 with a powerpc 8540 (e500v1)
CPU but the sysroot provided by the toolchain only support the 8548 (e500v2)
variant. Allow to select CS 2012.03 only with BR2_powerpc_8548.

Also re-add the previous CS toolchain handling for pixman and liquid-dsp.

[1] http://lists.busybox.net/pipermail/buildroot/2015-December/148308.html

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

hidapi: add libiconv handling

hidapi uses iconv functions, which on uClibc without locale support is
provided by the external libiconv library. This commit adds the
necessary libiconv handling to make the hidapi package build on
!locale toolchains.

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

hidapi: fix build with toolchains lacking C++ support

hidapi currently doesn't build on non-C++ capable toolchains due to
hidtest being a C++ source file, even if it in facts contains only C
code. This commit adds a patch that fixes this in the hidapi source
code.

Fixes:

  http://autobuild.buildroot.org/results/6ce/6ce0a4b8c7acb857005350a57c313b493bc6e2b7/

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

gst1-plugins-bad: fix openh264 handling

Commit 301e8ffbb25 (gst1-plugins-bad: update configure options) added a sub
option for openh264, but didn't propagate the toolchain dependencies or add
any configure options.

Fix this and while we're at it also add a help text for the option.

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

package/gstreamer1: Allow to build OpenJPEG plugin

The OpenJPEG library is packaged in buildroot (version 1) and there is a
plugin "openjpeg" using it in gst1-plugins-bad. This commit add the option
for building this plugin. It provide a JPEG2000 encoder and decoder.

[Peter: use 'select' instead of 'depends on', sort alphabetically]
Signed-off-by: Vincent Dehors <vincent.dehors@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

netatalk: bump to version 3.1.8

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

iptables: add parallel build fix patch

Fixes:
http://autobuild.buildroot.net/results/5c4/5c4ca595c4bf744aa00fc58851eeccf81f9c6ecc/

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

bind: disable libjson support

It conflicts with jsoncpp, bind probes for json/json.h first, but that
header is installed by jsoncpp, which is completely different from
json-c.
Since it's not clear who's correct here (there might be some other
json-c predecessor/version that installs there as well) and the same
functionality (stats channel) is provided by libxml2 as well, just
disable libjson support completely.

Fixes:
http://autobuild.buildroot.net/results/226/2262c9b46663ea7a45e128a5fd7ff30417c2c2a7/build-end.log
(indirectly, it was probing aboslute directories while searching for it)

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

qt: add missing select BR2_PACKAGE_TSLIB

Fixes:

  http://autobuild.buildroot.org/results/36c/36c56d10abcf96afd389f2ebe8db4dc13659997b/

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

dbus: audit support needs both audit and libcap-ng

As was suggested by the <pkg>_DEPENDENCIES variable, the audit support
in DBus requires both audit and libcap-ng. However, it didn't take
care of the fact that libcap-ng must be enabled in the configuration
to depend on it, causing some build failures with the newly added
check.

DBus configure.ac confirms that both packages are needed to enable
audit support, so we simply fix the condition to only be true when
both BR2_PACKAGE_AUDIT *and* BR2_PACKAGE_LIBCAP_NG are true.

Fixes:

  http://autobuild.buildroot.org/results/239/23953cc66faecb65e9ebf1f6980924f823d736a2/

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

php: add missing select BR2_PACKAGE_LIBMCRYPT

Fixes:

  http://autobuild.buildroot.org/results/b0c/b0c9ea0d577a74549476ddbdb0d2ecd09311c765/

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

openbox: add missing propagated dependencies from pango

pango needs atomics support and C++ support, and those dependencies
where not propagated to the new openbox package.

Should fix:

  http://autobuild.buildroot.org/results/d3b/d3b17335f2e5720a69b4635602fef30a5ae34660/

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

lttng-babeltrace: add missing select BR2_PACKAGE_POPT

lttng-babeltrace unconditionally uses "popt" in its <pkg>_DEPENDENCIES
variable, but does not select it. This commit fixes this
inconsistency.

Fixes:

  http://autobuild.buildroot.org/results/274/2748a02b090ce078fc880bd821df3c913b8c8dfd/

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

sshfs: use SSHFS_VERSION variable in URL

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

libfuse: use LIBFUSE_VERSION variable in URL

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

vboot-utils: ssize_t needs unistd.h

Fixes
http://autobuild.buildroot.net/results/95c8708411fbb116f313a0fdeccacf92cf9e3d8d/
http://autobuild.buildroot.net/results/cf4aa60f9fc9c48ff13d3233994b77b7c0950afe/
http://autobuild.buildroot.net/results/6f7dcafde693f06dfc88c922c5f8dd70ae2e30b5/
http://autobuild.buildroot.net/results/a2e95784d32c8842f6947801fd0d8c48f68b84e2/
http://autobuild.buildroot.net/results/f69ae4c9eee27df43389ae76048fb4ddd5a3eca6/

Signed-off-by: Alex Suykov <alex.suykov@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

vboot-utils: MTD_CHAR_MAJOR definition

Fixes
http://autobuild.buildroot.net/results/a6360fd21314899321071c34fcb4892c5eed9bcc/
http://autobuild.buildroot.net/results/0af3aa4a429c45d3a2609913ec1040f32378f3c8/
http://autobuild.buildroot.net/results/cc4940fd6f51b31d71014f5aedee5223180eb36e/

Signed-off-by: Alex Suykov <alex.suykov@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

python-paho-mqtt: add new package

Signed-off-by: Davide Viti <zinosat@tiscali.it>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/tpm-tools: fix build with locale support

tpm-tools needs to link with libintl when locale support is enabled.

However, tpm-tools' build-system internally builds a couple of static libraries
used to build the final programs.
When linking these libraries or the final programs, we need to explicitly tell
the linker to link with libintl in order to resolve all the missing symbols.

Fixes:
  http://autobuild.buildroot.org/results/560/56020558ec298ac4007bf7f6ce030b3f2fa1157d/
  http://autobuild.buildroot.org/results/17f/17fc6ca79765fd9ecdacec4f5a2b58caa554d48d/
  http://autobuild.buildroot.org/results/2d2/2d2c655429df62da2b8746a564fc064609e96001/
  http://autobuild.buildroot.org/results/d52/d5259b83e7a269bca019d024366aa201f5a174fd/
  http://autobuild.buildroot.org/results/11c/11cb0d1e51459e75e5f376cb07ce0188f8f050c4/
  http://autobuild.buildroot.org/results/bad/bad9a97b4fff00732f2efb043bbb76a99963c491/
  http://autobuild.buildroot.org/results/2f8/2f8bd534e6fcd9d356fe9087cb1394b9b2c4192c/
  and many others...

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Cc: Noé Rubinstein <nrubinstein@aldebaran.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

mbedtls: disable assembly for some scenarios

It's affected in the same way as polarssl, but use an enhanced trick
compared to 130ca81b since it's required by the codebase. Fixes:
http://autobuild.buildroot.net/results/584/584d1d12be9f2bbf45ccbdf93b6b442524059dcc/
http://autobuild.buildroot.net/results/d13/d13a06ba3fc0d7f08a03c04fcb484761da7af3f7/

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

bind: bump to version 9.10.3-P2

Leave the LTS series for the latest stable version for libressl
compatibility.
Unfortunately this means threads are now required, but this shouldn't be
a problem for a fully-featured resolver.

Drop 0001-disable-tests.patch since it's no longer required, genrandom
isn't run unless the tests are called upon.

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

wireshark: update help text

Since commit 00be67d84696 (wireshark: enable GUI options, 2015-12-02) the GUI
is enabled when the required libraries are present. Update the config help
text accordingly.

Cc: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

wireshark: security bump to version 2.0.1

Fixes:
wnpa-sec-2015-31 - NBAP dissector crashes
wnpa-sec-2015-37 - NLM dissector crashes
wnpa-sec-2015-39 - BER dissector crashes
wnpa-sec-2015-40 - Zlib decompression crash
wnpa-sec-2015-41 - SCTP dissector crashes
wnpa-sec-2015-42 - 802.11 decryption crash
wnpa-sec-2015-43 - DIAMETER dissector crash
wnpa-sec-2015-44 - VeriWave file parser crashes
wnpa-sec-2015-45 - RSVP dissector crash
wnpa-sec-2015-46 - ANSI A & GSM A dissector crashes
wnpa-sec-2015-47 - Ascend file parser crash
wnpa-sec-2015-48 - NBAP dissector crash
wnpa-sec-2015-49 - RSL dissector crash
wnpa-sec-2015-50 - ZigBee ZCL dissector crash
wnpa-sec-2015-51 - Sniffer file parser crash
wnpa-sec-2015-52 - NWP dissector crash
wnpa-sec-2015-53 - BT ATT disector crash
wnpa-sec-2015-54 - MP2T file parser crash
wnpa-sec-2015-55 - MP2T file parser crash
wnpa-sec-2015-56 - S7COMM disector crash
wnpa-sec-2015-57 - IPMI dissector crash
wnpa-sec-2015-58 - TDS dissector crash
wnpa-sec-2015-59 - PPI dissector crash
wnpa-sec-2015-60 - MS-WSP dissector crash

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

python-serial: bump to 3.0

Add hash file, switch to setuptools and change project's home page to
GitHub.

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

netbsd-queue: actually use the Config.in file

In commit e84fd04e88bcda2771847958b3b5153e476db45f, Yann added a
Config.in symbol for netbsd-queue in a brand new Config.in file. But
that Config.in file was never included anywhere, making it somewhat
useless.

Let's include it under Libraries -> Other, even if its actual
inclusion location doesn't matter much since it's just a hidden
Config.in option anyway.

Fixes:

  http://autobuild.buildroot.org/results/d23/d23a1c51c4d7a260364aeef576fdf5a407abc2f6/

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

minidlna, libsvg: do not use libjpeg directly, use jpeg

libjpeg should not be selected directly, as it is one implementation
of the virtual jpeg package. Consequently, this commit moves minidlna
and libsvg to use the jpeg virtual package instead.

The build has been tested with both libjpeg and jpeg-turbo variants.

Fixes:

  http://autobuild.buildroot.org/results/a12/a12760dea9e9d7ca62d14d0821539cc4b0c76835/

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

x11r7/xfont_*: add missing select BR2_PACKAGE_XFONT_FONT_UTIL

Fixes:

  http://autobuild.buildroot.org/results/d63/d63b45c24d01bfb3409e90569a0708f1d4af905d/

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

libcdaudio: disable on nios2, binutils assertion failure

Fixes:

  http://autobuild.buildroot.org/results/539/5397c51bc6915a082e873dee2d298f433172ef4d/

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

mesa3d: add missing select of xproto_glproto

When X.org is available, mesa3d has xproto_glproto in its
dependencies, but it forgets to select it. This commit fixes this
mistake, which fixes the following autobuilder failure:

  http://autobuild.buildroot.org/results/378/378bbf8520d67a67b5906ec177079542158a6be7/

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

collectd: fix incorrect select of libmodbus

BR2_PACKAGE_MODBUS does not exist, we meant to select
BR2_PACKAGE_LIBMODBUS.

Fixes:

  http://autobuild.buildroot.org/results/661/6617ea58424bf55640a70bee8fc3a6a8d098e3b2/

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

toolchain-external: select netbsd-queue for musl toolchains

Following the introduction of the check that target packages must have
their Config.in option enabled, we started to see failures related to
netbsd-queue. Yann fixed the internal toolchain case in commit
e84fd04e88bcda2771847958b3b5153e476db45f. This commit fixes the
similar issue, but for the external toolchain case.

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

core/pkg-virtual: do not check they are neabled

Virtual packages do not have a Kconfig symbol, so we must not check that
they are enabled.

Fixes:
    http://autobuild.buildroot.org/results/287/28713478cc6edf8e5d5c3e830fee86a42f0afa8d/
    ... and numerous similar failures ...

[Thomas:
 - use "$($(PKG)_IS_VIRTUAL)" != "YES" instead of !
   "$($(PKG)_IS_VIRTUAL)" = "YES"]

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/netbsd-queue: add Kconfig symbol like for all target packages

Select that package from musl, too, since it is a dependency.

Fixes:
    http://autobuild.buildroot.org/results/15c/15c9a80fb2754ed1866b59d5e62d02691b57834e/
    ... and numerous similar issues ...

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Sergio Prado <sergio.prado@e-labworks.com>
Cc: Luca Ceresoli <luca@lucaceresoli.net>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/c-libraries: need linux-headers

Now that we check that a target package in the _DEPENDENCIES of another
package has to be enabled in config, all target packages must have a
kconfig symbol.

Add a Kconfig symbol for linux-headers, and select it from the packages
that depends on it (C libraries).

Also remove the now-misleading comments "for legal-info" from the C
libraries.

Fixes:
    http://autobuild.buildroot.org/results/2a9/2a9e5d27b34357819b44f573a834da1ba5079030/
    ... and numerous similar 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>

iptables: bump to version 1.6.0

Add option for nftables compat utilies and for bpf compiler/nfsynproxy
configuration tool.

Drop symlink trickery since it's not required any more.

Switch homepage to proper one and drop the kernel v2.4+ note since
that's not supported by buildroot anyway.

Drop most patches (except musl build) since they're upstream.

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

package/xtables-addons: require kernel modules from Kconfig

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

package/ti-gfx: require kernel modules from Kconfig

And reorder 'depends' before 'select'.

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

package/rtai: require kernel modules from Kconfig

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

package/racehound: require kernel modules from Kconfig

And reorder 'select' after 'depends'

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

package/nvidia-driver: require kernel modules from Kconfig

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

package/linux-fusion: require kernel modules from Kconfig

And reorder 'depends' after 'bool'.

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

linux: add blind kconfig option to require kernel modules

Currently, packages that need the kernel to have support for laodable
modules have two ways to require it:

  - either the use the kernel-module infra, which does it automatically,

  - or they do not use it, and they need to require it manually by
    setting the corresponding Makefile variable; however, they must only
    set it when they are actually enabled, which makes for a slightly
    cumbersome and ugly code, like:

        ifeq ($(BR2_PACKAGE_FOO),y)
        LINUX_NEEDS_MODULES = y
        endif

Introduce a new blind Kconfig option that packages can select to signify
they need kernel modules. That Kconfig option is then used to set the
Makefile variable.

It makes it cleaner:

  - code is simpler (one Kconfig line instead of a Makefile if-block,

  - this is handled at the Kconfig level, which is where we usually
    handle such dependencies.

Packages will be updated in follow-up commits.

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

graph-depends: correct is_dep() comment

The uncached variant is called is_dep_uncached(), not is_dep_full().

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

graph-depends: optimize remove_transitive_deps()

For large configurations, the execution time of
remove_transitive_deps() becomes really high, due to the number of
nested loops + the is_dep() function being recursive.

For an allyespackageconfig, the remove_extra_deps() function takes 334
seconds to execute, and the overall time to generate the .dot file is
6 minutes and 39 seconds. Here is a timing of the different
graph-depends steps and the overall execution time:

  Getting dependencies:   42.5735 seconds
  Turn deps into a dict:   0.0023 seconds
  Remove extra deps:     334.1542 seconds
  Get version:            22.4919 seconds
  Generate .dot:           0.0197 seconds

  real 6m39.289s
  user 6m16.644s
  sys 0m8.792s

By adding a very simple cache for the results of is_dep(), we bring
down the execution time of the "Remove extra deps" step from 334
seconds to just 4 seconds, reducing the overall execution time to 1
minutes and 10 seconds:

  Getting dependencies:  42.9546 seconds
  Turn deps into a dict:  0.0025 seconds
  Remove extra deps:      4.9643 seconds
  Get version:           22.1865 seconds
  Generate .dot:          0.0207 seconds

  real 1m10.201s
  user 0m47.716s
  sys 0m7.948s

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Tested-by: Gustavo Zacarias <gustavo.zacarias@free-electrons.com>
[yann.morin.1998@free.fr:
  - rename is_dep() to is_dep_uncached(), keep existig code as-is
  - add is_dep() as a cached-version of is_dep_uncached()
  - use constructs more conform with 2to3
  - use exceptions (EAFP) rather than check-before-use (LBYL) to be more
    pythonist; that even decreases the duration yet a little bit more!
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

bcm2835: new package

[Thomas:
 - addition of 'depends on BR2_arm' as suggested by Yann.
 - addition of a comment in the Config.in help text that says that
   this library is a pure abomination.
 - fix the hash, since apparently upstream changes the released
   tarballs after they are published.
 - fix license, it's GPLv2, not GPLv2+.]

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

mongoose: bump to version 6.1

- provide only libmongoose.a
- rework package description

[Peter: move under Libraries->Networking]
Signed-off-by: Davide Viti <zinosat@tiscali.it>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/mke2img: set sparse_super for revision 1

The sparse_super feature is more efficient in terms of speed and
available space, so set it by default if supported, i.e. for revision 1.

This gives a huge speed gain for large resize2fs operations, especially
for expansion.

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