fs/common: get rid of post-target rules

The only users of post-target rules were ext2, cpio and initramfs.

Of those, ext2 and cpio were changed to use post-gen hooks, while
initramfs was not even using the generic rootfs infra and was fixed
to no longer reference post-target rules.

Besides, the comment in the infra was really misleading: it referenced
initramfs implying it was the sole user of that feature, even though
initramfs was not using the fs infra.

Furthermore, using post-target rules was inherently broken for top-level
parallel builds, because filesystems had to ensure the ordering by
themselves. Of the two real users of post-target rules (cpio and ext2),
one did enforce rules ordering (apparently correctly), while the other
forgot to do so.

We can get rid of post-target rules altogether, now.

Add a legacy check, to catch out-of-tree (e.g. br2-external) users of
post-target rules, and instruct them to switch to post-gen hooks instead.

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

fs/cpio: use a post-gen hook rather than a post-target rule

We already have a mechanism for running stuff after the filesystem is
generated, and that's called post-gen hooks.

Use those hooks.

Note: for cpio (and unlike ext2 previously), the dependency chain was
correct, in that the post-target rule correctly depended on the image
rule. Nonetheless, we still want to fix it for consistency.

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

fs/ext2: use a post-gen hook rather than a post-target rule

post-target rules are probably not resilient to parallel builds, given
that they do not depend on the image being generated first.

Beside, we already have a mechanism for running stuff after the
filesystem is generated, and that's called post-gen hooks.

Use those hooks.

Note: this basically reverts 75b6303 (rootfs-ext2: make the symlink as a
_POST_TARGET) since we've now re-introduced post-gen hooks.

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

fs/initramfs: cleanups, enhance comments

rootfs-initramfs is not using the generic fs infrastructure, because
there is virtually nothing to do to build the initramfs image: there is
no actual image to be built to begin with.

The only purpose of rootfs-initramfs is to ensure the rootfs.cpio image
is built and then that the Linux kernel is rebuilt with that rootfs.cpio
as initramfs source.

Using variables of the fs infra like if it were used is misleading. It
looked nice as long as there was the possibility that rootfs-initramfs
would one day use the fs infra. But there's no way that will happen any
time soon.

Furthermore, the linux' rule linux-rebuild-with-initramfs now already
depends on rootfs-cpio by itself, so we need not duplicate this
dependency in rootfs-initramfs.

Still, we want to advertise that the dependency is on rootfs-cpio, so
we get nice dependency graphs (and not expose the internal
linux-rebuild-with-initramfs rule to the users).

So, remove the variables and directly define the rules.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Romain Naour <romain.naour@openwide.fr>
Reviewed-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

linux: meddle not in the internals of filesystems

Currently, the rule to rebuild the Linux kernel with an initramfs
directly depends on the path of the file of the intermediate cpio image.

This is inherently "bad" from a purity point of view; linux.mk should
not have to delve into the fs internals.

Rather, make it directly depend on the "frontal" rule that generates the
cpio image.

Drop the comment for linux-rebuild-with-initramfs, it was misleading
(talking about generating "the initramfs list of files", which is not
what was done, since we use a cpio as source of initramfs, not a list of
files).

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

linux: split overly-long dependency line for readability

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

core: sort packages and eliminate duplicates in show-targets

Currently, enabling more than one filesystem image will make
'show-targets' list a few host packages more than once.

This is because all filesystem images add the same set of
host-packages to their dependencies, which are then added as-is
to the package list.

Thus, host-fakeroot, host-makedevs and, if needed, host-mkpasswd will
appear as many times as there are filesystem images enabled.

Fix that by sorting the package list, thus eliminating duplicates from
that list. Also sort the rootfs list for good measure. Sort the two
separately, so that rootfses are last.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Reviewed-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/efl: bump to 1.20.6

https://www.enlightenment.org/news/efl-1.20.6

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

musl: enable or1k architecture

With Linux kernel >= 4.13.x musl or1k can be used
with Qemu.

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

qemu-or1k: update to 4.14.x

Or1k support improved, f.e. musl can be tested inside qemu.

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

uclibc: enable NPTL for or1k

With 1.0.27 NPTL for or1k is improved and can be used
by default.

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

binutils: allow the selection of upstream binutils 2.29 for ARC

As of today quite a lot of ARC stuff got merged in upsream
such that we may finally use upstream realease for building
and even running Linux kernel and rootfs built with vanilla
binutils.

Essentially upstream release is still a bit behind our
GitHub-based releases but let's try to use vanilla version
and if we're lucky enough with just a small set of back-ported patches
we'll be able to use vanilla release(s) for everything in Buildroot.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
[Thomas: fixup the condition to select the special binutils site for ARC.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gcc: allow the selection of upstream GCC 7.x for ARC

Upstream gcc now has support for ARC, as of 7.x, with 7.2 being
usable, so let's allow its selection.

GCC from Synopsys GitHub still has a lot of fixes and improvements
most of which will be a part of the next upstream release but let's
see how it goes and if we see vanilla GCC doesn't require way too many
back-ported patches we'll be able to get rid of Synopsys GitHub here
at all.

As of today the only really required back-port is the one for non-uClibc
libc.

Back-ported from upstream master with help of:
-------------------->8--------------------
git format-patch 927c0132251f -1 -- . ':!gcc/ChangeLog' ':!libgcc/ChangeLog'
-------------------->8--------------------

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

libostree: new package

OSTree is an upgrade system for Linux-based operating systems

Signed-off-by: Marcus Folkesson <marcus.folkesson@gmail.com>
[Thomas: use SPDX license code, fix license, it's GPL-2.0+ not GPL-2.0.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Revert "package/qt5webkit-examples: remove obsolete package"

This reverts commit 940bfe2f4336a3fea1ec6805db1f90ce83bd4a78 with the
following changes:

 - new hashes for 5.6.3 and 5.9.1 (there is no 5.9.2).
 - BR2_PACKAGE_QT5BASE_LICENSE_APPROVED has been removed
 - add GPL-3 to licences
 - uses SPDX identifier for licences
 - add missing dependencies (qt5base, qt5webkit, qt5declarative)
 - removes /usr when qmake host-tools is invoked at configure

Signed-off-by: Gaël PORTAY <gael.portay@savoirfairelinux.com>
[Thomas: add entry in the DEVELOPERS file.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/x11r7/xapp_xfs: bump version to 1.2.0

xfs was ported to use libXfont 2.x instead of 1.x:
https://cgit.freedesktop.org/xorg/app/xfs/diff/configure.ac?id=fedc90e8e3f6c36204de5562bfb726270064a4f1

Added all hashes provided by upstream.

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

rauc: bump to version 0.2

Version 0.2 of rauc provides several bug fixes and minor feature
enhancements.

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

toolchain/buildroot: add comment for glibc NaN-2008 headers requirement

Commit 4a5140ecf (toolchain/buildroot: glibc requires kernel headers >=
4.5 with NaN-2008) added a restriction on kernel headers for glibc when
the architecture is using naN-2008.

However, such a restriction is usually associated to a comment explaining
the restriction, so the user knows what is happening.

That comment was forgotten in 4a5140ecf. Add it now.

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

package/qt5: simplify and cleanup _SITE variables

The variable QT5_SNAPSHOTS_SITE was introduced in commit 9d5ac9698 to
fetch sources of obsolete modules. It was used to download qt5webkit
tarballs.

The variable is not referenced anymore since the qt5webkit tarballs are
downloaded from the Qt community release site.

Since the introduction of qt-5.9 in aa1c40ba3, the QT5_SITE variable
was duplicated for no good reason.

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

musl: bump to 1.1.18

Patch 0002-arm-atomics-asm-with-new-binutils.patch is upstream as of
commit b261a24256792177a5f0531dbb25cc6267220ca5.

Patch 0003-Makefile-include-per-arch-Makefile-before-Makefile is
upstream as of commit 45ca5d3fcb6f874bf5ba55d0e9651cef68515395.

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

uclibc: update to 1.0.27

All patches are upstream.
C-sky support added. Lot of or1k NPTL bugfixes.

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

package/solatus: new package

Sample quest and testing quest are not included in SOLARUS_LICENSE
since it's notinstalled by default.

Remove Werror when building with debugging symbols.

Add libpng runtime dependency since solarus games require png image
support.

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Julien BOIBESSOT <julien.boibessot@armadeus.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/physfs: bump to 3.0.1

Remove upstream patch.

Use the official archive and add a hash file.
Also add a license file hash.

Remove LGPL.txt license file since src/lzma directory has been removed
from the sources.

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Julien BOIBESSOT <julien.boibessot@armadeus.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gnutls: bump to version 3.5.16

Add link to tarball signature.

Add license hashes.

Cc: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libmicrohttpd: bump to version 0.9.57

Drop libgcrypt dependency for SSL support; with newer version of gnutls
libgcrypt is not needed.

Drop the LIBS fix for static linking with gnutls that has been
introduced in commit 2d3d6d258bb (libmicrohttpd: fix static linking with
gnutls). gnutls does not support static linking anymore.

Fix a comment typo.

Add license hash.

Cc: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

test-pkg: filter empty lines and comments in CSV file

In preparation for the addition of comments in the CSV file listing
toolchain configurations, we filter out such lines when reading the
CSV file in utils/test-pkg.

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>

utils/genrandconfig: filter empty lines and comments in CSV file

In preparation for the addition of comments in the CSV file listing
toolchain configurations, we filter out such lines when reading the
CSV file in utils/genrandconfig.

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>

core: check files are not touched by more than one package

Currently, we do nothing about packages that touch the same file: given
a specific configuration, the result is reproducible (even though it
might not be what the user expected) because the build order is
guaranteed.

However, when we later introduce top-level parallel build, we will no
longer be able to guarantee a build order, by the mere way of it being
parallel. Reconciliating all those modified files will be impossible to
do automatically. The only way will be to refuse such situations.

As a preliminary step, introduce a helper script that detects files that
are being moified by two or more packages, and reports them and the
impacted packages, at the end of the build.

The list being reported at the end of the build will make it prominently
visible in autobuilder results, so we can assess the problem, if any.

Later on, calling that helper script can be done right after the package
installation step, to bail out early.

Thanks Arnout for the pythonist way to write default dictionaries! ;-)

Note: doing it in python rather than a shell script is impressively
faster: where the shell script takes ~1.2s on a minimalist build, the
python script only takes ~0.015s, that is about 80 times faster.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Baruch Siach <baruch@tkos.co.il>
Cc: Peter Seiderer <ps.report@gmx.net>
[Thomas: rename script without .py extension.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

core/pkg-generic: store file->package list for staging and host too

Currently, we store the list of files installed in target/ and associate
each of them to the package that installed it.

However, we sometimes may need to know what package installed which file
in staging/, for example to debug header collision, or in host/, to
debug what package installed what host tool.

Enhance the step instrumentation to also generate the list for staging/
and host/.

We maintain backward compatibility, for external scripts that wanted to
parse the previously existing list, by not renaming the target-related
package list. Only the staging- and host-related lists are named after
staging and host.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
[Thomas: fix missing word in .mk comment.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

core/pkg-generic: redirect only when listing package's installed files

There is no need to redirect again and again for each new file added to
the list; we can just redirect once and for all.

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>

core/pkg-generic: remove intermediate file-list files

To compute the list of files added by a package, we first store the list
of files before the install, do the install, list the files after the
install, and finally compare the two lists. The two lists are stored in
dot-files, hidden in the package's build dir.

We currently keep those two files, and only list the files installed in
target/

In followup patches, we'll also list files installed in staging/ as well
as files installed in host/.

Rather than add even more internal, hidden files in the package build
dir, we'll just re-use the same two temporary files to store the before
and after lists.

So, remove them after the comparison is done.

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>

rygel: new package

Rygel is a home media solution (UPnP AV MediaServer) that allows you to
easily share audio, video and pictures to other devices.

Additionally, media player software may use Rygel to become a
MediaRenderer that may be controlled remotely by a UPnP or DLNA
Controller.

Rygel achieves interoperability with other devices in the market by
trying to conform to the very strict requirements of DLNA and by
converting media on-the-fly to formats that client devices can handle.

Most Rygel functionality is implemented through a plug-in mechanism.

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[Thomas:
 - use SPDX license codes
 - add hashes for license files
 - move Config.in comment at the end of the Config.in file to not
   break the indentation of the sub-options.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gupnp-av: fix static library linking with gupnp

gupnp and gupnp-av defines two internal functions with the same name:
xml_util_get_element and xml_util_get_child_element_content, as a result
an application such as rygel can't be build statically.
To fix this issue, rename both functions as
xml_util_get_element2 and xml_util_get_child_element_content2

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

gnupg2: bump to version 2.2.2

Cc: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gnupg2: don't preserve the old gpg2 executable name

Commit e82fadab236a (gnupg2: bump to version 2.2.0) added a configure
option to keep the old 'gpg2' executable name to avoid conflict with the
gnupg package. It turns out that gnupg depends on !BR2_PACKAGE_GNUPG2
since commit 2cadb26e6d4 (gnupg: make gnupg and gnupg2 mutually
exclusive). Drop this configure option.

Rename the config option that controls the removal of gpgv2, now gpgv,
to match the new name. Add legacy config symbol handling.

Cc: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

qt5webengine: enable build for 5.6.3

The build was broken in 5.6.2 and was disabled.

It compiles fine since 5.6.3.

The two additional patches are useful at run-time.

The first one avoid the need to specify the path to the SSL certificate
directory (using an additional environment variable).

The second one is the same used in 5.9.x (plus resolved conflicts). It uses the
process's context to get handles on EGL and GLESv2 libraries. Those libraries
are linked to Qt WebEngine at compile time.

The patch is particularly usefull for RPI boards since the raspberrypi userland
package does not provide the libEGLv2.so.2 and libGLES.so.1 symlinks. Both
library paths are hardcoded in Qt WebEngine.

Signed-off-by: Gaël PORTAY <gael.portay@savoirfairelinux.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

postgresql: bump to 10.1

- Add a hash for the license file.

- PostgreSQL 10.0 and above will default to checking for /dev/urandom if an
  SSL library is not found, which will fail when cross compiling.
  Since /dev/urandom is guaranteed to be provided on Linux systems,
  add ac_cv_file__dev_urandom=yes to the configure environment if a SSL library
  is not selected.

Signed-off-by: Adam Duskett <Adamduskett@outlook.com>
[Thomas: minor tweaks to the /dev/urandom comment in the .mk file.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/avahi: bump version to 0.7

Added license hash, removed license text in avahi.mk added in 2006:
https://git.buildroot.net/buildroot/commit/package/avahi?id=21e97d3e155f4d9604f729698cdfff3c40eefbe7

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

dbus: bump to version 1.12.2

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

package/{mesa3d, mesa3d-headers}: bump version to 17.2.6

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

python-jaraco-classes: add missing dependency

python-jaraco-classes needs host-python-setuptools-scm.

Fixes:
http://autobuild.buildroot.net/results/3a7000002b86663a56c0341d192f131a4a8f4936
http://autobuild.buildroot.net/results/553d6fdeb6ea5359b1c7d4eeb844da7a588f02e7

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

package/samba4: bump version to 4.7.3

Rebased 0001-libreplace-disable-libbsd-support.patch.

Added iconv-related entry to samba4-cache.txt to fix configure error:
Cross answers file
 /home/buildroot/test-samba4/armv5-ctng-linux-gnueabi/build/samba4-4.7.3/cache.txt
 is incomplete

with this defconfig

BR2_TOOLCHAIN_BUILDROOT_WCHAR=y
BR2_PACKAGE_LIBICONV=y
BR2_PACKAGE_SAMBA4=y
BR2_PACKAGE_SAMBA4_AD_DC=y
BR2_PACKAGE_SAMBA4_ADS=y
BR2_PACKAGE_SAMBA4_SMBTORTURE=y

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

linux-firmware: Bump to the latest version

Update to commit 17e6288135d4.

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

utils/diffconfig: add diffconfig utility

Diffconfig is a simple utility for comparing two configuration files.
See usage in the script for more info.

Borrowed from the Linux kernel source code and adapted to Buildroot.

Signed-off-by: Marcus Folkesson <marcus.folkesson@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

python-schedule: new package

Signed-off-by: Joseph Kogut <joseph.kogut@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

hostapd: keep previous patches when DRIVER_RTW set

Changed HOSTAPD_PATCH= to HOSTAPD_PATCH+= to keep previously added
patches.

Signed-off-by: Alexander Mukhin <alexander.i.mukhin@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/eudev: bump version to 3.2.5

Added license hash.

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

package/php: bump version to 7.1.12

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

package/nano: bump version to 2.9.0

Removed patch applied upstream, added license hash.

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

axel: bump to version 2.16

- Change upstream URL.
- Add optional dependency on libressl/openssl.
- Switch to autotools-package.

Signed-off-by: Ismael Luceno <ismael@iodev.co.uk>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/python-mwclient: require pytest-runner only when necessary

Fixes
http://autobuild.buildroot.net/results/09c/09c0e590448f1fc069d1d00564c202d2b009d59a/
http://autobuild.buildroot.net/results/79e/79ef04d2e8c91231028e23f4bb35df685efa55ca/
http://autobuild.buildroot.net/results/7a4/7a4a4aecbe6dd315062bafe97b291645d3d6ced1/

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

arch/arm: add armv8.1a cores

The armv8.1a generation is a cumulative extension to armv8a. It adds new
extensions, and makes some previously optional ones now mandatory.

Since gcc correctly enables the appropriate extensions based on the core
name, we don't really need to introduce a separate config for armv8.1a,
and we can piggyback on armv8a.

All those new cores are aarch64 only (gcc fails to build in arm mode).

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>

arch/arm: add some non-cortex armv8a cores

Some need gcc-5, some gcc-6 and some gcc-7.

The thunderx familly does not build in 32-bit mode (gcc complains
that the CPU is unknown, and even gcc master only knows them as
aarch64-only).

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>

arch/arm: add some armv8a cortex variants

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>

arch/arm: add cortex-A32

The cortex-A32 is an armv8a core, but it lacks the optional AArch64
extensions, so can only work in 32-bit mode.

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>

arch/arm: armv8 is really armv8a

For armv8, there are different profiles: A, M and R, like there is for
armv7.

So, rename our internal symbol to mirror what we do for armv7.

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>

arch/arm: simplify hiding non 64-bit cores

Now that the cores are all oredered correctly, we can just enclose all
the non 64-bit cores inside a big if-block, rather than have each of
them have the dependency.

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>

arch/arm: re-order cores choice

Currently, the logic for ordering the ARM cores in the choice is all
but obvious. ;-)

Reorder the choice by architecture generation, starting with armv4,
ending with armv8.

Add a comment before each generation, just for ease of use. Add a
separate comment for armv7a and armv7m.

Finally, order cores alphabetically inside the same generation (except
for armv7m cores, listed after all armv7a cores).

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>

toolchain/buildroot: glibc requires kernel headers >= 4.5 with NaN-2008

From sysdeps/unix/sysv/linux/mips/configure.ac in glibc:

if test -z "$arch_minimum_kernel"; then
  if test x$libc_cv_mips_nan2008 = xyes; then
    arch_minimum_kernel=4.5.0
  fi
fi

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Romain Naour <romain.naour@openwide.fr>
Cc: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

arch/mips: inverse the mfpxx logic

Currently, the possibility to choose the floating point mode (32, xx or
64) is conditional on having a sufficiently recent gcc version.

Which means that the architecture selection depends on the gcc version.

But that's opposite to what we've always done in Buildroot: the software
versions are conditional to the architecture options. There is nothing
we can do about the hardware: it is there, we can't change it, while we
can restrict ourselves to using software that is working on said
hardware.

Thus, we inverse the logic, to move the condition onto the software
side: whenever mfpxx is selected, we restrict the toolchain selection to
at least a gcc-5.

And now, the blind BR2_TOOLCHAIN_HAS_MFPXX_OPTION symbol is no longer
needed, so we get rid of it.

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

arch/mips: inverse the NaN logic

Currently the possibility to choose the NaN encoding is conditional to
having a sufficiently recent gcc version.

Which means that the architecture selection depends on the gcc version.

But that's opposite to what we've always done in Buildroot: the software
versions are conditional to the architecture options. There is nothing
we can do about the hardware: it is there, we can't change it, while we
can restrict ourselves to using software that is working on said
hardware.

Thus, we inverse the logic, to move the condition onto the software
side: whenever NaN-2008 are selected, we restrict the toolchain
selection to at least a gcc-4.9.

But now, the option with the NaN type is always set, so we must enclose
the code in gcc.mk inside a HAS_NAN_OPTION condition, as is already done
for the external toolchain case.

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

package/gcc: slight cleanup and reorg in remaining arch depends

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>

arch/arm: some variants need different gcc versions

Take the conditions currently specified in the gcc version choice.

Also, the conditions explained in the commit log for 78c2a9f7 were not
all properly applied, especially the a57-a53 combo needs gcc-6, but
78c2a9f7 forgot to add the condition to gcc-4.9.

gcc-4.9 was excluded for cortex-a17 and a72, but the CodeSourcery
external toolchain, which uses 4.8, was not excluded for those two
cores. Now it is.

Remove the arch condition from gcc and the external toolchains.

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>

arch/mips: some variants need different gcc versions

We use the conditions currently expressed in the gcc version choice.

We leave the musl vs mips64 conditions in gcc, because the "fault"
really is on gcc, which does not recognise the mips64+musl tuples,
so the fix lies within gcc, and the current conditions are fitting.

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>

arch/bfin: needs gcc >= 6

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>

toolchain/external: hide versions too old for the current arch

Hide the toolchains if the arch requires a gcc version more recent
than the one they provide.

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>

toolchain/external-custom: hide versions too old for the current arch

When an architecture expresses a requirement on the gcc version, limit
the version choice in the custom external toolchain.

The rationale being that there is no point in offering that version to
the user if we know before-hand that the gcc version will not work for
that architecture.

All versions below the minimum we support is just made conditional to
that minimum as well, including the "older" entry.

However, this means that the "older" entry is no longer available when
the architecture requires a minimum gcc version. A user who wants to use
a toolchain with a gcc older than the minimum will have no choice but to
realise the toolchain is not suitable (or lie and we would catch that
when checking the gcc version anyway).

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/gcc: hide versions too old for the current arch

Begin the conversion from hard-coded dependencies on architectures, to
architecture-specified version requirement, using the newly introduced
BR2_ARCH_NEEDS_GCC_AT_LEAST_XXX symbols.

Hard-coded dependencies will be removed progressively, as archs are
individually converted over to using the new symbols.

We do not change the architecture-specific versions for ARC and
OpenRISC, because there is no point in doing so for those, as they use
special, non-upstream versions anyway.

Signed-off-by: "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>

arch: introduce minimal required gcc version

Some CPU variants require that a recent-enough gcc be selected. For
example, ARM's cortex-a35 requires gcc-5, while cortex-a73 requires
gcc-7. Same goes for other architectures, of course.

Currently, we hard-code every such conditions in the gcc version choice,
as well as in the individual external toolchains.

However, as we add even more CPU variants, the conditions are getting
more and more complex to write and maintain.

Introduce new symbols, that architectures can select if they have a
specific requirement on the gcc version. gcc and external toolchains
can then properly depend on those symbols.

The burden of maintaining the requirements on the gcc version now falls
down to the architeture, instead of being split up in gcc and all the
external toolchains.

As the oldest gcc version to handle, we can either choose gcc-4.9, as
the oldest version we support in our internal toolchain, or choose
gcc-4.8, as the oldest external toolchain we support (except for the
custom ones, but they'll be handled specifically in upcoming changes).
We choose to go back up to gcc-4.8.

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>

hostapd: add support for Realtek driver

Since kernel drivers for Realtek wireless chips use non-standard
interfaces, upstream hostapd does not support them. One have to apply
an external patch for hostapd to work with these chips. See:
https://github.com/pritambaral/hostapd-rtl871xdrv

A configuration option is added to enable support for Realtek chips,
and it's turned off by default.

Signed-off-by: Alexander Mukhin <alexander.i.mukhin@gmail.com>
Tested-by: Angelo Compagnucci <angelo.compagnucci@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

mtd: be explicit about config options for the host

Instead of letting auto-detection do its job, be explicit about the
fact that we want the JFFS2 and UBIFS utilities when building the host
variant of mtd.

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

python-cheroot: bump version to 5.10.0

Fix licence file name and add checksum.

Fixes:
http://autobuild.buildroot.net/results/94d/94dee5b01e0f7693d4e6e05074e7d93e92236e2c

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

libmaxminddb: bump to version 1.3.0

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

tinycbor: bump to version 0.4.2

- Remove patch (already in version)
- Add LICENSE hash
- Change github repo name from 01org to intel

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

libupnp: bump to version 1.6.24

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

motion: fix build on musl

Some toolchains (musl) have pthread_setname_np but not pthread_getname_np.
The first patch fixes check on pthread_setname_np and the second one add
a check for pthread_getname_np

Fixes:

  http://autobuild.buildroot.net/results/65534775c5977e2424c5f5c63c46f9d0f39d7e1b

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

DEVELOPERS, .gitlab-ci.yml: fix typoes in imx6sx-sdb board name

The addition of a new defconfig in commit
459e3320dc234f96b93661bed1a753e477aeab17 ("configs/imx6sx-sdb: Add new
defconfig") introduced changes in the DEVELOPERS file and
.gitlab-ci.yml file that were not matching the defconfig name. This
commit fixes those issues.

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

package/git: add optional support for pcre2

Renamed --with-libpcre to --with-libpcre1. Currently --with-libpcre
activates pcre1 support but this can change in the future to pcre2:
https://github.com/git/git/blob/df7fd961a9d9ba60840ffc0868d36cc3db2aec74/configure.ac#L258

Please note that we cannot use --with-/--without because it will lead
to an error reported by configure, for example

    --with-libpcre1 --without-libpcre2

will produce

configure: error: Only supply one of --with-libpcre1 or --with-libpcre2!

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

package/python-networkmanager: Bump to version 2.0.1

Added new runtime dependency on python-six.

Dropped dbus-python as build time dependency (it's only runtime
dependency).

Signed-off-by: Petr Vorel <petr.vorel@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

refpolicy: bump to 2.20170805

In addition to a simple bump, the following extra changes have occured:
  - Change the refpolicy site to the official release URL.
  - Remove REFPOLICY_SITE_METHOD and REFPOLICY_GIT_SUBMODULES as the contrib
    submodule is included in the release tarball.
  - Refpolicy is now compatible with python3, as such, remove host-python.
    from the dependencies and add a check for python3 or python in it's place.
  - Add upstreamed 0001-fix-regex-escape-sequence-error.patch to fix building
    against python3.6.
  - Add sha256 license hash to hash file.

Signed-off-by: Adam Duskett <Adamduskett@outlook.com>
Tested-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/python-mwclient: bump version to 0.8.6

Version 0.8.2 added OAuth support so we need python-requests-oauthlib
as runtime dependency from now on. This package also has a runtime
dependency on python-requests so all we need is to update the select
command in Config.in.

Removed patch applied upstream.

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

package/python-requests-oauthlib: new package

Needed for python-mwclient version bump to 0.8.6.

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

package/python-oauthlib: new package

Needed for python-mwclient version bump to 0.8.6.

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

uboot: Use BR2_TARGET_UBOOT_SPL_NAME for TPL name

Since the BR2_TARGET_UBOOT_SPL_NAME option accepts a space-separated
list of binaries, the same option can be reuses for TPL binaries as
well. This commit updates the string and help text to indicate that
the same option can be used for SPL and TPL.

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

package/waylandpp: new package

Building waylandpp for the target requires a wayland-scanner++ binary
built for the host.

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
[Thomas: use 'depends on BR2_PACKAGE_WAYLAND' instead of a select.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

python-cherrypy: bump version to 12.0.1

Add licence file checksum and new runtime dependencies.

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

python-cheroot: bump version to 5.9.0

Omit 5.9.1 as is has broken Python files.

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

python-jaraco-classes: new package

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

package/dovecot-pigeonhole: bump version to 0.4.21

Added license hash.

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

package/dovecot: bump version to 2.2.33.2

Added license hashes.

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

configs/mx53loco: Bump U-Boot and kernel versions

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

libupnp18: new package

Branch 1.8.x of libupnp is not compatible with branch 1.6.x so add a
dedicated package and make it depends on !BR2_PACKAGE_LIBUPNP as
suggested by Thomas Petazzoni and Arnout Vandecappelle during review
of "libupnp: add 1.8.3 version" patch.

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[Thomas: fix the dependencies of the Config.in comment.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

netcat-openbsd: add source and license hashes

Signed-off-by: Gaël PORTAY <gael.portay@savoirfairelinux.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

configs/imx6-sabreauto: Bump kernel and U-Boot versions

Bump kernel to version 4.14 and U-Boot to 2017.11.

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

configs/imx6ulevk: Add new defconfig

Add support for imx6ulevk_defconfig that allows booting a mainline
kernel and mainline U-Boot.

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

configs/armadeus_apf28: bump kernel version to latest 4.9 LTS

As there is now a defconfig inside kernel sources for mxs platforms, use it.

Signed-off-by: Julien BOIBESSOT <julien.boibessot@armadeus.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

openntpd: add optional dependency on LibreSSL

OpenNTPD looks for a crypto library with tls_config_set_ca_mem, which
is available in LibreSSL but not OpenSSL. If tls_config_set_ca_mem is
found, crypto support is added to the build.

Because this is not currently checked, crypto support might not be
added due to the order of the build. Add a small check to see if
BR2_PACKAGE_LIBRESSL is selected, and if so, add a dependency for it.

Also add a note about this in the Config.in help section.

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

openntpd: update to 6.2p3

Other changes:

  - Remove upstream patch.
  - Change locally computed hash to hash from the SHA256 URL.
  - Add sha256sum for license file.

Note that we need to keep <pkg>_AUTORECONF = YES even though the patch
touching the Makefile.am has been removed, because our libtool patch
doesn't apply on the bundled ltmain.sh.

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

util-linux: explicitly use /usr/{bin, sbin, lib}, with merged usr

Prevent the installation from attempting to move shared libraries from
${usrlib_execdir} to ${libdir} if both paths are the same, which leads
to error messages like this:

  mv: '$(TARGET_DIR)/usr/lib/libfoo.so.1' and '$(TARGET_DIR)/lib/libfoo.so.1' are the same file

That error is not fatal but let's avoid possible future problems.

Signed-off-by: Carlos Santos <casantos@datacom.ind.br>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

python-raven: new package

Changes since v2:
- remove newlines from Config.in
- add LICENSE hash

Changes since v1:
- reword commit to match upstream style
- fold the DEVELOPERS modifications into the same commit
- add Config selections for required python core modules
- bump version to 6.3.0

Tested in a clean chroot, events are generated for exceptions, and
reported to the server configured for the client.

Signed-off-by: Joseph Kogut <joseph.kogut@gmail.com>
Reviewed-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>