test-pkg: get configs from in-tree toolchain-configs.csv

Now we have the toolchain config fragments in the buildroot directory
itself, it is no longer necessary to fetch it from the toolchain URL.

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

support: add autobuild toolchain config fragments

We currently have a list of toolchain configurations that are used by
the autobuilders at [1]. However, this makes it a little more difficult
for people to use these configurations, and also to have a different
list of configurations for different branches. For example if a new
architecture is introduced, the 2017.02.x branch doesn't have support
for this architecture yet so it shouldn't try to run those configs.

Therefore, include the autobuild config fragments directly in Buildroot,
so they can be branched together with the rest. We create a new
directory under support/ to store them.

Generated with
wget -nd --no-parent --recursive http://autobuild.buildroot.net/toolchains/configs/

The index.html file is removed.

The toolchain-configs.csv file is adapted so the URLs become relative
paths pointing to the config fragments.

[1] http://autobuild.buildroot.net/toolchains/configs/toolchain-configs.csv

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

genrandconfig: use minimal.config

This has a number of side-effects which must be handled.

The lines in minimal.config may be overridden by the random lines added
by amending the configuration, so is_toolchain_usable() shouldn't take
those into account, or indeed the random lines added. Therefore, make
a copy of the config before appending minimal.config and the random
lines. While we're at it, rename the variable to the more appropriate
toolchainconfig.

minimal.config sets BR2_INIT_NONE=y, but we really also want to test
with BR2_INIT_BUSYBOX=y. Therefore, add a random line to use the
busybox init system. We set its probability rather high. The
probabilities of systemd and eudev are increased since they're now
in the else branch of BR2_INIT_BUSYBOX, which halves the probability
that we even get there.

We now also generate configurations without busybox. Previously,
busybox was almost always selected due to BR2_INIT_BUSYBOX=y. Only if
systemd is selected there was a chance to build without busybox.

We now set BR2_SYSTEM_BIN_SH_NONE=y, the other /bin/sh options are
never tested. However, this is not really something that is relevant
to test in the autobuilders.

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

genrandconfig: fix the case when outputdir is 'output'

When outputdir is 'output' (the default), genrandconfig didn't work
correctly because it expects the configfile in outputdir, while
Buildroot puts it in the buildroot directory.

Fix this by explicitly checking if outputdir == buildrootdir/output.
Because abspath is used for both paths, string comparison works
reliably.

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

genrandconfig: calculate configfile only once

The path to the .config file is calculated in several places - replace
it with a single calculation, and pass configfile as an argument
to is_toolchain_usable and fixup_config. These functions also don't
need outputdir any more.

This makes it easier to fix the case when configfile is not in
outputdir.

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

genrandconfig: pass outputdir and buildrootdir as arguments

The --instance argument is just an artifact of genrandconfig's
history as part of autobuild-run. It is much more logical to pass
the output directory and the buildroot directory as arguments, with
sane defaults.

This also allows us to remove the hack of creating a symlink in the
instance directory if it doesn't exist yet.

Note that the default outputdir 'output' doesn't work yet, because in
that case Buildroot will put the config file in the buildroot directory
instead of the output directory. This will be fixed in a follow-up
patch.

After this change, the script should be called from autobuild-run as:

    subprocess.call([os.path.join(srcdir, "utils/genrandconfig"),
                     "-o", outputdir, "-b", srcdir,
                     "--toolchains-url", kwargs['toolchains_url']],
                    stdout=devnull, stderr=log)

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

genrandconfig: calculate buildrootdir in __main__

This prepares for passing buildrootdir as an argument.

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

genrandconfig: calculate outputdir in __main__

This prepares for passing outputdir as an argument.

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

genrandconfig: verbose output and use stderr

The output of genrandconfig is currently very terse, which is annoying
for debugging the script or generally seeing what is going on. Also the
timing information added by log_write isn't very useful when the script
is used stand-alone.

In the new setup, (verbose) output goes to stdout and error output goes
to stderr. Also the "INFO: generate the configuration" message is
eliminated - it should go in the autobuild-run script.

We also add an explicit message when a toolchain can't be used after
the first defconfig, otherwise autobuild-run will just silently
restart.

Note that, since the output of make is no longer redirected to
/dev/null, we get one more message on stderr that will be recorded in
the autobuilder's log file: KCONFIG_SEED=0xXXXXXXXX.

This approach allows us to optimise the error handling to use
exceptions, where appropriate, which can be caught at the top level and
converted to an error message to stderr. This, in turn, allows us to use
subprocess.check_call, which eliminates a lot of conditions.

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

genrandconfig: move instantiation of SystemInfo down

The SystemInfo class is instantiated globally and passed down to all
functions, but it is really only used in fixup_config. So instead,
instantiate it there.

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

genrandconfig: replace kwargs with explicit arguments

kwargs is a left-over from the use of docopt, it's better to use
argparse's Namespace object directly.

In addition, most functions use just one or two fields of args, so
these can just as well be passed directly as arguments to the function.
Particularly for outputdir it doesn't make sense to reconstruct it all
the time.

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

genrandconfig: fix (some) pep8 warnings

Warnings fixed:
E731 do not assign a lambda expression, use a def
 -> urlopen_closing is defined with a def. urlopen is not used
    elsewhere so inlined.
E302 expected 2 blank lines
E501 line too long
 -> long lines due to a long string are NOT split
E701 multiple statements on one line (colon)
E722 do not use bare except'
 -> use "except Exception", so KeyInterrupt and SystemExit are still
    passed. We never intended to catch those.
E741 ambiguous variable name 'l'
 -> variable name is replaced with the much more descriptive
    toolchains_csv
E271 multiple spaces after keyword
E231 missing whitespace after ','

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

genrandconfig: use subprocess.check_output instead of Popen

Popen is more complicated and more difficult to understand.

check_output raises an exception if the exit code is non-zero, but
that's probably what we want if ldd can't be executed.

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

utils/genrandconfig: new script

This script will be used by the autobuild-run script to generate the
configuration to test. It is put in the utils directory because it can
also be called directly to allow users to test things.

For now, it is a direct copy of the relevant functions from the
autobuild-run script. The only changes are:
- unneeded import statements are removed;
- code/decode wrappers are limited to decode_byte_list;
- __main__ handling is added.

For now, the only supported arguments are the ones needed for
autobuild-run.  Follow-up patches will refactor things and also change
the way the script is called. In this version, it can be called from the
autobuild-run script as:

  subprocess.call([os.path.join(srcdir, "utils/genrandconfig"),
                   "-i", str(kwargs['instance']),
                   "--toolchains-url", kwargs['toolchains_url']],
                  stdout=log, stderr=log)

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

package/leptonica: bump version to 1.74.4

Removed patch 0001, not needed anymore after
https://github.com/DanBloomberg/leptonica/commit/4476d162cc191a0fefb2ce434153e12bbf188664
fixes the problem in a similar way.

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

paxtest: install helper applications/libs to /usr/lib/paxtest

RUNDIR installs the test applications and shared libraries into a
different folder then /usr/lib. This is desired as there are a lot of
test apps which would get installed into /usr/lib without
organization. Instead, set RUNDIR=/usr/lib/paxtest to install the test
apps and libraries in a sub folder. The genpaxtest script accounts for
this path and handles a LD_LIBRARY_PATH update as part of the paxtest
script's exeuction.

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

DEVELOPERS: add Ricardo Martincoski for support/testing

While at it, also move my professional entry near my personal one.

Signed-off-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/lcdapi: bump version to 0.11

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

package/lbreakout2: bump version to 2.6.5

Changed _SITE to https, added upstream md5 & sha1 hashes.

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

dieharder: fix link issue with inline function not declared static

Fixes:

  http://autobuild.buildroot.net/results/b629754c6a820446ff38df8202ea1ed0041bc4ac
  http://autobuild.buildroot.net/results/e02325e06866618d9d3ee90600dc3326465c56a1
  http://autobuild.buildroot.net/results/c1db73dcb25ea1db4be0f9d6ce2bf2d02f5bd5bb
  http://autobuild.buildroot.net/results/bd93120ee7cbfeb4fe7cbcd7f845f131743caf05
  http://autobuild.buildroot.net/results/273ba504de31bc17fd41e91ee5d6c0b34797a4f9
  http://autobuild.buildroot.net/results/37920b26f9c4853a0d620eb4a33b50b53e548888
  http://autobuild.buildroot.net/results/ee668405ed234fbbd644a01d49e8d9d41d216cf6
  http://autobuild.buildroot.net/results/5b76d62ad03d0cbe483792b32ea14ce7d7432983
  http://autobuild.buildroot.net/results/cf08d42be8fcb659d59288e2cedf3f18b660e8a6
  http://autobuild.buildroot.net/results/e1309fd2eea5daf854f4314b92ec441092239cd5

Signed-off-by: Julien Viard de Galbert <julien@vdg.name>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

zstd: new package

Add package to provide Zstandard compression tools
(see https://facebook.github.io/zstd)

Minimal config snippet for utils/test-pkg is as follows:

BR2_PACKAGE_ZSTD=y

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
[Thomas:
 - use "config" instead of "menuconfig" in Config.in
 - add missing final newline in Config.in
 - pass DESTDIR=$(TARGET_DIR) only at install time
 - wrap too long lines in the .mk file
 - remove useless empty newline at end of .hash file.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/joe: bump version to 4.4

Removed patch 0001 after bessel functions in uClibc were enabled by
default:
https://git.buildroot.net/buildroot/commit/package/uclibc?id=de56d58e5d26d78161ca3b7a1a2a4decfe4e89db

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

nginx-upload-module: add openssl lib dependency

When building against nginx 1.11.2+, requires additional
-lcrypto dependency. (Nginx changed crypto approach
and dependencies for openssl are no longer default)

Upstream pull request:
https://github.com/vkholodkov/nginx-upload-module/pull/93

Fixes:
http://autobuild.buildroot.net/results/e25f8ba8d8743e47a77707cf582e58477e1860e6
http://autobuild.buildroot.net/results/c4aa6e22033be934bd311cd9761f646d44618dc0
http://autobuild.buildroot.net/results/75e935d248b46df11c32c95526423a2287ac8c62

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/keyutils: bump version to 1.5.10

Upstream does not provide hashes anymore, replace with self-computed
sha256 hash.

Removed patches applied upstream, renumbered remaining patches:

0001-allow-building-of-the-shared-library-to-be-suppressed.patch
https://git.kernel.org/pub/scm/linux/kernel/git/dhowells/keyutils.git/commit/?id=a4deb71ddc05e951c8be8d46615beed9d408a5c8

0004-Makefile-for-buildroot.patch
https://git.kernel.org/pub/scm/linux/kernel/git/dhowells/keyutils.git/commit/?id=d83b9b827c6c4bc2377dfa073cf5c837b87465e8

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

package/kexec: bump version to 2.0.15

Removed patch applied upstream:

0001-vmcore-dmesg-Define-_GNU_SOURCE.patch
https://git.kernel.org/pub/scm/utils/kernel/kexec/kexec-tools.git/commit/?id=2f6f6d6fef7872647cacc6741ac35ac2b4df7ed5

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

package/kbd: bump version to 2.0.4

Removed patches applied upstream

0001-add-configure-flag-to-disable-tests.patch
https://git.kernel.org/pub/scm/linux/kernel/git/legion/kbd.git/commit/?id=baeb5aa827d956bd06492775dc5bd9f89d394149

0002-Link-against-libintl-when-needed.patch
https://git.kernel.org/pub/scm/linux/kernel/git/legion/kbd.git/commit/?id=b08caa2e3db53ea4263d0236c83a0df11a40a569

Autoreconf is not necessary anymore.

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

qt5location: fix plugins build dependency

Do not build the plugins before the dependency on module positioning for
the plugin subdir position is available (add Add upstream patch
0001-Fix-plugins-build-dependency.patch [1]).

Fixes [2]:

  cp -dpfr .../output/host/aarch64-buildroot-linux-gnu/sysroot/usr/lib/qt/plugins/position .../output/target/usr/lib/qt/plugins/
  cp: cannot stat '.../output/host/aarch64-buildroot-linux-gnu/sysroot/usr/lib/qt/plugins/position': No such file or directory

[1] https://code.qt.io/cgit/qt/qtlocation.git/patch/?id=3ac051c4549575634cecc706175b019f4ed4c3bf
[2] http://autobuild.buildroot.net/results/bc1/bc13abf3bb2fe1c991aec2334ee658c9641d1fd5

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

package/jsoncpp: bump version to 1.8.1

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

package/jo: bump version to 1.1

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

package/boost: bump version to 1.64.0

Removed patches applied upstream.

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

package/kismet: bump version to 2016-07-R1

Changed _SITE from git clone to tarball provided by upstream.
Rebased patches 0003 & 0004.

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

orc: update project url

The original url http://code.entropywave.com/orc is dead (server not found).

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

orc: bump version to 0.4.27

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

mariadb: bump version to 10.1.25

release notes: https://mariadb.com/kb/en/mariadb-10125-release-notes/
changelog: https://mariadb.com/kb/en/mariadb-10125-changelog/

Signed-off-by: Ryan Coe <bluemrp9@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

configs/warp7: Bump U-Boot and kernel versions

Bump U-Boot to 2017.07 and kernel to version 4.12.3.

As imx_v6_v7_defconfig in 4.12 selects CONFIG_CFG80211_WEXT=y,
remove the linux.fragment file.

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

gst-omx: bump version to 1.12.2

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

gst1-rtsp-server: bump version to 1.12.2

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

gst1-vaapi: bump version to 1.12.2

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

gst1-libav: bump version to 1.12.2

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

gst1-validate: bump version to 1.12.2

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

gst1-plugins-ugly: bump version to 1.12.2

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

gst1-plugins-bad: bump version to 1.12.2

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

gst1-plugins-good: bump version to 1.12.2

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

gst1-plugins-base: bump version to 1.12.2

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

gstreamer1: bump version to 1.12.2

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

package/jpeg-turbo: bump version to 1.5.2

Changed _SITE to https.

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/nodejs: bump version to 8.2.1

https://nodejs.org/en/blog/release/v8.2.1/

Signed-off-by: Martin Bark <martin@barkynet.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

support/testing: add test of BR2_CCACHE with an external toolchain

We reuse TestExternalToolchainBuildrootuClibc and add ccache to its
configuration.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

s6-networking: enable SSL if libressl is selected

If LibreSSL is selected, enable the associated backend to build the
s6-tls* tools.

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

bind: bump version to bugfix release 9.11.1-P3

BIND 9.11.1-P3 addresses a TSIG regression introduced in the 9.11.1-P2
security bump:

https://lists.isc.org/pipermail/bind-announce/2017-July/001057.html

Also add a hash for the license file while we're at it.

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

grep: remove musl build workaround

Commit de903755d79e (grep: fix build failure for musl) added
--with-included-regex to configure options in musl builds. The commit log
mentions build failure because grep configure doesn't detect the missing libc
regex support. Unfortunately, the commit log adds no details on said build
failure, or any autobuilder reference. The autobuilder didn't record
any grep build failure for version 2.20 that was current at the time.

Build with musl succeeds nowadays, so remove the workaround.

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

grep: standardise pcre optional dependency

Add the 'disable' config option in the 'else' clause of the optional
dependency as is the common pattern.

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

grep: bump to version 3.1

Add license file hash.

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

xvisor: bump version to 0.29

Bump version to 0.29, drop deprecated patch and update dependencies.

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/x265: bump version to 2.5

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

package/x265: disable altivec on ppc64

Disable altivec support until gcc problems are fixed:
https://bitbucket.org/multicoreware/x265/issues/320/fail-to-build-on-power8-le#comment-34076791

Fixes
http://autobuild.buildroot.net/results/419/41910d44ff98c60a6bb9fd3b6a10bd4d0b98d646/

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

testing/infra/basetest: move jlevel logic to constructor

As suggested by Arnout in [1].

While at it, simplify the logic by always appending the BR2_JLEVEL and
defaulting to 0 (the value copied from Config.in is used for 5 years now
and is very unlikely to change).

[1] http://patchwork.ozlabs.org/patch/790525/

Suggested-by: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Tested-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

support/testing: fix run-tests -j

Since commit cf3cd4388a652c9af27ef1c35622e2d0a55b99a9 the -j option is
silently ignored.

The configuration lines are processed using '\n'.join().
This function adds intervening occurrences of the separator, but the
resulting string does not end at a separator.
 >>> "n".join(["a","b"])
 'anb'
It results in a defconfig that does not end in a newline.

When BR2_JLEVEL is added by -j logic to the defconfig it ends up
concatenated to the last line of the defconfig.
 BR2_TOOLCHAIN_EXTERNAL_CODESOURCERY_ARM=yBR2_JLEVEL=7
The resulting .config has the default BR2_JLEVEL=0.

Instead of just workaround this problem by adding a newline before
BR2_JLEVEL when -j is used, make the defconfig to end in a newline since
it is a more future-proof solution.

Signed-off-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Yann E. MORIN <yann.morin.1998@free.fr>
Tested-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

testing/infra/builder: dump defconfig to log

The defconfig is composed on-the-fly by test infra + tests.

Dump it to the logfile before running 'make olddefconfig' so it can
easily analysed when debugging.

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

package/heimdal: security bump to version 7.4.0

Fixes security bugs CVE-2017-11103 & CVE-2017-6594

Changed upstream tarball location as noted in the release notes:
http://www.h5l.org/releases.html?show=7.4.0

--with-db-type-preference= is needed to fix a build error:

  CCLD     otp
../../lib/otp/.libs/libotp.a(otp_db.o): In function `otp_get_internal':
otp_db.c:(.text+0x32): undefined reference to `__roken_dbm_fetch'
otp_db.c:(.text+0xd9): undefined reference to `__roken_dbm_store'
../../lib/otp/.libs/libotp.a(otp_db.o): In function `otp_db_open':
otp_db.c:(.text+0x1c9): undefined reference to `__roken_dbm_open'
../../lib/otp/.libs/libotp.a(otp_db.o): In function `otp_db_close':
otp_db.c:(.text+0x205): undefined reference to `__roken_dbm_close'
../../lib/otp/.libs/libotp.a(otp_db.o): In function `otp_delete':
otp_db.c:(.text+0x23e): undefined reference to `__roken_dbm_delete'
../../lib/otp/.libs/libotp.a(otp_db.o): In function `otp_put':
otp_db.c:(.text+0x388): undefined reference to `__roken_dbm_store'
collect2: error: ld returned 1 exit status

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

configs/nanopi_m1_*: bump to final u-boot 2017.07

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

board: Add support for nanopi-m1-plus

Add initial support for nanopi-m1-plus board
with below features
- U-Boot 2017.07-rc3
    nanopi-m1-plus dts support is present in rc3 version
- Linux 4.11.5
- linux patches to add nanopi-m1-plus dts
- Default packages from buildroot

Signed-off-by: Chakra Divi <chakra@openedev.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/kvm-unit-tests: fix syntax

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

ndisc6: patch no longer required

Now that we have 0001-remove-__FAVOR_BSD.patch in uClibc-ng (which is
upstream and will be part of uClibc-ng 1.0.26), we no longer need the
work-around patch in the ndisc6 package.

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

pkg-generic: make PKG_DL_DIR equal to DL_DIR

This variable is currently not used by anyone.
The value is changed to match the path to DL_DIR.

The next patch will introduce the use of PKG_DL_DIR for packages that
use specific EXTRACT_CMDS.

And then we will be able to change the value of PKG_DL_DIR to create a
new directory structure in DL_DIR with a subdirectory for each package.

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

package/aespipe: fix host compile

Building host-aespipe fails on Debian stretch at linking stage:

/usr/bin/gcc -L/home/buildroot/br6/output/host/lib -L/home/buildroot/br6/output/host/usr/lib -Wl,-rpath,/home/buildroot/br6/output/host/usr/lib -o aespipe aespipe.o aes-amd64.o md5-amd64.o md5-2x-amd64.o aes-intel64.o sha512.o rmd160.o
/usr/bin/ld: aes-amd64.o: relocation R_X86_64_32S against `.rodata' can not be used when making a shared object; recompile with -fPIC

The same problem apparently exists on recent Ubuntu and Gentoo.

Fix is also used in Debian:
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=837393

[Peter: add comment explaining why]
[Arnout: use host-cc-option to discover if -no-pie is available;
 cfr. 57b628a932b9b4a3c4bf80f4c82a81da5adcb173]
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/Makefile.in: add host-cc-option macro

This macro allows to test if HOSTCC supports a specific option. It is
needed to pass '-no-pie' on recent Debian, Ubuntu and Gentoo hosts.

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

package/gnuradio: link with libatomic when needed

Fixes
http://autobuild.buildroot.net/results/528/52825918bf65b23241cd68d99513350703469294/

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

package/kvm-unit-test: available for big.LITTLE arm cores too

Since it supports each variants individually, it should also supports
the big.LITTLE configurations.

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

arch/arm: add big.LITTLE cpu variants

The big.LITTLE configurations can be optimised for by gcc, and a few
users wonder what they should choose when they have such CPUs.

Add new entries for those big.LITTLE configurations.

Note: the various combos were added in various gcc versions, but only
really worked in later versions:

    Variant   | Introduced in | First built in
    ----------+---------------+----------------
    a15-a7    | 4.9           | 4.9
    a17-a7    | 5             | 5
    a57-a53   | 4.9           | 6
    a72-a53   | 5             | 6

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Thomas De Schampheleire <patrickdepinguin@gmail.com>
Cc: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/kvm-unit-tests: introduce _ARCH_SUPPORTS

Move all architecture options to their own symbol, so that it is easier
to add more variants in the future.

The dependency on cross-gcc >= 4.5 is only valid for i386, as we use the
host gcc for x86_64. Adapt the dependency accordingly.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Cyril Bur <cyrilbur@gmail.com>
[Thomas:
 - add comment about host gcc >= 4.5 dependency on x86-64
 - remove bogus "depends on ... if ..." construct.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

efibootmgr: fix build with gcc 7.x

Now that the build of efivar with gcc 7.x has been fixed by commit
0ca30170345a81f5f21e4ef4424b1f186cde1988 ("efivar: fix build with gcc
7"), efibootmgr fails similarly with gcc 7.x.

This commit backports an upstream patch that fixes this issue.

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

testing/tests/package: add basic unit test for IPython

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
[Thomas: update .gitlab-ci.yml.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

testing/tests/package/test_python: allow to change timeout

Depending on Python implementation used for testing, time it takes to
perform a given test can vary pretty significantly. To accout for that
allow individual test functions to specify different timeout value.

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

testing/tests/package/test_python: do not hardcode interpreter name

In order to be able to leverage the same test code for testing
different python interpreters (or wrappers around CPython) allow child
classes of TestPythonBase to override the name of the executable used
to run tests.

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

testing/tests/package/test_python: add TestPython3

Add Python3 version of TestPython2 to make sure both versions of
Python get unit-tested.

Modify the code of libc_time_test() to support that change (convert
the code to use Python3-style "print").

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
[Thomas: update .gitlab-ci.yml.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/kodi-inputstream-adaptive: new package

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
[Thomas: use SPDX license code.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/kodi-inputstream-rtmp: new package

Select the new package by the legacy option for easier update.

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
[Thomas: use SPDX license code.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

cppcms: fix build with ICU

Since the bump of ICU to 59.1, it needs C++11. cppcms has been updated
with the appropriate dependency on gcc >= 4.8, but it also needs
-std=c++11. Therefore, this patch complements the CXXFLAGS passed to
cppcms by adding what icu-config reports, in a manner similar to what
was done to fix the same issue in the PHP package in commit
214a76d045f906c1a8d687b7c7e7bc85c5b09906 ("package/php: fix icu
support").

Fixes:

  http://autobuild.buildroot.net/results/f05d9999f1f3248b6a50f14de476108028ed047e/

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

testing/tests/package/test_python: refactor TestPythonBase

Convert TestPythonBase to a true base class that only provides code
implementing various tests without defining tests themselves in a
"discoverable" form.

To retain correct testing functionality, add TestPython2 derived class
that uses code from TestPythonBase to define actual runnable test.

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Reviewed-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
[Thomas: fix typo in commit log, update .gitlab-ci.yml, both pointed
by Ricardo.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

testing/infra/emulator: allow to specify pexpect timeout

Some commands take more than 5 seconds to complete under QEMU, so add
provisions to allow individual unit-test to specify different duration
to avoid false negative test failures.

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Reviewed-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

system: move remounting / to the corresponding init systems

Currently, remounting / read-write (or not) is done by the skeleton
package when the init system is either busybox or sysvinit, by
registering a target-finalize hook; it is not done at all for systemd.

Move registering this target-finalize hook to both of busybox and
sysvinit. Leave systemd alone, we'll take care of it later.

Rename the macro to a more meaningful name, and move it to system.mk
with the other such macros.

Yet a little bit less init-system knowledge in the skeleton.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
[Thomas: remove not-so-useful comments, as pointed by Arnout.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

system: move setting getty to the corresponding init systems

Currently, setting the getty is done:
  - by the skeleton package when the init system is either busybox or
    sysvinit;
  - by the systemd package when the init system is systemd;
both by registering a target-finalize hook.

This is not very consistent.

Move setting the getty out of the skeleton and into the package that
provides the init system, by registering a per-package target-fialize
hook.

This offloads yet a bit more out of the skeleton, so that it is easier
to properly separate the skeletons for the various init systems.

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

system: provide package-wide system variables and macros

Some macros, soon some variables, currently defined in the skeleton are
going to be used by other packages.

Some of those variables will be used as Makefile conditions (e.g. in
ifeq() conditions), so they *must* be defined before being used.

Since the skeleton package, starting with an 's', is included quite
late, those variables would not be available to most packages.

Offload the existing macros into the new system/system.mk file, that is
included early, before any package is. Rename the macros to appropriate
names.

Future commits will add new macros and variables in that file.

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

package/skeleton: make SKELETON_LIB_SYMLINK a macro

Rather than duplicate that code in all skeleton packages, make this a
macro, so that it is easier to use.

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

package/ffmpeg: disable build of nvidia hardware acceleration support

ffmpeg always enables support for nvenc/cuda even if their support
libraries are not present:

External libraries providing hardware acceleration:
cuda cuvid nvenc
[...]
Enabled hwaccels:
h264_cuvid mjpeg_cuvid mpeg2_cuvid vc1_cuvid vp9_cuvid
hevc_cuvid mpeg1_cuvid mpeg4_cuvid vp8_cuvid

This leads to a crash in freeswitch git master when transcoding video
streams on a system without nvidia hardware:

2017-07-22 15:06:27.306760 [INFO] avcodec.c:1077 initializing encoder 352x288
2017-07-22 15:06:27.306760 [NOTICE] avcodec.c:828 NVENC HW CODEC ENABLED

This patch disables the support of nvidia hardware acceleration support
for now until the needed packages are added to buildroot. For details
about this please refer to https://developer.nvidia.com/ffmpeg

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

package/skeleton: add macro to rsync skeleton directory

The rsync command is currently called from a single location. In
follow-up patches, it will be called from various places then.

It will be easier to update and fix this command if it is defined in a
single location, rather than if duplicated in many places.

Define a macro that skeleton packages can call to install their skeleton
directories. Since at least one occurrence of it will want to install in
staging, the macro must accept a destination path, rather than hard-code
target as the destination.

Since it is to be used by other skeleton packages, define it early,
outside of any conditional block.

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

core/pkg-generic: add variable to skip skeleton dependency

We currently skip the skeleton dependency by checking if the current
package is the skeleton.

We are going to introduce more skeleton-related packages, so we
need a way to exclude the skeleton dependency for those, or we'd
get a circular dependency, for the same reason we need to skip
the toolchain dependency.

Instead of checking for all the skeleton-providing packages in the core
infra, add a new package options so that packages can express they do
not need the dependency on the skeleton, like we have an option to avoid
the depednency on the toolchain. The only packages that will use that
option are probably the skeletons, so we need not document this
variable, like we did not document the option to exclude the dependency
on the toolchain.

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

paxtest: new package

PaX regression test suite

Signed-off-by: David Graziano <david.graziano@rockwellcollins.com>
Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

configs/nitrogen*: bump kernel and u-boot revisions

Here are the main U-Boot modifications:
- Rebase on top of mainline 2017.03 release
- Fix sparse image flashing issues

Here are the main Linux modifications:
- Add mma8451 driver support
- Fix possible TiWi spurious interrupts issue
- Fix Ethernet interrupt parent issue
- Add USB Type C TUSB320 controller support
- Add SOM2 i.MX6DL and i.MX6QP variants

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

package/openal: bump version to 1.18.0

Patches 0001 to 0004 were upstream backports, so they have been
dropped. Patch 0005 has been upstreamed as commit
https://github.com/kcat/openal-soft/commit/af5fb3d6e73ac5e7f015011974c7a7443ae6744b. Therefore,
all 5 patches are dropped.

Signed-off-by: André Hentschel <nerv@dawncrow.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

support/scripts/fix-rpath: typo fix and unneeded "; " removed

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

support/scripts/fix-rpath: check if the patchelf utility is executable

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

core: introduce "sdk" target to make a relocatable SDK

We use a separate make target to build a relocatable SDK. We first
sanitize the RPATH in host tree. Next we also sanitize the
staging tree. Therefore "sdk" must depend on world.

Sanitizing staging is not really needed, in the sense that any rpath
in there is simply not going to be used. We want to sanitize staging
for the following reasons:

- To avoid leaking references to the original output directory. This
  way, we can validate that the SDK is relocatable by running a simple
  "grep -r ${BASE_DIR} ${HOST_DIR}". Obviously RPATH sanitization is
  not sufficient (e.g. also the references to source files have to be
  stripped), but it's a step in the right direction. This reason is
  obviously only relevant for the SDK.

- To make sure that when an executable is copied to target that it
  actually executes correctly. Since within Buildroot we never copy
  stuff from staging to target, this is clearly only relevant for
  the SDK.

Finally we install the script "relocate-sdk.sh" into the top directory
of the SDK (HOST_DIR) and the SDK location path is stored in the file
"HOST_DIR/share/buildroot/sdk-location"-

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

package/libtirpc: security bump to version 1.0.2

Fixes CVE-2017-8779:
http://git.linux-nfs.org/?p=steved/libtirpc.git;a=commitdiff;h=dd9c7cf4f8f375c6d641b760d124650c418c2ce3

Rebased patches 0001, 0002 & 0006.
Removed patch 0007, applied upstream:
http://git.linux-nfs.org/?p=steved/libtirpc.git;a=commit;h=4f1503e84b2f7bd229a097335e52fb8203f5bb0b
Renumbered patch 0008.

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

libmemcached: fix build with gcc 7.x

This commit adds a patch to the libmemcached package that fixes the
build with gcc 7.x. Since libmemcached is barely maintained upstream,
the patch comes from the Fedora packages.

Fixes:

  http://autobuild.buildroot.net/results/872b8e0e6a24cbc96e3ad9e0b8b47acdf6160ce0/

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

collectd: fix build with gcc 7.x

This commit backports an upstream collectd patch that fixes a build
issue with gcc 7.x.

Fixes:

  http://autobuild.buildroot.net/results/2441e2a69d013a6376a90d375e15991e8cb816bd/

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

python-requests: bump to version 2.18.1

Add new runtime dependencies and select Python SSL support.

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

python-flask: bump to version 0.12.2

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

python-pathvalidate: bump to version 0.16.0

Upstream project has dropped all runtime dependencies.

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