package/gcc: use the Xtensa variables

... instead of re-computing them over-and-over-again.

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

package/gdb: use the Xtensa variables

... instead of re-computing them over-and-over-again.

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

package/binutils: use the Xtensa variables

... instead of re-computing them over-and-over-again.

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

arch/xtensa: add definitions for the Xtensa overlay

Currently, we have four packages that need to use the Xtensa overlay:
binutils, gcc-initial, gcc-final and gdb. However, they all use three
different ways to check for, and use the overlay. Furthermore, two of
them define the same variable: XTENSA_CORE_NAME is defined both by
binutils and gdb

In order to sanitise that, have the Xtensa architecture export the
appropriate variables: the name of the core and the path to the overlay
tarball. Packages can then check and use those variables.

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

core: include arch-specific definitions

Allow architectures to define variables and helper macros.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

toolchain/helpers.mk: remove unused argument of check_arm_abi

The check_arm_abi function takes as second argument the path to the
cross-readelf, but does not use it. Therefore, this commit gets rid of
this unnecessary argument.

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

toolchain-external: improve musl external check

The current test to verify if the toolchain uses musl or not is based on
checking if /lib/libc.so or /lib/libm.so exist in the sysroot. However,
some toolchains (notably Crosstool-NG ones) put these libraries in
/usr/lib/.

To fix this, build a minimal C program and check if the program
interpreter contains /lib/ld-musl.

Signed-off-by: Ilya Kuzmich <ilya.kuzmich@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

toolchain-external: fix definition of TOOLCHAIN_EXTERNAL_READELF

TOOLCHAIN_EXTERNAL_READELF is defined to
$(TOOLCHAIN_EXTERNAL_CROSS)readelf$(TOOLCHAIN_EXTERNAL_SUFFIX), where
TOOLCHAIN_EXTERNAL_SUFFIX is .br_real for Buildroot
toolchains. However, this is bogus, because readelf is not wrapped by
the Buildroot toolchain wrapper, so "<arch>-readelf.br_real" never
exists.

Therefore, it should simply be defined as
$(TOOLCHAIN_EXTERNAL_CROSS)readelf. Currently,
TOOLCHAIN_EXTERNAL_READELF is not used anywhere, so it wasn't visible,
but a follow-up commit will make use of it.

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

jo: new package

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

htop: use correct ncurses config script name

The ncurses config script can have different names depending on the
configuration (e.g. wchar support enabled/disabled). So, use
$(NCURSES_CONFIG_SCRIPTS) from the ncurses package as it contains the
proper name.

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

slang: use correct ncurses config script name

The ncurses config script can have different names depending on the
configuration (e.g. wchar support enabled/disabled). So, use
$(NCURSES_CONFIG_SCRIPTS) from the ncurses package as it contains the
proper name.

Fixes:

  http://autobuild.buildroot.net/results/c97554c05e121c76407be53ba37924cfcb658e9d/
  http://autobuild.buildroot.net/results/b0a31a3491a30ee0d6e6e7dbba68a86f0e00943d/
  http://autobuild.buildroot.net/results/a77241141cf8710fa945523ef62327798e3c4824/
  [...]

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

xenomai: bump version to 3.0.3

- Bump Xenomai version to 3.0.3

- Remove patch 0001-bfin-remove-inline-keyword.patch for version
  2.6.4, as it is already included in upstream

- Add upstream patch fixing gcc 6.x build issue of cyclictest.

Signed-off-by: Pawel Sikora <sikor6@gmail.com>
[Thomas: add patch fixing gcc 6.x build issue.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

linux-firmware: Add missing firmware for the carl9170 driver

For a long time now Atheros USB AR9170 devices have been using the
carl9170 driver which relies on the carl9170-1.fw firmware file.  This
firmware is available in linux-firmware but it wasn't currently
installed by Buildroot.

Signed-off-by: Karoly Pados <kp@tec4data.at>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

python3: Bump version to 3.6.1

Update all appropriate version numbers as well as SHAs and MD5s as well
as repbasing BR's patches on top of 3.6.1 codebase (new github repo
tree, v3.6.1 tag was used). Note that patch:

   [PATCH] Change the install location of _sysconfigdata.py

was dropped due to the fact taht build system now adds platform
specific suffix to sysconfigdata's name, so each platform's file
should have a unique name and distutils now allows to specify which
sysconfigdata is used via _PYTHON_SYSCONFIGDATA_NAME

see:

    https://github.com/python/cpython/commit/c4b53afce491142b80b228a21a05de5dcfd8d36f
    https://github.com/python/cpython/commit/92dec548ff1494b86f08bd3753ca91a9330b4ea9

and patches:

    [PATCH] distutils/sysconfig: use sysconfigdata
    [PATCH] setup.py: do not add invalid header locations
    [PATCH] Do not harcode invalid path to ncursesw headers

was dropped since it looks like it made it's way upstream, see:

    https://github.com/python/cpython/commit/409482251b06fe75c4ee56e85ffbb4b23d934159
    https://github.com/python/cpython/commit/1351c31aa9651b278d7ef8ec79af3b646a520235
    https://github.com/python/cpython/commit/e13c3201fb66c4c211b4ebb7604d1435bedc1015

respectively.

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

qt: add enable for truetype font installation

Currently, qt.mk installs DejaVu/Vera TrueType fonts when Qt is configured for
freetype support. However, some users may prefer to limit the installed fonts
to only Qt's prerendered fonts (micro, fixed, helvetica, etc.). Or, there may
be a preference to use seperately installed TrueType fonts. This commit adds
a switch to explicitly control the installation of the TrueType fonts, similar
to the existing prerendered font enables.

Signed-off-by: Danomi Manchego <danomimanchego123@gmail.com>
Reviewed-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

vnstat: bump to version 1.17

Also remove patch that has been merged upstream, disable AUTORECONF
since we're no longer patching configure.ac and get rid of hack
involving the install target as the 'install-data-hook' in Makefile.am
now avoids running 'vnstat --showconfig' on the host in case of
cross-compilation.

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

ustream-ssl: bump version

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

uhttpd: bump version

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

libuci: bump version

Change upstream location

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

ubus: bump version

rebase 0001-Install-server-and-client-examples patch
on latest version.

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

libubox: bump version

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

openvpn: Add explicit support for external lz4

2.4.0 added lz4 support. Make that choice explicit or the bundled compat-lz4
"stub" will be used.

Signed-off-by: Jeroen Roovers <jer@airfi.aero>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

openvpn: bump version to 2.4.1

--disable-eurephia configure option doesn't exist, so remove it.

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/ntp: fix build for toolchains without ssp

ntp defaults to use ssp support
http://bk1.ntp.org/ntp-stable/sntp/harden/README

by using these flags
http://bk1.ntp.org/ntp-stable/sntp/harden/linux

If the toolchain lacks SSP support this patch forces ntp to use an
empty set of flags: http://bk1.ntp.org/ntp-stable/sntp/harden/default

Fixes
http://autobuild.buildroot.net/results/1d5/1d58bd8745b22c8eb71fea4c7255d3ace69f6f7a/

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

lz4: version v1.7.5

Signed-off-by: Jeroen Roovers <jer@airfi.aero>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

busybox: fix mdev.conf

mdev does not set "/dev/snd" and "/dev/input" group and permission
propery, because with commit c3cf1e30a3022453311a7e9fe11d94c7a381640e
(May 2013!) the behavior of mdev has changed.

The device name is now taken directly from the uevent file and does no
longer match the old rule.

Fix the rules for "/dev/snd" and "/dev/input" according to the example
given in [1].

[1] http://lists.busybox.net/pipermail/busybox/2015-February/082297.html

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

libinput: bump version to 1.7.0

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

lft: bump version to 3.78

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

upmpdcli: add user to the audio group

upmpdcli runs as user and group upmpdcli. In order to allow it to access
the ALSA devices add it to the audio group.

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

tcpreplay: bump version to 4.2.1

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Acked-by: Chris Packham <judge.packham@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

fs/ext2: Remove support for auto-calculation of rootfs size

Auto-calculation of the rootfs size cannot be done in a reliable way
as it depends on the host filesystem and is broken on non ext4 host
(see bugs [1] [2]). So let the user specify the size he wants for his
rootfs.

[1] https://bugs.busybox.net/show_bug.cgi?id=8831
[2] https://bugs.busybox.net/show_bug.cgi?id=9496

Signed-off-by: Sébastien Szymanski <sebastien.szymanski@armadeus.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
[Thomas: as suggested by Arnout, use 60 MB as the default size instead
of 64 MB.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

cjson: bump to version 1.4.4

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

package/enlightenment: bump to version 0.21.7

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

board/csky: fixup gdb instructions in readme.txt

The gdbinit file assumes the .dtb file is in the current directory, so
let's start gdb from output/images.

Signed-off-by: Guo Ren <ren_guo@c-sky.com>
[Thomas: improved commit description.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

kmscube: Bump to latest version to fix mips64el errors

Building kmscube for mips64el leads to the following build error:

cube-tex.c: In function 'get_fd_rgba':
cube-tex.c:230:2: error: 'for' loop initial declarations are only allowed in C99 mode
  for (int i = 0; i < texh; i++) {
  ^
cube-tex.c:230:2: note: use option -std=c99 or -std=gnu99 to compile your code
cube-tex.c: In function 'get_fd_y':
cube-tex.c:261:2: error: 'for' loop initial declarations are only allowed in C99 mode
  for (int i = 0; i < texh; i++) {
  ^
cube-tex.c: In function 'get_fd_uv':
cube-tex.c:292:2: error: 'for' loop initial declarations are only allowed in C99 mode
  for (int i = 0; i < texh/2; i++) {

I have sent a fix to mainline kmscube to address such issue, so bump to
the latest version.

Fixes: http://autobuild.buildroot.net/results/1f2/1f2ad0536254834582cf7fbce6a38dbb46a5e249/build-end.log
Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

tcpreplay: Bump version to v4.2.0

Signed-off-by: Chris Packham <judge.packham@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

nginx-naxsi: bump to 0.55.3

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

ruby: bump version to 2.4.1

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

janus-gateway: bump to v0.2.2

In addition, a new option sample-event-handler has been added. It has
been disabled by default in the .mk file, as it requires
libcurl. Support for it can be added later.

While at it, remove a trailing space.

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

imagemagick: bump version to 7.0.5-3

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

linux: bump default version to 4.10.5

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

linux-headers: bump 3.{2, 12}.x and 4.{1, 4, 9, 10}.x series

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

systemd: bump version to 233

--disable-kdbus no longer exists, so remove everything related to it.

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

vboot-utils: correct hash

I created the hash based on an old tarball in my dl dir which was
generated without the sorting introduced in commit 4f87aa7db5ee. This
time it is based on a properly created tarball.

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

git: fix build with with no threads

When building git with toolchain that don't have thread support, we
get following build errors:

      CC builtin/hash-object.o
  builtin/grep.c: In function 'grep_submodule_launch':
  builtin/grep.c:596:34: error: dereferencing pointer to incomplete type 'struct work_item'
    status = capture_command(&cp, &w->out, 0);
                                    ^
  builtin/grep.c: In function 'grep_submodule':
  builtin/grep.c:644:20: error: storage size of 'w' isn't known
     struct work_item w;
                      ^
  make[2]: *** [builtin/grep.o] Error 1
  make[2]: *** Waiting for unfinished jobs....

Add two upstream patches to fix this issue.

Fixes:
  http://autobuild.buildroot.net/results/94b/94bce9a99a5ce9894a6918774ab75e23d12c1394/

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

package/ntp: security bump to 4.2.8p10

Changed NTP_SITE to https to circumvent "URL transformed to HTTPS due
to an HSTS policy" during download.

For details about the bugs fixed see:
http://support.ntp.org/bin/view/Main/SecurityNotice#Recent_Vulnerabilities
http://www.kb.cert.org/vuls/id/633847

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

tesseract-ocr: needs wchar support

Fixes:
http://autobuild.buildroot.net/results/66e44564aac79c1c47755f6d158663317a5ffe09

Signed-off-by: Gilles Talis <gilles.talis@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

leptonica: make package description more accurate

Signed-off-by: Gilles Talis <gilles.talis@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

arch/Config.in.arc: fix BR2_READELF_ARCH_NAME

On ARC, depending on the CPU selected, the ELF machine number might be
different. Take this into account when defining BR2_READELF_ARCH_NAME.

Fixes:

   http://autobuild.buildroot.net/results/0a1957056e07cedc4839439c0ea8ebd1f646b904/

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

support/scripts/check-bin-arch: fix after "fix for filenames with spaces"

In commit 2f6c5e513cf6688f9343a25ac90138e9ad751eb5
("support/check-bin-arch: fix for filenames with spaces"), Yann
adjuste the check-bin-arch script to properly handle filenames with
spaces.

However, he also did a subtle change of the regexp that extracts the
path of the files. It was:

  "/^${package},(.+)$/!d; s//\1/;"

and Yann changed it to:

  "/^${package},\.(.+)$/!d; s//\1/;"

So the file paths used to start with a dot (like "./usr/share/foo"),
and now they no longer start with a dot (like "/usr/share/foo"). While
this modification is good and makes sense, the match for
/lib/firmware/ was not adjusted accordingly, and the follow-up patch
also ignoring /usr/share was not adjusted as well.

This commit fixes those /lib/firmware/ and /usr/share/ special cases,
which will fix:

  http://autobuild.buildroot.net/results/76a1475f4cdedb80426fb022ef2e644aa5625660/

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

augeas: new package

Augeas is a configuration editing tool. It parses configuration files in
their native formats and transforms them into a tree. Configuration
changes are made by manipulating this tree and saving it back into
native config files.

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Cc: Romain Naour <romain.naour@gmail.com>
[Thomas: fix Config.in comment dependency.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

vboot-utils: add hash

vboot-utils had a hash file, but is says "none" because it is a git
download.  Since our git downloads are not reproducible, we can actually
add a hash.

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

uhttpd: add hash

uhttpd had a hash file, but is says "none" because it is a git download.
Since our git downloads are not reproducible, we can actually add a
hash.

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

ubus: add hash

ubus had a hash file, but is says "none" because it is a git download.
Since our git downloads are not reproducible, we can actually add a
hash.

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

squashfs: correct hash

When squashfs was bumped to 3de1687d in commit ee90313c64, the hash
file was not updated. This wasn't noticed before since hashes are not
checked for git downloads.

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

linux-firmware: correct hash

When linux-firmware was bumped to 6d3bc8886 in commit 3ff5896ff, a hash
file was added but the hash was wrong. This wasn't noticed before since
hashes are not checked for git downloads.

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

fmc: correct hash file

It had sha256sum instead of sha256. This wasn't noticed before since
hashes for git downloads are not checked.

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

aer-inject: remove redundant _SITE_METHOD

The _SITE has a git:// URL, so _SITE_METHOD is not needed.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Tiago Brusamarello <tiago.brusamarello@datacom.ind.br>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

download/git: create GNU format tar files

On most distros, the tar format defaults to GNU. However, at build time
the default format may be changed to posix. Also, future versions of
tar will default to posix.

Since we want the tarballs created by the git download method to be
reproducible (so their hash can be checked), we should explicitly
specify the format. Since existing tarballs on sources.buildroot.org
use the GNU format, and also the existing hashes in the *.hash files
are based on GNU format tarballs, we use the GNU format.

In addition, the Posix format encodes atime and ctime as well as mtime,
but tar offers no option like --mtime to override them. In the GNU
format, atime and ctime are only encoded if the --incremental option is
given.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/libusb: bump version to 1.0.21

Removed patch applied upstream.

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

python-pathvalidate: bump version to 0.15.0

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

graph-depends: rename pkgutil.py to brpkgutil.py

pkgutil.py is also part of Python itself. Placing pkgutil.py as is
in a folder with other scripts that require original pkgutil will
break them. This is the case with scanpypi. So rename pkgutil.py
to brpkgutil.py to avoid naming collision.

Fixes: https://bugs.busybox.net/show_bug.cgi?id=9766
Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

tslib: update to 1.8

For those who are interested, the release notes:

https://github.com/kergoth/tslib/releases

Signed-off-by: Martin Kepplinger <martin.kepplinger@ginzinger.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

tslib: use main project page in Config.in help text

Signed-off-by: Martin Kepplinger <martin.kepplinger@ginzinger.com>
[Thomas: instead of adding tslib.org, just replace the github address by
the tslib.org one, which is the really upstream homepage.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ghostscript: drop the ijs subfolder to fix build

The ijs/ subfolder contains a bogus ltmain.sh file, which is actually a
broken symbolic link. Due to this, our logic to patch ltmain.sh files
fail with some patch versions:

  >>> ghostscript 9.21 Patching libtool
  File /home/buildroot/build/instance-0/output/build/ghostscript-9.21/ijs/ltmain.sh is not a regular file -- refusing to patch

Since we are anyway not using the ijs support (--without-ijs is passed),
do like we do for all other sub-components of ghostscript: get rid of
the ijs/ subfolder completely.

Fixes:

  http://autobuild.buildroot.net/results/ca683d69b7fb564788c8877c3bb6466390e976a8/

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>

ltp-testsuite: disable Open POSIX testsuite

The Open POSIX testsuite builds and installs to the target directory a
program called 't0' that isn't cross-compiled, which is bad.

Since the LTP build system is autoconf but not automake based, and the
Open POSIX testsuite is a sort of sub-project inside it, fixing the
issue is not trivial.

Therefore, we simply disable the Open POSIX testsuite entirely. Oddly
enough, --without-open-posix-testsuite doesn't work due to another bug,
so we simply have to remove --with-open-posix-testsuite.

Open POSIX testsuite cross-compilation issue reported at
https://github.com/linux-test-project/ltp/issues/144.

Configure script bug reported at
https://github.com/linux-test-project/ltp/issues/143.

Fixes:

  http://autobuild.buildroot.net/results/8326ba9eb257dfc92c1ad282ba6d3565e8250def/

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

package/dbus-cpp: fix cross-compilation

dbus-cpp has an ugly hack to not cross-build the tools when it detects
cross-compilation.

However, we already have a host variant that builds those tools (and we
anyway don't seem to need them to begin with).

Drop our patch that propagates the CXXFLAGS/LDFLAGS_FOR_BUILD. Replace
with a patch to cross-compile the tools.

Fixes:
    http://autobuild.buildroot.org/results/110/1100539caae6ef62c61a3b96bc54f7c0a215cc62/
    and many, many more.

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

apr: fix size of pid_t

pid_t is a signed 32bits integer on both 32bits and 64bits
architectures.
This fixes an issue with apache server which causes bad pid
to be written in PidFile

Signed-off-by: Julien Beraud <julien.beraud@spectracom.orolia.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

irssi: add patch to fix static linking

An upstream change which removed the option to build Irssi without SSL
support (v1.0.0 and later) also eliminated the use of PKG_CHECK_MODULES
to find the OpenSSL library, relying only on a test with AC_CHECK_LIB
for that purpose.

This unfortunately broke static linking since the flag to link with
zlib, used by OpenSSL, is missed completely.

The newly added patch therefore adds PKG_CHECK_MODULES back into the
configure script.

Fixes:

  http://autobuild.buildroot.net/results/e8b51f65525246d1205a26859d418393aaebf48c/
  http://autobuild.buildroot.net/results/35d952b3b36e295723bf8ed1badb4c9439201822/
  http://autobuild.buildroot.net/results/ea8af1458edf3ce0e2b1c444d2ae29ac1e5d2fbf/
  [...]

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

package/gnupg2: fix libusb detection

gnupg2 does not use pkgconf to detect the include dir of libusb.

Fixes
http://autobuild.buildroot.net/results/70c/70c9f5f8ae77fbda2b5eba3aacf831430dc59a1c/

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

vala: bump version to 0.34.7

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

support/script/check-bin-arch: ignore /usr/share

/usr/share normally should not contain binaries executable for the
target platform. However, it might contain ELF binaries for other
platforms, such as firmware files installed by Qemu or
pru-software-support.

Instead of special-casing each package, let's simply ignore /usr/share.

Fixes:

  http://autobuild.buildroot.net/results/6f3fea9f6adaef1573fbb0dd6903b5d99e470610/
  (pru-software-support)

  http://autobuild.buildroot.net/results/fe8892bc22a03299fc41e30bfea5e42166838f88/
  (qemu)

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

support/check-bin-arch: fix for filenames with spaces

Filenames with spaces will break the current for loop.

Fix that by using a while-read loop, fed with the list of files on
stdin, using process substitution.

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

poppler: bump version to 0.53.0

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

btrfs-progs: bump version to 4.10.1

Drop 0003-btrfs-progs-add-missing-limits-headers.patch since it's
already included in this release:
  https://github.com/kdave/btrfs-progs/commit/153d4e6c01b53855d01b5d0c99886ba31b48bad1

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

moarvm: bump version to 2017.03

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

memcached: bump version to 1.4.36

From the release notes
(https://github.com/memcached/memcached/wiki/ReleaseNotes1436):

Important bug fix that could lead to a hung slab mover.  Also improves
memory efficiency of chunked items.

[Peter: add release notes link / mention important bug fix]
Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libgtk3: bump version to 3.22.11

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

json-glib: bump version to 1.2.8

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

git: bump version to 2.12.1

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

openssh: security bump to version 7.5

From the release notes (https://www.openssh.com/txt/release-7.5):

Security
--------

 * ssh(1), sshd(8): Fix weakness in CBC padding oracle countermeasures
   that allowed a variant of the attack fixed in OpenSSH 7.3 to proceed.
   Note that the OpenSSH client disables CBC ciphers by default, sshd
   offers them as lowest-preference options and will remove them by
   default entriely in the next release. Reported by Jean Paul
   Degabriele, Kenny Paterson, Martin Albrecht and Torben Hansen of
   Royal Holloway, University of London.

 * sftp-client(1): [portable OpenSSH only] On Cygwin, a client making
   a recursive file transfer could be maniuplated by a hostile server to
   perform a path-traversal attack. creating or modifying files outside
   of the intended target directory. Reported by Jann Horn of Google
   Project Zero.

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

package/ghostscript: new package

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
[Thomas:
 - switch to version 9.21 now that it has been released
 - add a hash file
 - switch to Git formatted patches
 - use $(HOSTCC) instead of hardcoding "gcc", and use $(HOST_CFLAGS) and
   $(HOST_LDFLAGS) instead of hardcoding -L$(HOST_DIR)/usr/lib
   -I$(HOST_DIR)/usr/include
 - add entry to DEVELOPERS file.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/lcms2: add support for host variant

Needed for upcoming ghostscript package.

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

python3: generate reproducible .pyc

.pyc files contain the modification time of the corresponding .py
source. In order to make the build reproducible, we fix the modification
time of all .py before compiling .pyc files.

In addition, since pycompile relies on the modification time to know if
a file needs to be recompiled, it is safer to force recompilation of all
source files.

This work was sponsored by `BA Robotic Systems'.

Signed-off-by: Jérôme Pouiller <jezz@sysmic.org>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
[Thomas: do not register PYTHON_FIX_TIME as a
 PYTHON_TARGET_FINALIZE_HOOKS, instead call it inside
 PYTHON_CREATE_PYC_FILES before doing the byte compilation.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

python2: generate reproducible .pyc

.pyc files contain the modification time of the corresponding .py
source. In order to make the build reproducible, we fix the modification
time of all .py before compiling .pyc files.

In addition, since pycompile relies on the modification time to know if
a file needs to be recompiled, it is safer to force recompilation of all
source files.

This work was sponsored by `BA Robotic Systems'.

Signed-off-by: Jérôme Pouiller <jezz@sysmic.org>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
[Thomas: do not register PYTHON_FIX_TIME as a
 PYTHON_TARGET_FINALIZE_HOOKS, instead call it inside
 PYTHON_CREATE_PYC_FILES before doing the byte compilation.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-elecrons.com>

pycompile: allow to force compilation

By default, compile_dir() relies on the modification time to know if a
python file has to be built again. However in some circumstances (when
doing reproducible builds), modification times are not reliable. Thus,
this patch adds a way to force the rebuild of all python sources.

Signed-off-by: Jérôme Pouiller <jezz@sysmic.org>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

configs: add defconfig for Nationalchip gx6605s dev board

gx6605s is a nice SOC for dvbs2 DVB product, and C-SKY inside.

Signed-off-by: Guo Ren <ren_guo@c-sky.com>
[Thomas: drop post-image script copying the gdbinit file, and suggest
to use gdb -x option instead to point to the gdbinit file in board/.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gst1-plugins-bad: waylandsink needs wayland-protocols

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
----
Changes v1 -> v2:
  - use select instead of depends for BR2_PACKAGE_WAYLAND_PROTOCOLS
   (suggested by Thomas Petazzoni)
  - add select BR2_PACKAGE_WAYLAND_PROTOCOLS to missing second
    place (BR2_PACKAGE_GST1_PLUGINS_BAD_PLUGIN_WAYLAND)
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libxkbcommon: disable wayland support

Disable wayland support (only needed for the interactive-wayland
test program). This fixes a compile failure in case wayland
support is autodetected:

    GEN      xdg-shell-unstable-v5-protocol.c
  /bin/sh: /usr/bin/wayland-scanner: No such file or directory
  Makefile:2426: recipe for target 'xdg-shell-unstable-v5-protocol.c' failed
  make[3]: *** [xdg-shell-unstable-v5-protocol.c] Error 127
  make[3]: *** Waiting for unfinished jobs....
    GEN      xdg-shell-unstable-v5-client-protocol.h
  /bin/sh: /usr/bin/wayland-scanner: No such file or directory
  Makefile:2428: recipe for target 'xdg-shell-unstable-v5-client-protocol.h' failed
  make[3]: *** [xdg-shell-unstable-v5-client-protocol.h] Error 127

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

Makefile: add check of binaries architecture

As shown recently by the firejail example, it is easy to miss that a
package builds and installs binaries without actually cross-compiling
them: they are built for the host architecture instead of the target
architecture.

This commit adds a small helper script, check-bin-arch, called as a
GLOBAL_INSTRUMENTATION_HOOKS at the end of the target installation of
each package, to verify that the files installed by this package have
been built for the correct architecture.

Being called as a GLOBAL_INSTRUMENTATION_HOOKS allows the build to error
out right after the installation of the faulty package, and therefore
get autobuilder error detection properly assigned to this specific
package.

Example output with the firejail package enabled, when building for an
ARM target:

ERROR: architecture for ./usr/lib/firejail/libconnect.so is Advanced Micro Devices X86-64, should be ARM
ERROR: architecture for ./usr/bin/firejail is Advanced Micro Devices X86-64, should be ARM
ERROR: architecture for ./usr/lib/firejail/libtrace.so is Advanced Micro Devices X86-64, should be ARM
ERROR: architecture for ./usr/lib/firejail/libtracelog.so is Advanced Micro Devices X86-64, should be ARM
ERROR: architecture for ./usr/lib/firejail/ftee is Advanced Micro Devices X86-64, should be ARM
ERROR: architecture for ./usr/lib/firejail/faudit is Advanced Micro Devices X86-64, should be ARM
ERROR: architecture for ./usr/bin/firemon is Advanced Micro Devices X86-64, should be ARM
ERROR: architecture for ./usr/bin/firecfg is Advanced Micro Devices X86-64, should be ARM

Many thanks to Yann E. Morin and Arnout Vandecappelle for their reviews
and suggestions.

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

arch: add BR2_READELF_ARCH_NAME hidden config option

This config option corresponds to the string returned by readelf for
the "Machine" field of the ELF header. It will be used to check if the
architecture of binaries built by Buildroot match the target
architecture.

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

package/x11r7/xdriver_xf86-video-ati: bump version to 7.9.0

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/x11r7/xserver_xorg-server: bump to version 1.19.3

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Bernd Kuhls <bernd.kuhls@t-online.de>
[Thomas: fix hash file, as noticed by Bernd.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/mesa3d: add lm-sensors optional dependency

radeon-pci-0008
Adapter: PCI adapter
temp1:        +56.0 C  (crit = +120.0 C, hyst = +90.0 C)

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

configs/qemu: update xtensa linux to 4.9.16

This fixes the following nommu kernel build issue:

  arch/xtensa/kernel/setup.c: In function 'cpu_reset':
  arch/xtensa/kernel/setup.c:553:2: error: implicit declaration of function
    'local_flush_tlb_all' [-Werror=implicit-function-declaration]
    local_flush_tlb_all();

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

package/libkcapi: bump version to latest git hash 79cb80714ebc

Configure script is used to detect options supported by the target
compiler, but those options are then incorrectly used by the targets
compiled using a host compiler.

configure:
  checking whether CC supports -fstack-protector-strong... yes
make:
  /usr/bin/gcc (...)
  cc1: error: unrecognized command line option "-fstack-protector-strong"

Fixes:
  http://autobuild.buildroot.net/results/4c0/4c06dfa35ef52f6522619f797a9ae4e7a2e45d04/
  http://autobuild.buildroot.net/results/788/78886670cff406a09bd08808d9a6e328ee75634b/

Signed-off-by: Marcin Nowakowski <marcin.nowakowski@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

python-sortedcontainers: new package

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

gst1-vaapi: fix build when Gstreamer debugging is disabled

When Gstreamer debugging is disabled (BR2_PACKAGE_GSTREAMER1_GST_DEBUG)
we get following build errors:

  gstvaapisink.c: In function 'gst_vaapisink_init':
  gstvaapisink.c:1902:39: error: 'gst_debug_vaapisink' undeclared (first use in this function)
  gstvaapisink.c:1902:39: note: each undeclared identifier is reported only once for each function it appears in
  gstvaapidecode.c: In function 'gst_vaapidecode_init':
  gstvaapidecode.c:1365:63: error: 'gst_debug_vaapidecode' undeclared (first use in this function)
  gstvaapidecode.c:1365:63: note: each undeclared identifier is reported only once for each function it appears in
  gstvaapipostproc.c: In function 'gst_vaapipostproc_init':
  gstvaapipostproc.c:1811:7: error: 'gst_debug_vaapipostproc' undeclared (first use in this function)
  gstvaapipostproc.c:1811:7: note: each undeclared identifier is reported only once for each function it appears in

Add upstream patch to fix these build issues.

Fixes:
  http://autobuild.buildroot.net/results/653/653d9b056b04e589b8f5c57111635b0b3d7c8949

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

package/x265: bump version to 2.3

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

package/libbluray: bump version to 1.0.0

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