cjson: bump to version 1.7.4

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Merge branch 'next'

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

Kickoff 2018.05 cycle

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

czmq: bump to version v4.1.0

Signed-off-by: Asaf Kahlon <asafka7@gmail.com>
[Arnout: correct comment about AUTORECONF]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

docs/website/news.html: add 2018.02 announcement link

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

Update for 2018.02

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

support/tests: enhance the runtime systemd tests

Recent systemd bump has broken DBus dameon and DBus applications can no
longer find the daemon. So we want to catch those kind of failures
early.

We also want to check that the system as a whole is stable: no unit
should be failed.

Finally, ensure that we can read the jounrnal, even when we are doing our
tricks on read-only systems.

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

package/skeleton-systemd: invert factory logic

Currently, we handle the factory by redirectoring /var with a symlink at
build time, and with some trickery during the filesystem generation,
depending on whether we need to remount the filesystem read-write or
not.

However, this is causing quite some pain with the latest systemd, now that
they have moved their dbus socket to /run instead of /var/run.

As such, trying to play tricks with /var/run as a symlink is difficult,
because at times it is in .usr/share/factory/var/run (during build) and
then it is in /var/run (at runtime). So a relative symlink is not
possible. But an absolute symlink is not possible either, because we are
installing out-of-tree.

Oh the joys of cross-compilation... :-)

We fix all this mess by making /var a real directory from the onset, so
that we can use the runtime-expected layout even during the build.

Then, during filesystem generation, we move /var away to the factory,
and populate it as we used to do. This still requires a post-fs hook to
restore /var after the filesystem generation.

This leaves a situation that, should the filesystem generation fails,
/var will be left in an inconsistent state. But that is not worse than
what we already had anyway.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Trent Piepho <tpiepho@impinj.com>
Cc: Adam Duskett <aduskett@gmail.com>
Cc: Romain Naour <romain.naour@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/skeleton-init-systemd: work around for /var/lib not populating

When using a RO root with systemd, it is intended that /var/lib should be
populated at boot time by tmpfiles system mirroring it from
/usr/share/factory/var/lib.

However, this will only happen if /var/lib does not already exist at the
time systemd-tmpfiles runs.  If it does exist, then tmpfiles will
(silently) skip it and do nothing.

It turns out /var/lib will exist, because some part of systemd creates
/var/lib/systemd/catalog on boot before tmpfiles runs.

The fix used here is to also create tmpfiles entries for the contents of
/var/lib/* and /var/lib/systemd/*.  This way, when those directories
already exist, the entire tree is not skipped and instead the
not-yet-existing contents of /var/lib and /var/lib/systemd will be still
be mirrored from the factory dir.

And if /var/lib/systemd, or a prefix of that, stops getting created and
does not exist, it'll still mirror properly.

It does cause some warnings from systemd:
systemd[1]: Starting Create Volatile Files and Directories...
systemd-tmpfiles[148]: [/etc/tmpfiles.d/var-factory.conf:7] Duplicate line for path "/var/lib/systemd", ignoring.
systemd-tmpfiles[148]: [/etc/tmpfiles.d/var-factory.conf:8] Duplicate line for path "/var/lib/systemd/coredump", ignoring.

But they can be ignored.

IMHO, I think a better solution would be for systemd-tmpfiles to gain a
"merge tree" operation that is like "C" but doesn't abort if the
destination exists, but rather merges the source into it.

Signed-off-by: Trent Piepho <tpiepho@impinj.com>
[yann.morin.1998@free.fr: slight rework of commit title]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Tested-by: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux-headers: bump 4.{9, 14}.x series

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

core: drop no-longer used C.UTF-8 locale option

Its use has been globbed into the more generic
BR2_NEEDS_HOST_UTF8_LOCALE option now.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Julius Kriukas <julius@kriukas.lt>
Cc: Christian Stewart <kidovate@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/systemd: needs any UTF-8 locale

Not all distributions have the language-agnostic C.UTF-8 locale (Gentoo,
I'm frowning at you!).

Instead, use any UTF-8 locale provided by the system.

Reported-by: Christian Stewart <kidovate@gmail.com>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Julius Kriukas <julius@kriukas.lt>
Cc: Christian Stewart <kidovate@gmail.com>
Cc: Trent Piepho <tpiepho@impinj.com>
Cc: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

core: find a host UTF-8 locale

Some packages really want to use an UTF-8 locale, or they break.

However, there is no guarantee that any given locale is available on a
system. For example,, while most mainstream distros (Debian and
derivatives, Fedora...) do have the generic, language-agnostic C.UTF-8
locale, Gentoo does not provide it.

So, find the first UTF-8 locale available on the system, and take any
that is available. We however do favour using the user-set current
locale, then using the language-agnostic C.UTF-8, and eventually any
random UTF-8 locale.

Note: we only need to enforce LC_ALL, because setting it implies
everything else:
    http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap08.html#tag_08_02

    """
    1. If the LC_ALL environment variable is defined and is not null,
    the value of LC_ALL shall be used.
    """

[Peter: use same regexp as in dependencies.sh]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

support/dependencies: unbreak check for UTF-8 locale

Although the UTF-8 locales in mainstream distributions all are suffixed
with just 'utf8', the nomenclature is a bit ambiguous with the way they
are to be specified with the various LC_* variables, suffixed there with
'UTF-8'.

Also, POSIX, ISO, and IEC do not enforce any specific suffix in LC_*
variables:
    http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap08.html#tag_08_02

    """
    If the locale value has the form:
        language[_territory][.codeset]

    it refers to an implementation-provided locale, where settings of
    language, territory, and codeset are implementation-defined.
    """

To avoid any confusion, use a regexp that is a bit more lax when
matching locales.

Also, quote the regexp, so that the '?' and '$' are not interpreted by
the shell.

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

package/systemd: create groups required for udevd

udevd needs extra groups for its bundled rules:

    Mar 03 12:21:30 buildroot systemd-udevd[732]: Specified group 'render' unknown
    Mar 03 12:21:30 buildroot systemd-udevd[732]: Specified group 'kvm' unknown

Add those missing groups.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Julius Kriukas <julius@kriukas.lt>
Cc: Trent Piepho <tpiepho@impinj.com>
Cc: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/tor: security bump to version 0.3.1.10

Fixes CVE-2018-0490: null-pointer crash in directory authority protocol list
code.

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

libv4l: fix libdvbv5 musl compile

Add optional copy of TEMP_FAILURE_RETRY macro.

Fixes [1]:

  ../../lib/libdvbv5/.libs/libdvbv5.so: undefined reference to `TEMP_FAILURE_RETRY'

[1] http://autobuild.buildroot.net/results/7aea0cbb9e7fe7d9919c9be04ba4567ddcf4e15e

Patch submitted upstream:
https://www.mail-archive.com/linux-media@vger.kernel.org/msg127134.html

[Peter: add upstream submission link as suggested by Baruch]
Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

dhcp: add upstream security fixes

CVE-2018-5732: The DHCP client incorrectly handled certain malformed
responses. A remote attacker could use this issue to cause the DHCP
client to crash, resulting in a denial of service, or possibly execute
arbitrary code. In the default installation, attackers would be isolated
by the dhclient AppArmor profile.

CVE-2018-5733: The DHCP server incorrectly handled reference counting. A
remote attacker could possibly use this issue to cause the DHCP server
to crash, resulting in a denial of service.

Both issues are fixed in version 4.4.1. But we are close to release, so
backport the fixes instead of bumping version.

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

postgresql: security bump to 10.3

Helps mitigate CVE-2018-1058

see: https://www.postgresql.org/docs/current/static/release-10-3.html for more
bugfixes.

Signed-off-by: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

board/qemu/ppc64le-pseries: make Qemu command similar to ppc64-pseries

In particular:

 - Explicitly specify the CPU to be used, POWER8, which matches
   qemu_ppc64le_pseries_defconfig

 - Use hard disk emulation to access the root filesystem instead of an
   initrd.

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

board/qemu/ppc64le-pseries: put Qemu command on one line

This allows the toolchain building machinery used by
https://toolchains.bootlin.com to automatically re-use this Qemu
command line.

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

board/qemu/ppc64le-pseries: use qemu-system-ppc64

qemu-system-ppc64le doesn't necessarily exist: it isn't installed by
Qemu, and only created as a symlink to qemu-system-ppc64 by some
distributions (Ubuntu). Other distributions (Fedora) just have
qemu-system-ppc64.

But qemu-system-ppc64 is capable of running little-endian PPC64
systems, so use this one instead.

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

support/dependencies/dependencies.sh: check for Python argparse module

The script support/scripts/check-uniq-files uses the argparse Python
module. In most recent Python versions (starting with 2.7), the
argparse module is part of the standard library, and we already check
for the availability of Python in
support/dependencies/dependencies.sh.

However, when running on an ancient distribution with Python 2.6, the
argparse module is not part of the Python standard library, but
available as an external module. Without this module, the build fails,
because check-uniq-files, which is used in target-finalize, fails to
run.

To avoid this failure, this commit adds a check in
support/dependencies/dependencies.sh to verify that the argparse
module is available.

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

support/scripts/check-uniq-files: add indices in format string

Using {} in format strings is only supported in sufficiently recent
Python versions. Python 2.6 doesn't support this, and only format
strings with numbered arguments: {0}, {1}, etc.

Python 2.7:

$ python -c 'print("foo {}".format(12))'
foo 12
$ python -c 'print("foo {0}".format(12))'
foo 12

Python 2.6:

$ python -c 'print("foo {}".format(12))'
Traceback (most recent call last):
  File "<string>", line 1, in <module>
ValueError: zero length field name in format
$ python -c 'print("foo {0}".format(12))'
foo 12

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

mosquitto: unbreak build with websockets and !libopenssl

Fixes:
http://autobuild.buildroot.net/results/d69/d693f3e3f1c73ccf54ac7076623e436355a9d901/b

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

qemu_x86_64_defconfig: fix kernel header version after bump to 4.15

Closes #10831

Commit a952d24b9 (qemu_x86_64: use kernel 4.15) bumped the kernel version
but didn't update the kernel header series to match, causing build failures.

[Peter: reworded]
Signed-off-by: Ciro Santilli <ciro.santilli@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/efl: add missing licenses

ecore_con added with efl 1.17 is under MIT license (COPYING.NGINX-MIT)
ector added with efl 1.15 is under FTL license (COPYING.FTL)

While at it, add license file hash.

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

systemd: switch to C.UTF-8 locale when building

When BR2_REPRODUCIBLE is enabled or host uses non UTF-8 capable locale
building systemd fails with an error:

[1/1080] Generating systemd.bg.catalog with a meson_exe.py custom command.
FAILED: catalog/systemd.bg.catalog
/buildroot/output/host/bin/python3 /buildroot/output/host/bin/meson --internal exe /buildroot/output/build/systemd-236/build/meson-private/meson_exe_sed_232a0623cc7ce2cd67ec72ed784b76307102ed76.dat
Warning: You are using 'ANSI_X3.4-1968' which is not a Unicode-compatible locale.
You might see errors if you use UTF-8 strings as filenames, as strings, or as file contents.
Please switch to a UTF-8 locale for your platform.
...
UnicodeDecodeError: 'ascii' codec can't decode byte 0xd0 in position 1079: ordinal not in range(128)
package/pkg-generic.mk:247: recipe for target '/buildroot/output/build/systemd-236/.stamp_built' failed
make: *** [/buildroot/output/build/systemd-236/.stamp_built] Error 1

This patch changes default host system locale from C to C.UTF-8 when
building systemd package to fix this issue. It also introduces
BR2_NEEDS_HOST_C_UTF8_LOCALE flag that checks if this locale is available on
the host system. If locale is not available error message is show and build
process is stopped.

Signed-off-by: Julius Kriukas <julius@kriukas.lt>
[Thomas: use C.UTF-8 instead of en_US.UTF-8.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libepoxy: bump to version 1.5.0, add license hash

This release of Epoxy makes "0001-Forward-EGL-cflags-into-epoxy.pc.patch"
unnecessary. Other than bugfixes and a couple of functions added to the
API, this release is backwards-compatible with programs built against
the 1.4.x releases.

Signed-off-by: Adrian Perez de Castro <aperez@igalia.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/ffmpeg: fix build without pthreads

Fixes
http://autobuild.buildroot.org/results/4c7/4c77c15eaa2e44d0854cf24ab872e7b2ef83ebd6/

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

mono: only available for x86/x86-64 host architectures

Fixes:

http://autobuild.buildroot.net/results/3cab7b54390feb5f952407f1e7fa49b9633a5f76/
http://autobuild.buildroot.net/results/ad9fc1f8a598d2c1ee7e6f8884f897f7b8bbb657/

And many others.

Mono needs host-mono, which fails to build on the ppc64le autobuilder.
There is unlikely to be any real use cases of people building mono on !x86,
so only make it available on x86/x86-64 hosts, similar to how it was
recently done for erlang.

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Acked-by: Angelo Compagnucci <angelo.compagnucci@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

autofs: disable for musl

autofs requires nsswitch which musl libc does not implement. Commit
873040371b14 (autofs: allow to use libtirpc) removed the
BR2_TOOLCHAIN_HAS_NATIVE_RPC dependency, making autofs available for
musl. Disable musl explicitly.

Fixes:
http://autobuild.buildroot.net/results/a54/a549b62fbf03474ffd8586869637d083de275091/
http://autobuild.buildroot.net/results/b32/b3276b2cbd1e1c260f8df4c3945f51ddc09d492f/
http://autobuild.buildroot.net/results/10b/10b5c5b13971214d0439b715f1a46c78a5249309/

Cc: Waldemar Brodkorb <wbx@openadk.org>
Cc: Jonathan Ben Avraham <yba@tkos.co.il>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

dash: fix build with musl

musl libc requires the getopt.h header for the definition of 'optreset'.

Fixes:
http://autobuild.buildroot.net/results/c23/c233ade479d6124ed7b382f2aa9859fe509d7bca/
http://autobuild.buildroot.net/results/081/0813c13c2a978316e6fa9a3cdbd5512612f5c374/
http://autobuild.buildroot.net/results/182/18230384d8225fbd7fddabfcbd685eaa38e31b31/

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

dash: fix static link with libedit

libedit is an optional dependency of dash since commit 651af57cf011e
(dash: enable line editting if libedit is selected). Unfortunately, the
dash build system does not take into account the ncurses dependency of
libedit. Use pkgconf to get that right.

Fixes:
http://autobuild.buildroot.net/results/801/801e8bc06277978556a301ddf54f609169de37bb/
http://autobuild.buildroot.net/results/123/12367267735b9165afef77289460c247aaad494a/
http://autobuild.buildroot.net/results/aa7/aa7cc7315d778c15c5165ca6423257a84b957482/

Cc: Carlos Santos <casantos@datacom.ind.br>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/php: security bump to version 7.2.3

Fixes CVE 2018-7584: https://bugs.php.net/bug.php?id=75981

For details see release notes:
http://www.php.net/archive/2018.php#id2018-03-01-2

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

package/clamav: security bump to version 0.99.4

Fixes CVE-2012-6706, CVE-2017-6419, CVE-2017-11423, CVE-2018-1000085 &
CVE-2018-0202.

For details see upstream announcement:
http://lists.clamav.net/pipermail/clamav-announce/2018/000029.html

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

uboot: ensure host includes are searched before system default includes

Commit baae5156ce37e (uboot: use local fdt headers) changed the uboot logic
to pass the host include directories with -idirafter instead of -I, so
include files local to u-boot would be preferred over host includes.

This unfortunately breaks configurations using
BR2_TARGET_UBOOT_NEEDS_OPENSSL on hosts with incompatible openssl headers
installed in the system default include directories as explained here:

http://lists.busybox.net/pipermail/buildroot/2018-March/214651.html

The problem is that -idirafter directories gets added to the very end of the
search order, AFTER the system default directories.

Instead use -isystem which causes the directories to be added after -I but
before the system default directories.  With this in place, the include
directories of u-boot will first be scanned, followed by the host includes
and finally the system default include directories.

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

qt: mark as obsolete

qt 4.x is no longer supported upstream since December 2015:

https://blog.qt.io/blog/2015/05/26/qt-4-8-7-released/

So it shouldn't be used for new projects.  Add an obsolete warning to the
package prompt to make this clear.

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

package/busybox: bump version to 1.28.1

Added sha256 hash.

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

18xx-ti-utils: new package

These are spacial RF calibration utilities for TI Wilink 18xx
modules.

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

package/nasm: bump version to 2.13.03

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

package/libdrm: bump version to 2.4.90

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

package/x265: bump version to 2.7

Upstream switched to nasm:
https://mailman.videolan.org/pipermail/x265-devel/2018-March/011615.html

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

package/libvpx: bump version to 1.7.0

Switched _SITE to github, current _SITE is broken.

The URL
https://chromium.googlesource.com/webm/libvpx/+archive/v1.7.0.tar.gz

proposed by upstream
https://storage.googleapis.com/downloads.webmproject.org/releases/webm/index.html

provides archives with different hashes for each download so we use
Github instead.

Added license hashes.

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

package/libmicrohttpd: bump version to 0.9.59

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

mosquitto: security bump to version 1.4.15

Fixes CVE-2017-7651: Unauthenticated clients can send a crafted CONNECT
packet which causes large amounts of memory use in the broker.  If multiple
clients do this, an out of memory situation can occur and the system may
become unresponsive or the broker will be killed by the operating system.

The fix addresses the problem by limiting the permissible size for CONNECT
packet, and by adding a memory_limit configuration option that allows the
broker to self limit the amount of memory it uses.

The hash of new tarball is not (yet) available through download.php, so use
a locally calculated hash.

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

package/dovecot: security bump to version 2.3.4

Fixes CVE-2017-15130, CVE-2017-14461 & CVE-2017-15132:
https://www.dovecot.org/list/dovecot-news/2018-February/000370.html

Removed patch applied upstream:
https://github.com/dovecot/core/commit/a008617e811673064fd657acf517dc4a12493d29

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

systemd: correct README hash after bump to version 237

systemd-237 slightly changed the README, but the hash was not updated
leading to legal-info failing.  Fix it.

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

utils/scanpypi: fix 'downloas' typo in error message

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

gnutls: bump to version 3.5.18

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

lzip: bump to version 1.20

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

iw: bump to version 4.14

Drop IW_MAKE_OPTS, passing the CC, LD, and LDFLAGS in the environment
via TARGET_CONFIGURE_OPTS is enough.

Add license file hash.

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

gnupg2: bump to version 2.2.5

Cc: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

sdl2: rpi video support needs OpenGL ES

Fixes [1]:

  .../build/sdl2-2.0.7/src/video/raspberry/SDL_rpivideo.c: In function 'RPI_Create':
  .../build/sdl2-2.0.7/src/video/raspberry/SDL_rpivideo.c:126:39: error: 'RPI_GLES_DefaultProfileConfig' undeclared (first use in this function); did you mean 'RPI_GLES_DeleteContext'?
       device->GL_DefaultProfileConfig = RPI_GLES_DefaultProfileConfig;
                                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
                                         RPI_GLES_DeleteContext
  .../build/sdl2-2.0.7/src/video/raspberry/SDL_rpivideo.c:126:39: note: each undeclared identifier is reported only once for each function it appears in
  .../build/sdl2-2.0.7/src/video/raspberry/SDL_rpivideo.c: In function 'RPI_CreateWindow':
  .../build/sdl2-2.0.7/src/video/raspberry/SDL_rpivideo.c:281:17: error: 'SDL_VideoDevice {aka struct SDL_VideoDevice}' has no member named 'egl_data'; did you mean 'gl_data'?
       if (!_this->egl_data) {
                   ^~~~~~~~
                   gl_data
  .../build/sdl2-2.0.7/src/video/raspberry/SDL_rpivideo.c:286:10: error: 'SDL_WindowData {aka struct SDL_WindowData}' has no member named 'egl_surface'
       wdata->egl_surface = SDL_EGL_CreateSurface(_this, (NativeWindowType) &wdata->dispman_window);
            ^~
  .../build/sdl2-2.0.7/src/video/raspberry/SDL_rpivideo.c:286:26: warning: implicit declaration of function 'SDL_EGL_CreateSurface'; did you mean 'SDL_Vulkan_CreateSurface'? [-Wimplicit-function-declaration]
       wdata->egl_surface = SDL_EGL_CreateSurface(_this, (NativeWindowType) &wdata->dispman_window);
                            ^~~~~~~~~~~~~~~~~~~~~
                            SDL_Vulkan_CreateSurface
  .../build/sdl2-2.0.7/src/video/raspberry/SDL_rpivideo.c:288:14: error: 'SDL_WindowData {aka struct SDL_WindowData}' has no member named 'egl_surface'
       if (wdata->egl_surface == EGL_NO_SURFACE) {
                ^~

[1] http://autobuild.buildroot.net/results/9612d43b192bbb88214a11fe18f8b8da6ad10313

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux-headers: bump 4.{4, 9, 14, 15}.x series

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux: bump default to version 4.15.7

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/tvheadend: disable for microblazeel

tvheadend does not build on microblazeel, because of some assertion
failure in binutils:

    [...]/microblazeel-buildroot-linux-uclibc/bin/ld: FDE encoding in [...]/lib/gcc/microblazeel-buildroot-linux-uclibc/6.4.0/libgcc.a(_divdi3.o)(.eh_frame) prevents .eh_frame_hdr table being created.
    [...]/microblazeel-buildroot-linux-uclibc/bin/ld: FDE encoding in [...]/lib/gcc/microblazeel-buildroot-linux-uclibc/6.4.0/libgcc.a(_udivdi3.o)(.eh_frame) prevents .eh_frame_hdr table being created.
    [...]/microblazeel-buildroot-linux-uclibc/bin/ld: FDE encoding in [...]/lib/gcc/microblazeel-buildroot-linux-uclibc/6.4.0/libgcc.a(_umoddi3.o)(.eh_frame) prevents .eh_frame_hdr table being created.
    [...]/microblazeel-buildroot-linux-uclibc/bin/ld: BFD (GNU Binutils) 2.29.1 assertion fail elf32-microblaze.c:1494
    [...]/microblazeel-buildroot-linux-uclibc/bin/ld: [...]/sysroot/usr/lib/Scrt1.o: probably compiled without -fPIC?
    [...]/microblazeel-buildroot-linux-uclibc/bin/ld: final link failed: Bad value
    collect2: error: ld returned 1 exit status

Fixes:
    http://autobuild.buildroot.org/results/85475885a95f23f3dbc88e5b162108a458233bc4/
    [...]

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

wavpack: add upstream security fixes

Fixes the following security issues:

CVE-2018-6767: A stack-based buffer over-read in the ParseRiffHeaderConfig
function of cli/riff.c file of WavPack 5.1.0 allows a remote attacker to
cause a denial-of-service attack or possibly have unspecified other impact
via a maliciously crafted RF64 file.

CVE-2018-7253: The ParseDsdiffHeaderConfig function of the cli/dsdiff.c file
of WavPack 5.1.0 allows a remote attacker to cause a denial-of-service
(heap-based buffer over-read) or possibly overwrite the heap via a
maliciously crafted DSDIFF file.

CVE-2018-7254: The ParseCaffHeaderConfig function of the cli/caff.c file of
WavPack 5.1.0 allows a remote attacker to cause a denial-of-service (global
buffer over-read), or possibly trigger a buffer overflow or incorrect memory
allocation, via a maliciously crafted CAF file.

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

Update for 2018.02-rc3

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

bluez5_utils: Make systemd link correctly

The link in /etc/systemd/system/, named dbus-org.bluez.service, pointing to
../../../../usr/lib/systemd/system/bluetooth.service has one too many ".." in
it.  Likely the target was copied from the other link in
/etc/systemd/system/bluetooth.target.wants/, which is one level deeper.

Adjust link to be correct.

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

azmq: bump to version 1.0.2

Add license hash

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

bind: use BIND_PKGDIR vairable

Use the BIND_PKGDIR variable instead of package/bind.

Signed-off-by: Christopher McCrory <chrismcc@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

systemd: Fix relative ln add-wants wrapper

The patch to allow systemd to work with old "ln" versions that don't
support --relative didn't work properly in the the meson-add-wants.sh
script.

This results in all the links in systemd's "*.wants" directories being
broken, e.g.
/usr/lib/systemd/system/multi-user.target.wants/getty.target ->
  ../../../../usr/lib/systemd/system/getty.target
There is one too few ".." in that relative link.

The problem is that the script is called with the link name being either a
file or an existing directory.  In the latter case, ln creates the link in
the directory using the name of the target.  This means the link is one
level deeper than the relative link making code thinks.

The solution used is to only dirname the link, moving up a level, if it's
not a directory, to mimic ln's logic in how it creates links.

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

pcre2: add patch fixing a build issue on m68k

A pretty nasty difference in alignment rules between most
architectures and m68k lead pcre2 to fail building its 16-bit and
32-bit variants on m68k. This commit adds a patch that fixes that.

Fixes:

  http://autobuild.buildroot.net/results/f6a45df5cb80e3fd94d57163bd28a0014a02bf4/

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

qt5multimedia: fix build issue with mesa3d w/out xcb

The mesa's EGL/eglplatform.h header includes X11 headers unless the flag
MESA_EGL_NO_X11_HEADERS is defined[1].

A build issue happens when mesa3d is selected as then OpenGL EGL backend
but the XCB library is not selected.

This commit tells qmake to pass the cflag MESA_EGL_NO_X11_HEADERS to
make and prevent from including the missing X headers.

The issue QTBUG-66233 is opened in the Qt tracker[1].

Fixes:

In file included from
/home/gportay/src/buildroot/output-qt5.10/host/x86_64-buildroot-linux-gnu/sysroot/usr/include/EGL/egl.h:39:0,
                 from qsgvideonode_egl.h:48,
                 from qsgvideonode_egl.cpp:40:
/home/gportay/src/buildroot/output-qt5.10/host/x86_64-buildroot-linux-gnu/sysroot/usr/include/EGL/eglplatform.h:118:22:
fatal error: X11/Xlib.h: No such file or directory
 #include <X11/Xlib.h>
                      ^
compilation terminated.
Makefile:550: recipe for target '.obj/qsgvideonode_egl.o' failed

[1]: https://github.com/mesa3d/mesa/blob/79ee1b2ff0b85f4eeb4165d23a7943c28d3a3d93/include/EGL/eglplatform.h#L109-L125
[2]: https://bugreports.qt.io/browse/QTBUG-66233

[Peter: simplify logic]
Cc: Julien CORJON <corjon.j@ecagroup.com>
Signed-off-by: Gaël PORTAY <gael.portay@savoirfairelinux.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

qt5webkit: fix build issue with mesa3d w/out xcb

The mesa's EGL/eglplatform.h header includes X11 headers unless the flag
MESA_EGL_NO_X11_HEADERS is defined[1].

A build issue happens when mesa3d is selected as then OpenGL EGL backend
but the XCB library is not selected.

This commit tells qmake to pass the cflag MESA_EGL_NO_X11_HEADERS to
make and prevent from including the missing X headers.

The issue QTBUG-66233 is opened in the Qt tracker[1].

Fixes:

In file included from /home/gportay/src/buildroot/output-qt5.10/host/x86_64-buildroot-linux-gnu/sysroot/usr/include/EGL/egl.h:39:0,
                 from platform/graphics/opengl/Extensions3DOpenGLES.cpp:33:
/home/gportay/src/buildroot/output-qt5.10/host/x86_64-buildroot-linux-gnu/sysroot/usr/include/EGL/eglplatform.h:118:22: fatal error: X11/Xlib.h: No such file or directory
 #include <X11/Xlib.h>
                      ^
compilation terminated.
Makefile.WebCore.Target:93596: recipe for target '.obj/platform/graphics/opengl/Extensions3DOpenGLES.o' failed

[1]: https://github.com/mesa3d/mesa/blob/79ee1b2ff0b85f4eeb4165d23a7943c28d3a3d93/include/EGL/eglplatform.h#L109-L125
[2]: https://bugreports.qt.io/browse/QTBUG-66233

[Peter: simplify logic]
Cc: Julien CORJON <corjon.j@ecagroup.com>
Signed-off-by: Gaël PORTAY <gael.portay@savoirfairelinux.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

qt5wayland: fix build issue with mesa3d w/out xcb

The mesa's EGL/eglplatform.h header includes X11 headers unless the flag
MESA_EGL_NO_X11_HEADERS is defined[1].

A build issue happens when mesa3d is selected as then OpenGL EGL backend
but the XCB library is not selected.

This commit tells qmake to pass the cflag MESA_EGL_NO_X11_HEADERS to
make and prevent from including the missing X headers.

The issue QTBUG-66233 is opened in the Qt tracker[1].

Fixes:

In file included from
/home/gportay/src/buildroot/output-qt5.10/host/x86_64-buildroot-linux-gnu/sysroot/usr/include/EGL/egl.h:39:0,
                 from
/home/gportay/src/buildroot/output-qt5.10/host/x86_64-buildroot-linux-gnu/sysroot/usr/include/qt5/QtEglSupport/5.10.0/QtEglSupport/private/qt_egl_p.h:65,
                 from
/home/gportay/src/buildroot/output-qt5.10/host/x86_64-buildroot-linux-gnu/sysroot/usr/include/qt5/QtEglSupport/5.10.0/QtEglSupport/private/qeglstreamconvenience_p.h:55,
                 from
../../../../hardwareintegration/compositor/wayland-egl/waylandeglclientbufferintegration.cpp:56:
/home/gportay/src/buildroot/output-qt5.10/host/x86_64-buildroot-linux-gnu/sysroot/usr/include/EGL/eglplatform.h:118:22:
fatal error: X11/Xlib.h: No such file or directory
 #include <X11/Xlib.h>
                      ^
compilation terminated.
Makefile:656: recipe for target
'.obj/waylandeglclientbufferintegration.o' failed
make[8]: *** [.obj/waylandeglclientbufferintegration.o] Error 1

[1]: https://github.com/mesa3d/mesa/blob/79ee1b2ff0b85f4eeb4165d23a7943c28d3a3d93/include/EGL/eglplatform.h#L109-L125
[2]: https://bugreports.qt.io/browse/QTBUG-66233

[Peter: simplify logic]
Cc: Julien CORJON <corjon.j@ecagroup.com>
Signed-off-by: Gaël PORTAY <gael.portay@savoirfairelinux.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

domoticz: depends on NPTL

boost 1.66 uses pthread_condattr_setclock in
boost/asio/detail/impl/posix_event.ipp.

Boost is mostly a header-only library, in particular the asio stuff.
There are sub-options for the parts of Boost that are not purely
headers, but there are not any sub-options for the parts of Boost that
are just headers.

So, the options are either to add sub-options to Boost, or to make
Boost as a whole only available on NPTL toolchains. The latter is a bit
annoying as it would mean adding this dependency to all Boost reverse
dependencies, even those not using asio.

So, instead of updating boost, add a dependency to
BR2_TOOLCHAIN_HAS_THREADS_NPTL to domoticz. This is the best solution
that can be done today. It's not perfect, but it's good enough.

Fixes:
 - http://autobuild.buildroot.net/results/197baa15cbf7f4fc7b3ccc602515af3f375ed68f

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

mpv: security bump to 0.27.2

Fixes CVE-2018-6360: mpv through 0.28.0 allows remote attackers to execute
arbitrary code via a crafted web site, because it reads HTML documents
containing VIDEO elements, and accepts arbitrary URLs in a src attribute
without a protocol whitelist.

[Peter: Add CVE description]
Signed-off-by: Mahyar Koshkouei <mahyar.koshkouei@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

mpv: disable mali-fbdev support

Fixes:
http://autobuild.buildroot.net/results/2ce/2ce2d9be9e0699114e3bc3c0434ba05f64741f89/

A compile error occurs when attempting to compile mpv with mali support
(eg. when odroid-mali package is selected).

Signed-off-by: Mahyar Koshkouei <mahyar.koshkouei@gmail.com>
Signed-off-by: Mahyar Koshkouei &lt;mahyar.koshkouei@gmail.com&gt;<br/>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

asterisk: replace $(HOST_DIR)/usr/bin with $(HOST_DIR)/bin

Commit 0f9c0bf3d5cd did a global replace of $(HOST_DIR)/usr/bin to
$(HOST_DIR)/bin. But later, a new occurrence of $(HOST_DIR)/usr/bin snuck
into the sources via commit 05e306d8d33 which added asterisk.

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/{mesa3d, mesa3d-headers}: bump version to 17.3.6

Release note of this emergency release:
https://lists.freedesktop.org/archives/mesa-announce/2018-February/000403.html

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

linux-headers: bump 3.2.x series

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

qt5: bump latest version to 5.9.4

qt5connectivity:
 - removed
 0001-Example-heartrate-server-needs-only-core-and-bluetoo.patch
 (upstream comitted [1])

qt5quickcontrols2:
 - removed 0001-Example-flatstyle-needs-widgets-module.patch (upstream
 comitted [2])

qt5webengine:
 - removed 0002-Load-libEGL-and-libGLES2-symbols-implicitly.patch
 (upstream comitted, reverted and recomitted [3])

[1] https://codereview.qt-project.org/211810
[2] https://codereview.qt-project.org/212583
[3] https://codereview.qt-project.org/209547

Signed-off-by: Sébastien Szymanski <sebastien.szymanski@armadeus.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

DEVELOPERS: Update email address

I am leaving Xilinx so to avoid future bounces update my email address
to my personal email address.

Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Signed-off-by: Alistair Francis <alistair@alistair23.me>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

check-host-tar.sh: blacklist tar 1.30+

Tar 1.30 changed the --numeric-owner output for filenames > 100 characters,
leading to hash mismatches for the tar archives we create ourselves from
git.  This is really a fix for a bug in earlier tar versions regarding
deterministic output, so it is unlikely to be reverted in later versions.

For more details, see:
http://lists.busybox.net/pipermail/buildroot/2018-January/211222.html

To work around this issue, blacklist tar 1.30+ similar to how we do it for
pre-1.17 versions so Buildroot falls back to building host-tar.

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

dependencies.mk: check for valid host-tar before other host dependencies

host-{cmake,lzip,xz} needs host-tar to extract their source code tarball, so
we need to ensure that host-tar gets added to DEPENDENCIES_HOST_PREREQ
before these in case they are both needed, otherwise the tools will fail to
extract.

With the upcoming change to blacklist modern tar versions this situation is
likely to trigger more often.

The real solution to this issue is the <foo>_EXTRACT_DEPENDENCIES rework,
but that series is a bit too intrusive to add this close to 2018.02, so
therefore this hack.

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Reviewed-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

php: disable on configurations using BR2_BINFMT_FLAT

BR2_BINFMT_FLAT configurations require the use of elf2flt. However,
PHP uses -export-dynamic which breaks badly with elf2flt. Even a
simple program fails to build:

$ ./output/host/bin/arm-linux-gcc -Wl,-export-dynamic -o toto toto.c
/home/thomas/projets/buildroot/output/host/opt/ext-toolchain/arm-buildroot-uclinux-uclibcgnueabi/bin/ld.real: section .junk LMA [0000000000000000,0000000000000027] overlaps section .text LMA [0000000000000000,0000000000006d07]

-export-dynamic is clearly not useful for FLAT configurations, which
are always statically linked, but it's quite a bit of work to change
the PHP build system to use it only conditionall.

It looks more like an interaction bug between gcc (which wants to put
the .text section at address 0x0 in its linker script) and elf2flt,
which wants to put its .junk section (containing the .rel.text stuff)
also at address 0x0.

Fixes (works around) the "section overlap" part of:

  http://autobuild.buildroot.net/results/35cbed8927bb10500ecf2816aa728ea240a0be21/

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

php: add patch to fix build on static-only systems

This commit adds a patch to PHP to fix the build on static-only
system, a regression introduced in PHP recently.

Fixes:

  http://autobuild.buildroot.net/results/fbf7ebbb9502424727006f39e169ec1ee870186d/

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

rustc: disable for ARMV7A with EABI

The Rust project does not support armv7a EABI, only EABIHF [1]. So
disable rustc for this ARM variant.

Fixes:

- http://autobuild.buildroot.net/results/51c98ca5486e7e83725ababa437cbb8204ceb66b/

[1] https://forge.rust-lang.org/platform-support.html

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

erlang: enable host-erlang for x86/x86_64 architectures only

And make erlang depends on host-erlang.
This patch aims at avoiding the following autobuilder failures:
http://autobuild.buildroot.net/results/45edf95c0c44c9d553879e0cbb771098d7c63aa1
http://autobuild.buildroot.net/results/a36d00407a371d70b4551a9717ebd6ff852c8bca

Signed-off-by: Johan Oudinet <johan.oudinet@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

php: fix build on C libraries that don't provide AI_IDN

The PHP sockets module now provides access to AI_IDN functionality (by
simply declaring a bunch of constants available from PHP code). AI_IDN
(internationalized domain names) is not supported in uClibc or musl,
which breaks the build.

This commit adds a patch to PHP which makes the use of AI_IDN
conditional, in a way that is identical to what is already done for
AI_ALL. The patch has been submitted upstream.

Fixes:

  http://autobuild.buildroot.net/results/ef040cc45bb2789ead82a8c445db3376e36b5ac7/

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

scanpypi: fix licence detection handling for unknown licences

Check for match object not being None.

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

scanpypi: fix Py2/3 conversion leftover

Use urlparse from six package.

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

fs/cpio/init: preserve arguments

Use "$@" instead of $* to preserve arguments containing spaces.

The shell expands "$@" as "$1" "$2" "$3"... while it expands $@ as $1 $2
$3. With the second form, we loses spaces in positional parameters.

Signed-off-by: Gaël PORTAY <gael.portay@savoirfairelinux.com>
Reviewed-by: Adrian Perez de Castro <aperez@igalia.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux-headers: bump 4.{4, 9, 14, 15}.x series

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux: bump default to version 4.15.6

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

docs/manual: add documentation for the "make sdk" target

Signed-off-by: Daniel Serpell <daniel.serpell@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/zlib: zlib-ng only available on some archs

Upstream has code only for those architectures, with no fallback
generic code, so we must restrict zlib-ng to only those four archs...

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Nicolas Cavallari <Nicolas.Cavallari@green-communications.fr>
Cc: Adam Duskett <aduskett@gmail.com>
Cc: Stefan Fröberg <stefan.froberg@petroprogram.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
[Thomas: fix typoes in the Config.in option name.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/htop: remove 0xe2 from MetersPanel.c

This non ascii character trigger an issue with MakeHeader.py.

./scripts/MakeHeader.py MetersPanel.c
Traceback (most recent call last):
  File "./scripts/MakeHeader.py", line 32, in <module>
    for line in file.readlines():
  File "[...]/output/host/lib/python3.6/encodings/ascii.py", line 26, in decode
    return codecs.ascii_decode(input, self.errors)[0]
UnicodeDecodeError: 'ascii' codec can't decode byte 0xe2 in position 956: ordinal not in range(128)

Fixes:
http://autobuild.buildroot.net/results/8ed/8edb78b054a265447fd0e83f67cba5f978be4ed6

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

Revert "package/htop: remove 0xe2 from MetersPanel.c"

This reverts commit 93f080ba0a3d84bf3b23404e083b0a33cd5fa31d.

This patch was intended for the next branch, it shouldn't have been
applied to the master branch.

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

package/htop: remove 0xe2 from MetersPanel.c

This non ascii character trigger an issue with MakeHeader.py.

./scripts/MakeHeader.py MetersPanel.c
Traceback (most recent call last):
  File "./scripts/MakeHeader.py", line 32, in <module>
    for line in file.readlines():
  File "[...]/output/host/lib/python3.6/encodings/ascii.py", line 26, in decode
    return codecs.ascii_decode(input, self.errors)[0]
UnicodeDecodeError: 'ascii' codec can't decode byte 0xe2 in position 956: ordinal not in range(128)

Fixes:
http://autobuild.buildroot.net/results/8ed/8edb78b054a265447fd0e83f67cba5f978be4ed6

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

check-bin-arch: skip /lib/modules to allow 32-bit userland on 64-bit arch

The script check-bin-arch fails as follows on a config for PowerPC e6500
(64-bit CPU) with BR2_ARCH="powerpc" (32-bit userland desired):

ERROR: architecture for "/lib/modules/..../lib/libcrc32c.ko"
            is "PowerPC64", should be "PowerPC"

This situation is perfectly acceptable: the kernel is 64-bit and so are its
modules, even though userland is 32-bit.

To keep check-bin-arch and its caller simple, just skip /lib/modules/
entirely, like is done for /lib/firmware and some others.

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

mbedtls: fix API compatibility

Add upstream patch fixing API compatibility with previous releases.

Fixes (hiawatha):
http://autobuild.buildroot.net/results/ce6/ce6b4a50e6aafd06f82eaae688dd8720b982e9c2/
http://autobuild.buildroot.net/results/cde/cdec7ae3565d5b76a9bc50156c6244b44197534e/
http://autobuild.buildroot.net/results/9c1/9c1aec09c03f60bee9dc134da5a29e2671fc3b5e/

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

gcc: add the real gcc bugfix for m68k from upstream

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

uboot: revert "uboot: use local libfdt.h"

This reverts commit 3a6573ccee2624de0c604abf2c7df6704a4cf566.

It is no longer necessary after solving the problem differently.

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

uboot: use local fdt headers

After commit b8c3e941731d ("package/dtc: qemu system build need libfdt")
changed the dtc install target from 'install-bin' to 'install', uboot
compilation failures occurred because libfdt related headers were now
suddenly taken from output/host/include rather than from the uboot sources
itself.

Commit 3a6573ccee26 ("uboot: use local libfdt.h") solved this by patching
one specific uboot source file, tools/fdtgrep.c, to replace '<...>'-style
includes by '"..."'-style includes.

However, depending on the uboot version, this may not be enough: there may
be other references to fdt header files. In particular taking into account
that it is not uncommon to have vendor-provided uboot trees which have
custom changes.

The root of the problem is that the uboot.mk file passes the host compiler
as follows:
UBOOT_MAKE_OPTS += \
...
HOSTCC="$(HOSTCC) $(HOST_CFLAGS)" \
...
where HOST_CFLAGS contains the string '-I$(HOST_DIR)/include'
The uboot makefiles then use constructs of the form:
$(CC) $(CPPFLAGS) $(CFLAGS) .....
where CPPFLAGS may contain -I references pointing to local directories.

On the expanded compiler command-line, Buildroot's '-I$(HOST_DIR)/include'
is thus present _before_ any -I to local directories, and thus takes
precedence.  And that becomes a problem for header files present both
locally as in the Buildroot host directory, which is the case for libfdt.

To fix this problem without having to patch u-boot sources, use '-idirafter'
rather than '-I' to pass the Buildroot host include directory. '-idirafter'
is basically the same thing, but adds the specified directory at the end
of the include precedence chain, rather than at the beginning.

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libglib2: don't override ac_cv_func_strerror_r_char_p

libglib2 recently changed to use the result of the autoconf macro to
decide how to use strerror_r() in g_strerror() instead of embedding the
same preprocessor condition as in glibc's strings.h (upstream commit
c8e268bb was first included in release 2.53.4).

Following this change, if ac_cv_func_strerror_r_char_p is incorrectly
set to "no", the error string is an uninitialized buffer which cannot be
encoded as UTF-8.  The final result of this is that GLib functions that
are expected to fill in an error pointer on failure in fact leave this
pointing to NULL which is likely to cause a segfault in client
applications.

In fact the autoconf check compiles a test file but does not need to run
it, so the test is safe when cross-compiling and returns the correct
answer.  So remove this cached value and let the configure script figure
it out for itself, fixing g_strerror() on glibc systems.

Signed-off-by: John Keeping <john@metanate.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

scanpypi: add support for Python3

The script was changed via modernize utility. The only manual
made part was the handling of StringIO.

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

asterisk: security bump to version 14.7.6

Fixes the following security issues:

AST-2018-002: Crash when given an invalid SDP media format description

By crafting an SDP message with an invalid media format description Asterisk
crashes when using the pjsip channel driver because pjproject's sdp parsing
algorithm fails to catch the invalid media format description.

AST-2018-003: Crash with an invalid SDP fmtp attribute

By crafting an SDP message body with an invalid fmtp attribute Asterisk
crashes when using the pjsip channel driver because pjproject's fmtp
retrieval function fails to check if fmtp value is empty (set empty if
previously parsed as invalid).

AST-2018-004: Crash when receiving SUBSCRIBE request

When processing a SUBSCRIBE request the res_pjsip_pubsub  module stores the
accepted formats present in the Accept headers of the request.  This code
did not limit the number of headers it processed despite having a fixed
limit of 32.  If more than 32 Accept headers were present the code would
write outside of its memory and cause a crash.

AST-2018-005: Crash when large numbers of TCP connections are closed suddenly

A crash occurs when a number of authenticated INVITE messages are sent over
TCP or TLS and then the connection is suddenly closed.  This issue leads to
a segmentation fault.

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