package/freerdp: fix dependency on walyand

Currently, wayland support is enabled when the wayland package is
enabled, not when the FreeRDP wayland client is enabled.

But the dependency on libxkbcomon is only enforced from Config.in
when the FreeRDP wayland client is enabled., but is added to build
dependencies when the wayland package is enabled.

As such, we can end up in a situation where the FreeRDP wayland
client is disabled, the wayland package is enabled, and the
libxkbcommon package is also disabled, which casues the build to
fail with:

    Makefile:539: *** libxkbcommon is in the dependency chain of
    freerdp that has added it to its _DEPENDENCIES variable without
    selecting it or depending on it from Config.in.  Stop.

Change the build dependency to actually be on the FreeRDP client
being enabled.

Fixes;
    http://autobuild.buildroot.net/results/dc3e11f7076a8355f3d2f9cb49c6325dcf7084bd

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

qt5webkit: fix URL for 5.6 download

The Qt community releases are not stored under submodules path
component.

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

linux: bump default to version 4.13.16

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

linux-headers: bump 3.{2, 10}.x and 4.{1, 4, 9, 13}.x series

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

dovecot: add applicable licenses

List all code licenses mentioned in COPYING.

Cc: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Fix makefile include order by using sort/wildcard.

The 'include' directive in GNU make supports wildcards, but their
expansion has no defined sort order (GLOB_NOSORT is passed to glob()).
Usually this doesn't matter. However, there is at least one case where
it does make a difference: toolchain/*/*.mk includes both the
definitions of the external toolchain packages and
pkg-toolchain-external.mk, but pkg-toolchain-external.mk must be
included first.

For predictability, use ordered 'include $(sort $(wildcard ...))'
instead of unordered direct 'include */*.mk' everywhere.

Fixes [1] reported by Petr Vorel:

  make: *** No rule to make target 'toolchain-external-custom', needed by '.../build/toolchain-external/.stamp_configured'.  Stop.

[1] http://lists.busybox.net/pipermail/buildroot/2017-November/206969.html

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Tested-by: Petr Vorel <petr.vorel@gmail.com>
[Arnout: also sort the one remaining include, of the external docs]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/google-breakpad: replace references to 'struct ucontext' with 'ucontext_t'

In glibc, since
https://sourceware.org/git/?p=glibc.git;h=251287734e89a52da3db682a8241eb6bccc050c9
the 'struct ucontext' tag has been replaced with 'struct ucontext_t'.
The tag itself is anyway not POSIX - only the 'ucontext_t' typedef is
specified. And that type has existed since at least 1997 in glibc.

Therefore, replace references to 'struct ucontext' with 'ucontext_t',
which works in all versions of glibc, uClibc and musl.

Fixes:
[arm]     http://autobuild.buildroot.net/results/6380341dbb6c114e4452c5cda37da6b44b80d178
[aarch64] http://autobuild.buildroot.net/results/4ecf770df7c984a62082d59f8fab632d3efbe06b
[mipsel]  http://autobuild.buildroot.net/results/e1473a12cf38ccf4dd3ed0f26a8ff9e6b57f0810

Signed-off-by: Romain Naour <romain.naour@gmail.com>
[Arnout: improve commit message]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

parted: properly link with libiconv

parted uses libiconv but doesn't link with it. Add a patch to add it
to LIBADD of the library that uses it. iconv was already checked in
configure.ac, but only if i18n is enabled, so the iconv check is also
added unconditionally in configure.ac.

Also add an optional dependency on libiconv, so it is reproducible.

This was not detected in the autobuilders, since it only occurs when
libiconv exists (otherwise uClibc stubs will be used). libiconv
depends on !BR2_ENABLE_LOCALE and parted depends on BR2_USE_WCHAR. We
don't have such a configuration in the autobuilders.

Upstream status: sent to mailing list
http://lists.alioth.debian.org/pipermail/parted-devel/2017-November/005131.html

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

parted: convert patches to git formatting

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

linux: bump default to version 4.13.15

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

linux-headers: bump 4.{4, 9, 13}.x series

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

DEVELOPERS: Add i2c-tools to Angelo Compagnucci

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

shairport-sync: security bump to version 3.1.4

The bundled tinysvcmdns library is affected by CVE-2017-12087 [1]:

> An exploitable heap overflow vulnerability exists in the tinysvcmdns library
> version 2016-07-18. A specially crafted packet can make the library overwrite
> an arbitrary amount of data on the heap with attacker controlled values. An
> attacker needs send a dns packet to trigger this vulnerability.

shairport-sync has incorparated upstreams fixes in [2].

[1] https://bugs.launchpad.net/bugs/cve/2017-12087
[2] https://github.com/mikebrady/shairport-sync/commit/1dbdf94811b8315705dbac5ba9199d417231c5d3

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

lttng-tools: fix build errors and warnings for musl/uClibc-ng

Add 0001-Fix-detect-dlmopen-and-disable-corresponding-tests-i.patch to
detect if dlmopen() exists and disable parts of the build accordingly if
not. This fixes a bug when building with musl or uClibc-ng.

Add 0002-Fix-src-common-pipe.h-include-sys-types.h-for-ssize_.patch to
fix an #include bug when building with musl.

Add 0003-Fix-warning-src-bin-lttng-utils.c-cast-incompatible-.patch to
remove compilation warnings.

All three new patches are in upstream's master branch as of this date
and will be part of an eventual release.

Fixes:

  http://autobuild.buildroot.net/results/e37a5c69e2a9f9cd7c0705331e205c1ee20808e7/
  (musl)

  http://autobuild.buildroot.net/results/b2670e6d1928649aefa7fbc748858e6036585f01/
  (uclibc)

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

boost: do not allow selecting fiber module on mips32/mips64

The fiber module uses the cpu_relax() macro, for which the MIPS
implementation uses the "pause" instruction, only available since
mips32r2 and mips64r2. In order to avoid build failures on
mips32/mips64, we disallow the selection of the fiber module for such
architecture variants.

This solution was suggested by Arnout Vandecappelle.

Fixes:

  http://autobuild.buildroot.net/results/0439790b6f472e3d4b2d5431a05aa5b408f62e56/

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

DEVELOPERS: Add myself to libepoxy

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

Revert "runc: bump to v1.0.0-rc4"

This reverts commit 3861ce09d0acd6425873332300c6caa75863017f.

As reported on IRC, runc v1.0.0-rc4 is not compatible with the docker-engine
version we are using, so lets revert for 2017.11:

20:27 < eschu> It looks like 25 days ago buildroot had runc updated to
               1.0.0-rc4, which breaks compatibility with the docker-engine
               version 17.05.  Is there somewhere issues like this are
               reported?  Or is it just expected on non-release versions of
               buildroot?

20:47 < Jacmet> eschu: ahh, so the new runc doesn't work with the old
                docker-engine?  what kind of errors do you get exactly?

20:49 < eschu> Jacmet:  yes, runc made a change from -console to
               --console-socket, which results in "Incorrect Usage" error
               "flag provided but not defined: -console"

20:51 < eschu> Jacmet:  basically docker-engine 17.05 uses -console, but
               support for that was cut off in runc around the first 1.0.0
               versions.  I'll check the patches out, or might just
               downgrade to buildroot 2017.08 release

20:51 < agb> Jacmet: unfortunately this issue has hit a number of distributions
             https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=877146

20:53 < Jacmet> agb: thanks.  I'll send a revert for 2017.11 and hopefully
                we can make progress on the other docker packages for
                2018.02

Cc: Christian Stewart <christian@paral.in>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libupnp: add LICENSE hash and DEVELOPERS entry

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

qt5connectivity: fix examples compile without Qt gui module

Add upstream suggested patch to fix heartrate-server (needs only
core and bluetooth modules).

Fixes [1]:

  make[3]: Entering directory `.../build/qt5connectivity-5.9.2/examples/bluetooth'
  cd heartrate-server/ && ( test -e Makefile || .../output/host/bin/qmake -o Makefile .../build/qt5connectivity-5.9.2/examples/bluetooth/heartrate-server/heartrate-server.pro ) && /usr/bin/make -f Makefile
  Project ERROR: Unknown module(s) in QT: gui
  make[3]: *** [sub-heartrate-server-make_first] Error 3
  make[3]: Leaving directory `.../build/qt5connectivity-5.9.2/examples/bluetooth'

[1] http://autobuild.buildroot.net/results/387a21fe7fc672e5668b0afddc138cbb7197d234

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
[Thomas: move patch to 5.9.2 subfolder, since it only applies to 5.9.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

qt5declarative: fix examples compile without OpenGL support

Add upstream patch to fix examples compile without OpenGL support, in
the Qt 5.9 version of qt5declarative.

Fixes [1]:

  In file included from fbitem.h:49:0,
                   from mainwindow.cpp:42:
  logo.h:52:11: error: 'GLfloat' does not name a type
       const GLfloat *constData() const { return m_data.constData(); }
             ^
  logo.h:57:15: error: 'GLfloat' has not been declared
       void quad(GLfloat x1, GLfloat y1, GLfloat x2, GLfloat y2, GLfloat x3, GLfloat y3, GLfloat x4, GLfloat y4);
                 ^
  [...]
  Makefile:610: recipe for target '.obj/mainwindow.o' failed
  make[5]: *** [.obj/mainwindow.o] Error 1
  make[5]: Leaving directory '.../build/qt5declarative-5.9.2/examples/quick/quickwidgets/qquickviewcomparison'
  Makefile:70: recipe for target 'sub-qquickviewcomparison-make_first' failed
  make[4]: *** [sub-qquickviewcomparison-make_first] Error 2
  make[4]: *** Waiting for unfinished jobs....

[1] http://autobuild.buildroot.net/results/cef4abbe043f3aaad03c73e07cfab9d24c3d0fc9

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

localedef: fix xlocale.h related compile failure

Add upstream patch 'Don't include <xlocale.h>'.

Fixes Bug-10501 ([1]):

In file included from ./include/locale.h:1:0,
                 from /usr/include/libintl.h:103,
                 from ./include/libintl.h:2,
                 from glibc/locale/programs/charmap.c:25:
glibc/locale/locale.h:146:11: fatal error: xlocale.h: No such file or directory
 # include <xlocale.h>

[1] https://bugs.busybox.net/show_bug.cgi?id=10501

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

samba4: security bump to version 4.6.11

Fixes the following security issues:

 - CVE-2017-14746:
   All versions of Samba from 4.0.0 onwards are vulnerable to a use after
   free vulnerability, where a malicious SMB1 request can be used to
   control the contents of heap memory via a deallocated heap pointer. It
   is possible this may be used to compromise the SMB server.

 - CVE-2017-15275:
   All versions of Samba from 3.6.0 onwards are vulnerable to a heap
   memory information leak, where server allocated heap memory may be
   returned to the client without being cleared.

   There is no known vulnerability associated with this error, but
   uncleared heap memory may contain previously used data that may help
   an attacker compromise the server via other methods. Uncleared heap
   memory may potentially contain password hashes or other high-value
   data.

For more details, see the release notes:
https://www.samba.org/samba/history/samba-4.6.11.html

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

Revert "gst1-imx: fix v4l2 plugins disable options"

This reverts commit 56b611a972218e33e9fc4b48399138889a688a8e. This
patch was intended for the next branch, not the master branch.

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

gst1-imx: fix v4l2 plugins disable options

Latest bump to v0.13.0 didn't take care of the following upstream
commit:

  7bfbf71 v4l2: Rename to v4l2video and add switches to enable/disable sink/source

To address that, this commit:

 - Uses the new name of the v4l2 source configure script option

 - Introduce a new Config.in option to control the build of the v4l2
   sink support

Fixes:

  http://autobuild.buildroot.net/results/5c404a2a28a4052056e5c9a7edda40641fe55f9e

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

pdbg: bump to latest upstream version to fix build issues

This incorporates a fix supplied by Thomas to resolve build issues on
mips and blackfin.

The patch is dropped as it has been applied upstream.

Fixes:

  http://autobuild.buildroot.net/results/829bee459d7c27392b1c55ae3067607434f0b7f0/
  (mips64el)

  http://autobuild.buildroot.net/results/de51ba7e44a138aec877811d77d3d9b95bfce6b2/
  (blackfin)

  http://autobuild.buildroot.net/results/61ec486869b01f5c77e2cec57b60a1d3ca42af79/
  (mipsel)

Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

pdbg: update help text

Update the help text to reflect features added in recent versions.

Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

jose: fix build with old compilers

This commit adds two patches to the jose package that fix the build
with old compilers such as gcc 4.7.

Fixes:

  http://autobuild.buildroot.net/results/00d5a4b809922f8fa42c0de9d461c9754d2f8098

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

package/feh: Fix build issue with some older cross-compilers

by adding -std=gnu99 to CFLAGS.

Fixes:
http://autobuild.buildroot.net/results/392b58cca2671e206c85a530221ce7b2d42f31fb/
http://autobuild.buildroot.net/results/94527fcd70fa85298c366537013c939534beead0/
http://autobuild.buildroot.net/results/bf31199cdaa18971a9b599c26fbefac73b2d6f30/
http://autobuild.buildroot.net/results/5b65c305b50664e02d5f8022278055c80739342d/
http://autobuild.buildroot.net/results/ed3b9fb8d09cd677af4b8825566877c552926ae5/

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

oracle-mysql: fix build with gcc 7.x

This adjusts an existing patch to also fix gcc 7.x build issues on the
server side of MySQL.

Fixes:

  http://autobuild.buildroot.net/results/9e3a0b5b8f2506fb2ce2fcc3d5d3a494a14439cb

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

kvm-unit-tests: bump version to fix build issue

Since binutils commit 1a9ccd70f9a7 ("Fix the linker so that it will not
silently generate ELF binaries with invalid program headers. Fix readelf
to report such invalid binaries.") the behaviour has changed and
binutils tries to create a program header segment for which there is no
space. Thanks to Nicholas Piggin and his patch to Linux:
https://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux.git/commit/?h=next&id=ff45000fcb56b5b0f

Cedric fixed this in kvm-unit-tests commit 95062c20790d ("powerpc: add
-n to LDFLAGS") so bump to that version.

Since the patch 0001-remove-werror.patch needs to be adjusted, we take
this opportunity to reformat is as a Git patch.

Fixes:

  http://autobuild.buildroot.net/results/80825b492abf61313637b109d6a0944b38d5f739

Signed-off-by: Cyril Bur <cyrilbur@gmail.com>
[Thomas: add necessary hash change, refresh patch that was not
applying anymore.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

board: bananapi-m1: Fix to use dtb instead of dts

boot.cmd using dts file instead of dtb which was
added in below commit, so fix the same.

"board: Add Bananapi M1 support"
(sha1: 5f50fb8d1df51b622537da015f8c3b7b6dcbbc35)

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

qt: add patch to fix build on non-x86 platforms

This commit adds a patch to Qt that adjusts its build system to
properly support building on non-x86 platforms. The problem was due to
the PLATFORM and XPLATFORM variables having the same value when
building on a ppc64le system targetting an uncommon architecture
(Xtensa or ARC for example). We fix this problem by making sure
PLATFORM and XPLATFORM will always be different.

Fixes the following build failures occuring on gcc112 (a ppc64le
autobuilder machine):

  http://autobuild.buildroot.net/results/527747a2fbb31195bb6fdb37cfdaa624e69b7d68/
  (qt on xtensa)

  http://autobuild.buildroot.net/results/7db1265f6f34be3a498881564292517a8fc0e44e/
  (qt on ARC)

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

libfastjson: indicate explicitly which gcc -std option to use

This commit fixes the following build issue of libfastjson with old
enough compilers (4.8) and wchar disabled:

json_object.c: In function 'fjson_object_object_delete':
json_object.c:385:3: error: 'for' loop initial declarations are only allowed in C99 mode
   for (int i = 0 ; i < FJSON_OBJECT_CHLD_PG_SIZE ; ++i) {
   ^

The code of libfastjson requires C99. If your compiler is recent
enough (gcc 5.x), then no problem, it is C99 by default, no additional
flags are needed.

If your compiler is older (for example gcc 4.8), then -std=c99 or
-std=gnu99 is explicitly needed to tell the compiler to accept C99
constructs. Testing the compiler for the availability of such flags is
done by libfastjson configure script. However, the test program used
by the configure script uses some wchar_t types, and therefore the
test checking for C99 availability fails on toolchains with wchar
disabled. From config.log:

configure:3928: checking for /home/test/buildroot/output/host/usr/bin/i586-buildroot-linux-uclibc-gcc option to accept ISO C99
[...]
configure:4077: /home/test/buildroot/output/host/usr/bin/i586-buildroot-linux-uclibc-gcc -std=gnu99 -c -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64  -Os  -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 conftest.c >&5
conftest.c:54:3: error: unknown type name 'wchar_t'
   const wchar_t *name;
   ^

So, just like we did in libv4l in commit
f01396a158f14c53b781c35f7ff29da0bea8c8d6 ("libv4l: fix uclibc-ng
configure/compile"), let's hint directly the configure script that it
should use -std=gnu99. This fixes the build of libfastjson with old
compilers and wchar disabled.

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

support/config-fragments: update ARC pre-built toolchain

This is needed to get the autobuilders to use a toolchain that
includes the fix merged in 9d544feb8a277a43abec889450f917017d666e49
("fwup: fix for ARC toolchain").

In addition, this new toolchain version also fixes for real the RPATH
issue that should have been fixed by
f90f28a6dfe85439d11e5fd83f2462ff29502e11, but wasn't done properly.

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

fwup: fix for ARC toolchain

Building fwup package with ARC toolchain causes internal compiler error.
Adding patch for ARC gcc solves this issue. In the next version of toolchain,
this problem has already been taken into account.

Fixes:
http://autobuild.buildroot.net/results/a1b/a1bd5084309ced6c2d1b311617e5d9f0c362557a//
http://autobuild.buildroot.net/results/dd3/dd3000bd99d71383ce82cd6b01b543cd9660130e//

Signed-off-by: Evgeniy Didin <didin@synopsys.com>
Cc: Alexey Brodkin <abrodkin@synopsys.com>
Cc: arc-buildroot@synopsys.com
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

qt5webkit: backport upstream icu-59 fix from 5.9 branch to 5.6.3

icu-59.x slightly changed the API, breaking qt5webkit builds. For details, see:

https://bugreports.qt.io/browse/QTBUG-60532

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

qt5webkit: correct download URL and hash for 5.6 variant

Commit 06a4975d4bd03 (qt5: bump LTS version to 5.6.3) added an empty hash
for the 5.6.3 variant of qt5webkit, causing failures.

It also forgot to adjust the download URL as the qt5webkit tarballs are no
longer available under official_releases/ like the other submodules, but only
under community_releases/.

Fix both issues.

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

DEVELOPERS: update email address

Prefer microchip email to atmel one.

Signed-off-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

chrony: fix build with musl

The SCM_TIMESTAMPING_PKTINFO macro is missing from the socket.h header
that musl libc provides as of version 1.1.16 that we currently use in
Buildroot. Patch the code to add this macro locally until we update
musl.

Fixes:
http://autobuild.buildroot.net/results/aab/aab3210d6d0c60fb07415d9bbdb09f2d4091446c/
http://autobuild.buildroot.net/results/fc8/fc8440db025f51c176240ca127bc68a367add21c/

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

libpjsip: fix ssl support

Currently, ssl support is implicitely disabled in the initial configure
options. This overrides the check for openssl below.

libpjsip is also currently only compatible with libopenssl. Change
the check to LIBOPENSSL instead of openssl, and depend on libopenssl.

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

linux: bump default to version 4.13.13

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

linux-headers: bump 4.{4, 9, 13}.x series

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

package/samba4: bump version to 4.6.10

Quoting release notes
https://www.samba.org/samba/history/samba-4.6.10.html

"This is an additional bugfix release to address a possible data
 corruption issue. Please update immediately! For details, please see

  https://bugzilla.samba.org/show_bug.cgi?id=13130

 Samba 4.6.0 and newer is affected by this issue."

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

DEVELOPERS: Add myself to iproute2 and iputils

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

mp4v2: add upstream patch fixing build issue with gcc 7+

Fixes:
http://autobuild.buildroot.net/results/fc0/fc0f51f3ee1364f55248e8fe713995bc07f877c6/
http://autobuild.buildroot.net/results/b0e/b0e68ad2aeb7c191cdaa9856a36bd2cae4bcd5f0/
http://autobuild.buildroot.net/results/a66/a66c89e3fb3f1d9802581aaffad87a547c6e67f5/

And many more.

Add an upstream patch fixing invalid code that now triggers a build error on gcc7+.

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

wireguard: bump to version 0.0.20171111

Contains a number of important fixes. From the relase notes:

https://lists.zx2c4.com/pipermail/wireguard/2017-November/001935.html

  * netlink: plug memory leak
  * qemu: check for memory leaks

  There was a small memory leak on the netlink configuration layer that's now
  been fixed.

  * netlink: make sure we reserve space for NLMSG_DONE

  A rather important change - due to an upstream kernel bug, that's existed
  since the advent of netlink itself, sometimes wg(8) failed to receive valid
  data back from kernelspace, resulting in "ENOBUFS" when trying to dump all
  peers. This patch works around it while we wait for upstream to commit the
  fix.

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

package/glibc: switch to downloading a tarball

Currently, we do a full git clone of the repository, which takes quite
some time, especially on slow networks.

This was done like that because the initial patch was using the official
repository as the source of the download, and that repository did not
offer remotely-generated tarballs.

But now we've switched to using a mirror on github, which does provide
such a tarball, which provides faster downloads.

Use that.

However, the tarball from github differs from the one we were generating
locally, because the paths inside are different. WE used to create a
archive with paths starting with glibc-glibc-2.26-73-g4b692dfb95[...],
while github does away with the git-describe prefix, and generates paths
that start with just glibc-4b692dffb95[...]. The content are exactly
identicall (checked with a diff), though.

Update the hash accordingly.

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

support/config-fragments: update ARC toolchain

This commit updates the br-arcle-hs38 configuration fragment to use a
slightly updated ARC toolchain in which the RPATH of the ar and ranlib
tools have been manually fixed to be correct. This is a temporary
workaround until the issue gets resolved properly in the RPATH fixing
logic, and is needed to fix a number of build issues.

Fixes:

  http://autobuild.buildroot.net/results/0ae0d6eba6e2cd8d2298cc4ba6380b8c3e9dd86a/
  (libnl)

  http://autobuild.buildroot.net/results/fa9bdad8d8b57d1db7f22abdb063ade470e9f938/
  (pcsc-lite)

  http://autobuild.buildroot.net/results/fca2ad71c7d446e4738834a1f220d950bce5fddd/
  (alsa-lib)

  http://autobuild.buildroot.net/results/945c60ff942dfd3daa4da27f5e19470e3d97aed4/
  (libcap)

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

Revert "pkg-download: remove explicit PKG_VERSION from github helper"

This reverts commit 1e5a8916b2ab4c9c99548fa6fbd4855eee323881.

The idea was that the version string can be derived because we know the
package name.

However, this patch does not account for the fact that $(pkgname) always
points to the latest pacakge scanned, which in all other situation we're
using it, is the current package, because it is used inside one ot he
xxx-inner macros that are $(eval)ualed. So $(pkgname) is evaluated
"early" and gets the expected value.

However, the github value is not in one of those macros, so it gets
evaluated "late", when doing the actual download. So, by that time,
$(pkgname) will expand to the last package scanned, which is actuall the
manual (without a br2-external tree).

That would require that the _SITE variable be assigned with the :=
assignment operator. This is weird, because that would make it the only
variable to require that, but only when using the github helper, which
is even less obvious and would cause a lot of trouble...

The obvious fixup would seem to be to use $(PKG) instead, because that
already contains the upper-case package name that vcan be used as a
prefix to variables.

However, that does not work either, because we have a check that forbids
a trsailing slash in _SITE, check that is done in pacakge/pkg-generic,
inside the xxx-inner macro, during the $(eval) call.

And at that time, PKG is not yet defined, because it is only defined for
an actual recipe.

So we can't seem to have a workable solution. So, just revert the patch.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Marcin Nowakowski <marcin.nowakowski@imgtec.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Update for 2017.11-rc2

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

package/x11r7/xapp_xdriinfo: fix libgl dependency

Fixes
"mesa3d is in the dependency chain of xapp_xdriinfo that has added it
 to its _DEPENDENCIES variable without selecting it or depending on it
 from Config.in."
http://autobuild.buildroot.net/results/d8a/d8aeed2f64e21a277eb0bc5dc08d2339a14c682e/

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

libpjsip: fix static build failures due to name clash

Several packages have a similar md5.c file, and each has a function
named byteReverse(). This generates link errors when building
statically ("multiple definition of `byteReverse'").

Fix by applying a patch from upstream:
  https://trac.pjsip.org/repos/changeset/5688

Fixes:
  http://autobuild.buildroot.org/results/5d7/5d72e0f8517a555399978d5a0e9f7efd0a278189/
  http://autobuild.buildroot.org/results/c47/c47ccbb9b40011cf0d79c7040bed061ddefd9629/
  http://autobuild.buildroot.org/results/419/419ab2c0e034cc68991281c51caa8271b0fadbab/

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Cc: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/libplist: needs threads

Upstream added a mandatory pthread check:
https://github.com/libimobiledevice/libplist/commit/eec2e855b4f192cb1808d1f02b6bc8935a979025

Fixes
http://autobuild.buildroot.net/results/863/863bb43db222f8e63d60b1fc4a53299575727de1/

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

package/freerdp: fix build of wayland client

Currently, FreeRDP uses pkg-config to find whether wayland is available.
When it is, it gets the CFLAGS and LDFLAGS and so on from pkg-config,
which is OK.

But I also gets the path to the wayland scanner from pkg-config. And
this points to the target one, when we really need the host one.

So we force-feed the correct path as a configure option.

Furthermore, wayland support needs libxkbcommon, so add this missing
dependency as well.

Fixes:
    http://autobuild.buildroot.org/results/bdfe233ade36a56d0f65ef3c3d1698fca548beed/

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/freerdp: fix gstreamer-1.x detection

Backport a patch from upstream.

Partially fixes:
    http://autobuild.buildroot.org/results/81aa66ddd88919295ccb5f34b527b737627263a7/

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/freerdp: optional systemd's journald support

FreeRDP now has support for writing to the journal, so we must ensure
proper dependency and correct options.

Partially fixes:
    http://autobuild.buildroot.org/results/81aa66ddd88919295ccb5f34b527b737627263a7/

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Arnout Vandecappelle <arnout@mind.be>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/freerdp: drop dependency on host-pkgconf

Currently, we only depend on host-pkgconf for gstreamer-0.x

However, we also need it at least for gstreamer-1.x and waylabnd, too.

But since FreeRDP is a cmake package, host-pkgconf is always pulled in
as a dependency by the cmake-package infrastructure.

So, drop the gstreamer-conditional host-pkgconf dependency.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Samuel Martin <s.martin49@gmail.com>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/ffmpeg: fix typo "nios" -> "nios2"

Typo added recently by
https://git.buildroot.net/buildroot/commit/package/ffmpeg?id=bd2544d3d3903935fc4dc0ec9c930720441e545f

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

python-numpy: also disable ARC HS with glibc

Building python-numpy on ARC HS with glibc fails due to missing FE_*
definitions in <fenv.h>. Therefore, python-numpy is only available on
ARC HS with a C library other than glibc.

Fixes:

  http://autobuild.buildroot.net/results/6b9a3310bb4f8c1fd7db0ef4476458b3eec2bf2e/

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

python-numpy: remove trailing whitespace

Remove trailing whitespace introduced by commit
855002f22b4795b7ff3e37880c3ed65011cd036c ("python-numpy: reformat
BR2_PACKAGE_PYTHON_NUMPY_ARCH_SUPPORTS")

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

ruby: security bump to version 2.4.2

Fixed the following security issues:

CVE-2017-0898: Buffer underrun vulnerability in Kernel.sprintf
CVE-2017-10784: Escape sequence injection vulnerability in the Basic
authentication of WEBrick
CVE-2017-14033: Buffer underrun vulnerability in OpenSSL ASN1 decode
CVE-2017-14064: Heap exposure in generating JSON

For more details, see the release notes:
https://www.ruby-lang.org/en/news/2017/09/14/ruby-2-4-2-released/

Drop now upstreamed rubygems patches and add hashes for the license files
while we're at it.

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

package/imagemagick: security bump to version 7.0.7-10

Version 7.0.7-3 fixes CVE-2017-15218:
Stop potential leaks in the JNG decoder

Changelog: https://www.imagemagick.org/script/changelog.php

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

wpa_supplicant: mesh support needs openssl

When BR2_PACKAGE_WPA_SUPPLICANT_MESH_NETWORKING is enabeld,
wpa_supplicant currently fails to build with:

../src/common/sae.o: In function `sae_cn_confirm.isra.2':
sae.c:(.text+0x168): undefined reference to `crypto_bignum_to_bin'
sae.c:(.text+0x198): undefined reference to `crypto_bignum_to_bin'
../src/common/sae.o: In function `sae_cn_confirm_ffc':
sae.c:(.text+0x214): undefined reference to `crypto_bignum_to_bin'
sae.c:(.text+0x22c): undefined reference to `crypto_bignum_to_bin'
../src/common/sae.o: In function `sae_cn_confirm_ecc':
sae.c:(.text+0x2a4): undefined reference to `crypto_ec_point_to_bin'
sae.c:(.text+0x2c0): undefined reference to `crypto_ec_point_to_bin'

[...]

This is due to the fact that the SAE code, used for the mesh network
support, needs OpenSSL support. Therefore, we ensure that
BR2_PACKAGE_WPA_SUPPLICANT_MESH_NETWORKING selects OpenSSL. Only
OpenSSL is supported, which is why
BR2_PACKAGE_OPENSSL_FORCE_LIBOPENSSL is selected as well.

No changes to the .mk files are needed, because we were already
handling OpenSSL as an optional dependency.

This problem was not yet caught by the autobuilders.

Reported-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Cc: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

mesa3d: fix build with BR2_SHARED_STATIC_LIBS

mesa3d does not allow to enable both static and shared libraries so if
BR2_SHARED_STATIC_LIBS is set, disable static

Fixes https://bugs.busybox.net/show_bug.cgi?id=10326

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

vboot-utils: fix ARCH detection

The package includes some target-specific code that is irrelevant
in a host package but gets built anyway. The target for this code
must be one of the supported ChromeOS targets.

Supplied Makefile apparently relies on the environment to provide
a valid target, with a simple fallback to host arch. This breaks
the build if no value is provided and the host arch is not among
the supported ones.

Should fix
http://autobuild.buildroot.net/results/d118a83b6c4f7f910d0d44c279f36251d7ba29e8/
and similar failures.

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

support/config-fragments/autobuild: update toolchains for 2017.11-rc1

All Buildroot pre-built toolchains have been rebuilt with Buildroot
2017.11-rc1, so that they have the latest version of
glibc/musl/uClibc, and also the latest gcc/binutils updates.

Specifically, this will fix the build failures on Blackfin that were
due to the missing accept4() support:

  http://autobuild.buildroot.net/results/8b5a72dd7cde685f6f68f46aeee8b1b60c96d559/
  (openobex)

  http://autobuild.buildroot.net/results/b19dd9ed29944d7f79c6f824669e3baaa0bb045a/
  (libiio)

In terms of changes to the toolchains:

 - AArch64 glibc toolchain changed to use 4.4 kernel headers instead
   of 4.1, in order to increase the variety of kernel header versions
   being tested.

 - Most configurations now use 4.13 kernel headers instead of 4.12
   (except the configurations that were explicitly using an older
   kernel headers version)

 - The mips64 n64 configuration is changed from using gcc 4.9 to gcc
   5, since another ARM configuration already tests gcc 4.9.

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

util-linux: fix link failures due to name clash

Several packages have a similar md5.c file, and each has a function
named byteReverse(). This generates link errors when building
statically ("multiple definition of `byteReverse'").

Fix by applying a patch from upstream.

Fixes:
  http://autobuild.buildroot.net/results/419ab2c0e034cc68991281c51caa8271b0fadbab/

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Cc: Adam Duskett <aduskett@gmail.com>
Tested-by: Carlos Santos <casantos@datacom.ind.br>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

snmp++: security bump to v3.3.10

>From the changelong:
Set the FD_CLOEXEC flag on sockets, so they are not "leaked" to
spawned processes

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

package/collectd: Specify FP layout based on endianness

Big-endian CPUs store floating point as big endian (at lest majority
of them do), so, in order for 'network' plugin to work correctly (and
potentially any user of htond() in collectd's codebase),
--with-fp-layout=endianflip as opposed to --with-fp-layout=nothing
needs to be specified during configuration phase.

Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

python-numpy: disable for ARCompact

This package fails when building for ARCompact due to toolchain issue.

Marking this with special comment "#ARC toolchain issues for
ARCompact" as the package is to be enabled as soon as the issue with
the ARC toolchain is resolved.

Fixes:

  http://autobuild.buildroot.net/results/4c4eaa90335756448007ae3c5ae3839fb17d4442/

Signed-off-by: Evgeniy Didin <didin@synopsys.com>
Cc: Alexey Brodkin <abrodkin@synopsys.com>
Cc: arc-buildroot@synopsys.com
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

python-numpy: reformat BR2_PACKAGE_PYTHON_NUMPY_ARCH_SUPPORTS

Current BR2_PACKAGE_PYTHON_NUMPY_ARCH_SUPPORTS variable layout going
to be inconvenient and ugly for adding new architectures or
restricting specific architectures. Lets reformat layout of this
variable.

Signed-off-by: Evgeniy Didin <didin@synopsys.com>
Cc: Alexey Brodkin <abrodkin@synopsys.com>
Cc: arc-buildroot@synopsys.com
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/openjpeg: fix build with threads disabled

Fixes
http://autobuild.buildroot.net/results/f16/f16a558c30199a92e93a31e0ba391b99a9bd689f/

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

postgresql: security bump to version 9.6.6

Fixes the following security issues:

CVE-2017-12172: Start scripts permit database administrator to modify
root-owned files.

CVE-2017-15098: Memory disclosure in JSON functions.

CVE-2017-15099: INSERT ... ON CONFLICT DO UPDATE fails to enforce SELECT
privileges.

See the announcement for more details:
https://www.postgresql.org/about/news/1801/

While we're at it, also add a hash for the license file.

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

package/ffmpeg: fix fenv build failure on ARCv2 with glibc

https://git.buildroot.net/buildroot/commit/toolchain?id=0633eb58a2912328508a76f8fe9875a0d113a880
added glibc support for ARCv2 which lacks fenv.h.

Code format was suggested by Arnout:
http://lists.busybox.net/pipermail/buildroot/2017-November/206631.html

Fixes
http://autobuild.buildroot.net/results/aec/aec300d7d4472421398a24202d6871c965bd1be5/

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
[Thomas: adjust how the Config.in dependency is expressed.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/ffmpeg: simplify BR2_PACKAGE_FFMPEG_ARCH_SUPPORTS logic

Suggested by Arnout:
http://lists.busybox.net/pipermail/buildroot/2017-November/206631.html

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

DEVELOPERS: add packages to Baruch Siach

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

webkitgtk: security bump to version 2.18.3

This is a maintenance release of the current stable WebKitGTK+ version,
which contains a minor rendering fix, another for the WebDriver
implementation, and security fixes for CVE-2017-13798, CVE-2017-13788,
and CVE-2017-13803.

Release notes:

    https://webkitgtk.org/2017/11/10/webkitgtk2.18.3-released.html

Signed-off-by: Adrian Perez de Castro <aperez@igalia.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libpjsip: security bump to 2.7.1

Also add hash for license file

See release notes for details:
https://trac.pjsip.org/repos/milestone/release-2.7.1

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

package/libuv: bump version to 1.16.0

See https://github.com/libuv/libuv/releases/tag/v1.16.0

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

package/nodejs: bump version to 8.9.1

See https://nodejs.org/en/blog/release/v8.9.1/

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

go-bootstrap: add BR2_PACKAGE_HOST_GO_BOOTSTRAP_ARCH_SUPPORTS

go-bootstrap is a host package that builds a first stage Go compiler,
later used to build the final Go compiler. However, this first stage
compiler only supports building on x86, x86-64 and arm as host
architectures, so we need to add the relevant architecture
dependencies to avoid having go-bootstrap built on other unsupported
platforms.

We do this by introducing BR2_PACKAGE_HOST_GO_BOOTSTRAP_ARCH_SUPPORTS
in a new package/go-bootstrap/Config.in.host file. This option is then
used by BR2_PACKAGE_HOST_GO_ARCH_SUPPORTS to make sure we can't enable
Go packages when the host architecture doesn't allow building the Go
compiler.

Fixes:

  http://autobuild.buildroot.net/results/cbd419c6ab6fa8a6d18dc137c91f895867e53b8a/

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

usb_modeswitch: update config.guess / config.sub

Fixes:
http://autobuild.buildroot.net/results/fbe/fbe21bd173e6566dd93dddd58d4a848de8a3465e/

The build system of the embedded jimtcl copy doesn't use autotools, but does
use an old version of gnuconfig which doesn't know all the architectures
supported by Buildroot, so update config.guess / config.sub like we do in
pkg-autotools.mk

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

jimtcl: update config.guess / config.sub

Fixes:
http://autobuild.buildroot.net/results/284/284876bd649cab26e5ecd696c03c71c077ea7411/

The build system doesn't use autotools, but does use an old version of
gnuconfig which doesn't know all the architectures supported by Buildroot,
so update config.guess / config.sub like we do in pkg-autotools.mk

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

ntp: no longer require openssl

4.2.8p10 no longer requires openssl to compile.

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

linux-headers: bump 4.{4, 9, 13}.x series

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux: bump default to version 4.13.12

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Manual: DEVELOPERS file integrity check

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

DEVELOPERS: fix wireless_tools entry

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

DEVELOPERS: add myself to packages

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

wireshark: bump version to 2.2.10 (security)

Security fixes since 2.2.7:

- wnpa-sec-2017-22
  Bazaar dissector infinite loop (Bug 13599) CVE-2017-9352
- wnpa-sec-2017-23
  DOF dissector read overflow (Bug 13608) CVE-2017-9348
- wnpa-sec-2017-24
  DHCP dissector read overflow (Bug 13609, Bug 13628) CVE-2017-9351
- wnpa-sec-2017-25
  SoulSeek dissector infinite loop (Bug 13631) CVE-2017-9346
- wnpa-sec-2017-26
  DNS dissector infinite loop (Bug 13633) CVE-2017-9345
- wnpa-sec-2017-27
  DICOM dissector infinite loop (Bug 13685) CVE-2017-9349
- wnpa-sec-2017-28
  openSAFETY dissector memory exhaustion (Bug 13649) CVE-2017-9350
- wnpa-sec-2017-29
  BT L2CAP dissector divide by zero (Bug 13701) CVE-2017-9344
- wnpa-sec-2017-30
  MSNIP dissector crash (Bug 13725) CVE-2017-9343
- wnpa-sec-2017-31
  ROS dissector crash (Bug 13637) CVE-2017-9347
- wnpa-sec-2017-32
  RGMP dissector crash (Bug 13646) CVE-2017-9354
- wnpa-sec-2017-33
  IPv6 dissector crash (Bug 13675) CVE-2017-9353
- wnpa-sec-2017-13
  WBMXL dissector infinite loop (Bug 13477, Bug 13796) CVE-2017-7702, CVE-2017-11410
  Note: This is an update for a fix in Wireshark 2.2.6 and 2.0.12.
- wnpa-sec-2017-28
  openSAFETY dissector memory exhaustion (Bug 13649, Bug 13755) CVE-2017-9350, CVE-2017-11411
  Note: This is an update for a fix in Wireshark 2.2.7.
- wnpa-sec-2017-34
  AMQP dissector crash. (Bug 13780) CVE-2017-11408
- wnpa-sec-2017-35
  MQ dissector crash. (Bug 13792) CVE-2017-11407
- wnpa-sec-2017-36
  DOCSIS infinite loop. (Bug 13797) CVE-2017-11406
- wnpa-sec-2017-38
  MSDP dissector infinite loop (Bug 13933)
- wnpa-sec-2017-39
  Profinet I/O buffer overrun (Bug 13847)
- wnpa-sec-2017-41
  IrCOMM dissector buffer overrun (Bug 13929)

Full release notes:

  https://www.wireshark.org/docs/relnotes/wireshark-2.2.10.html

Signed-off-by: André Hentschel <nerv@dawncrow.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

CHANGES: add toolchain info missing from 2017.11-rc1

This was missed from the release update.

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

Update for 2017.11-rc1

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

CHANGES: update with recent changes

Openssl is now a virtual package and new defconfigs.

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

CHANGES: #10361 has been resolved

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

prosody: build with -fPIC

The prosody configure script defines CFLAGS as -fPIC -Wall, but
because we pass our own CFLAGS, the -fPIC goes away, causing a build
failure when building shared libraries on x86-64:

/home/test/autobuild/run/instance-0/output/host/opt/ext-toolchain/bin/../lib/gcc/x86_64-amd-linux-gnu/6.2.0/../../../../x86_64-amd-linux-gnu/bin/ld: encodings.o: relocation R_X86_64_32 against `.rodata' can not be used when making a shared object; recompile with -fPIC

So let's pass this -fPIC flag explicitly in CFLAGS. We don't bother
passing it conditionally !BR2_STATIC_LIBS because other packages (e.g
lua) also pass -fPIC unconditionally.

Fixes:

  http://autobuild.buildroot.net/results/7f0eb285b75cdeacb21a1f9a9f64d2f0cbbf1d59/

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

uboot: use local libfdt.h

Use of libfdt.h by u-boot is currently using
the <system-path>.

Jan submitted the following upstream
https://patchwork.ozlabs.org/patch/833760/

Resolves a failure like this one observed on uboot-tools
http://autobuild.buildroot.net/results/347cde4b5c0e6ca76d354396385be4ec1294da73

[Peter: only sed if file exists]
CC: Jan Kundrát <jan.kundrat@cesnet.cz>
CC: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

uboot-tools: use local libfdt.h

The use of libfdt.h by u-boot is using the
<system-path>.  The libfdt part of uboot
is expriencing a lot of change and this patch
should be overcome by events post uboot-2017.11
release.

Jan submitted the following upstream
https://patchwork.ozlabs.org/patch/833760/

Resolves:
http://autobuild.buildroot.net/results/347cde4b5c0e6ca76d354396385be4ec1294da73
http://autobuild.buildroot.net/results/2b429f82bed1f03d69021e0f959da4e2db2b5a0b
http://autobuild.buildroot.net/results/b18012963926682ba192b9176bd3e35fb07cdf56

CC: Jan Kundrát <jan.kundrat@cesnet.cz>
CC: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

CHANGES: update with recent libssh2 changes

With the revert to the old behavior, mbedtls is no longer the default.

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

libssh2: revert to previous crypto defaults

The recent crypto handling change (commit 04a1031d3: package/libssh2: Add
selectable crypto libraries) had the unfortunate side effect that it no
longer automatically selects the most suitable crypto backend (E.G.  one
where the dependency is already enabled), so all users not wanting to use
the mbedtls backend need to explicitly configure this.

Fix this by inverting the logic so the crypto backend sub options use
'depends on' their dependencies instead of 'select', so only the available
backends are displayed.

Like before, default to openssl if no crypto backend dependencies are
currently enabled.

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

qt5base: add qmake.conf fixup for i.MX platforms

As stated in QTBUG-61156 [1], since Qt5.9 all the Qt apps fail to start
on i.MX platforms.

The reason is that the qmake.conf should apparently now specify which
eglfs backend to use.

It can be seen in the qmake.conf example for i.MX [2].

It has also been integrated into the Yocto meta-freescale layer [3].

Although this issue can be worked around by exporting a variable
(QT_QPA_EGLFS_INTEGRATION=eglfs_viv), this patch aims at providing
a more user-friendly approach by modifying qmake.conf when created.

[1]: https://bugreports.qt.io/browse/QTBUG-61156
[2]: https://github.com/qt/qtbase/blob/5.9/mkspecs/devices/linux-imx6-g%2B%2B/qmake.conf#L35
[3]: https://github.com/Freescale/meta-freescale/blob/master/dynamic-layers/qt5-layer/recipes-qt/qt5/qtbase/0015-Add-eglfs-to-IMX-GPU.patch

[Peter: simplify conditional, add comment explaining why]
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
Tested-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>