toolchain: granular choice for stack protector

Currently, we only support two levels of stach-smashing protection:
  - entirely disabled,
  - protect _all_ functions with -fstack-protector-all.

-fstack-protector-all tends to be far too aggressive and impacts
performance too much to be worth on a real product.

Add a choice that allows us to select between different levels of
stack-smashing protection:
  - none
  - basic   (NEW)
  - strong  (NEW)
  - all

The differences are documented in the GCC online documentation:
    https://gcc.gnu.org/onlinedocs/gcc-4.9.2/gcc/Optimize-Options.html

Signed-off-by: Steven Noonan <steven@uplinklabs.net>
[yann.morin.1998@free.fr:
  - rebase
  - add legacy handling
  - SSP-strong depends on gcc >= 4.9
  - slightly simple ifeq-block in package/Makefile.in
  - keep the comment in the choice; add a comment shen strong is not
    available
  - drop the defaults (only keep the legacy)
  - update commit log
]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
[Thomas:
 - only show the choice if the toolchain has SSP support
 - add details for the BR2_SSP_ALL option that it has a significant
   performance impact.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ustr: not available for static-only configurations

ustr now unconditionally builds a shared library, so we have to mark
it as not available for static-only builds. The only reverse
dependencies of ustr is libsemanage, which was anyway already not
available for pure static builds.

Fixes:

  http://autobuild.buildroot.org/results/734/734975f6ac041e8aece07205c0b8966fe51e7df1/

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

stunnel: use BR2_TOOLCHAIN_HAS_SSP instead of BR2_ENABLE_SSP

All packages use BR2_TOOLCHAIN_HAS_SSP to determine whether they can
use SSP suppport or not, except stunnel. Therefore, this commit makes
stunnel consistent with other packages. This is also necessary in
preparation to the removal of BR2_ENABLE_SSP in favor of a choice
between various SSP options.

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

arch/arm: add help text to BR2_ARM_ENABLE_VFP

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

toolchain/external: Arago armv7 toolchain really requires a VFPv3

The Arago armv7 toolchain really requires a VFPv3 unit, so only expose
it to the user when the CPU actually has such a VFP unit

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

arch/arm: only expose VFP in FP strategy when the CPU *has* a VFP unit

There's no point in offering the user an option to select an FP strategy
when the CPU does not actually have a VFP unit.

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

arch/arm: only expose EABIhf when the CPU *has* a VFP unit

There's no point in offering the user an option to select EABIhf when
the CPU does not really have a VFP unit.

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

arch/arm: add option to enable an optional VFP unit

Currently, the VFP selection for ARM is a little bit muddy:
  - some CPUs definitely do not have a VFP or NEON,
  - some CPUs definitely do have a VFP or NEON,
  - some CPUs may have a VFP or NEON.

However, we currently conflate the availability of the VFP/NEON with the
possibility to use them. Even is the user chooses a floating point
strategy with a 'lower' solution (i.e. VFPv2 when a VFPv3 exists, or not
using NEON when the CPU has it), some packages are still using the
CPU-defined HW availaibility rather thean the usr's selection.

Furthermore, for CPU that may have a VFP/NEON, there is no way for the
user to actually specify that the HW is indeed available; the user can
only specify the floating point strategy. This means that some packages
or some package versions, like nodejs for example, can not be properly
selected on some CPU cores, like Cortex-A9 which only may have a VFP.

Like we have an option to enable an optional NEON unit, add a similar
option to enable an optional VFP unit.

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

arch/arm: reorder NEON option

Stating whether to use the NEON extensions when it is optional in the
CPU really is completing the definition of the CPU we've just selected.

Move the ENABLE_NEON option just after the choice of the CPU variant,
and before any "software" option (ABI/VFP).

This will make sense in a moment, when we introduce a similar option for
enabling an optional VFP unit.

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

vboot-utils: new package

Chromium OS verified boot utilities.

Needed for signing kernel images and manipulating bootable
partitions on media intended for Chromebooks.

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

strace: fix building for arc, metag, nios2, or1k and tile

Following commit broke building for arc, metag, nios2, or1k and tile:
http://sourceforge.net/p/strace/code/ci/34683e3926d8c2daa368afb805da422ee7043396/

Now we're fixing that breakage with a back-port of upstream fix:
http://sourceforge.net/p/strace/code/ci/dd1a80c8d213eed95fe55b7ebcb07ee165dd8e4b/

Mentioned fix will be a part of the next strace release,
i.e. this patch must be removed on strace version bump.

Fixes http://autobuild.buildroot.org/results/f25/f25c23003ed282b935f5289c944615860265364a
and alike, see http://autobuild.buildroot.org/?reason=strace-4.11

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

package/luv: bump to version 1.8.0

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

package/upmpcli: bump to version 0.13.1

upmpdcli version 0.13.1 requires libupnpp version 0.13.1 to build successfully.

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

package/libupnpp: bump to version 0.13.1

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

package/mpd: bump to version 0.19.12

Drop patch "Makefile.am: fix static build with alsa" - applied upstream.

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

package/luvi: bump to version 2.5.1

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

package/boost: bump to version 1.60.0

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

trousers: remove inline for read_data/write_data

read_data and write_data functions are declared with inline, but
their definition is not in a .h file. This is causing errors when
linking:

tcsps.c:(.text+0x4ac): undefined reference to `read_data'
tcsps.c:(.text+0x1888): undefined reference to `write_data'
...

The patch is a partial backport of upstream commit
3297fcdf5ac0df868778e976b8b59e35542c1ee2.

Fixes:
http://autobuild.buildroot.net/results/1a53d929bcec840cd18185e54b1a6da8aa849269/
http://autobuild.buildroot.net/results/fad66e88b5a97004b21658d413f4d3a6d0d8838b/
http://autobuild.buildroot.net/results/c7e23b8ec2c69798482868e18311ebf2c01b65e0/
...

Signed-off-by: Sergio Prado <sergio.prado@e-labworks.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

python-ipaddress: new package

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

package/nodejs: correct 0.10.x option name

The nodejs options incorrectly started BR2_BR2_xxx.  This change corrects
the name for 0.10.x releases to BR2_PACKAGE_NODEJS_0_10_X

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

libgdiplus: bump to version 4.2

Signed-off-by: Sergio Prado <sergio.prado@e-labworks.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/dhcp: remove sleep from SysV init scripts restart

No sleep is required for the restart and force-reload operations to
succeed.

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

package/dhcp: fix SysV init scripts output text

Fix various messages displayed by these scripts:
 - make start-stop-daemon quiet in order to avoid extra messages like
   "stopped /usr/sbin/dhcpd (pid 174)" being output between the command
   description and its result,
 - fix the script names in the usage strings.

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

package/dhcp/S80dhcp-server: allow empty INTERFACES

The dhcpd daemon does not require network interface names to be
specified on the command line.

>From dhcpd(8):
"The names of the network interfaces on which dhcpd should listen for
broadcasts may be specified on the command line.  This should be done
on systems where dhcpd is unable to identify non-broadcast interfaces,
but should not be required on other systems.  If no interface names
are specified on the command line dhcpd will identify all network
interfaces which are up, eliminating non-broadcast interfaces if
possible, and listen for DHCP broadcasts on each interface."

dhcpd exits with "Not configured to listen on any interfaces!" only if
no requested (those in INTERFACES, or all if empty) non-broadcast
interfaces matching the subnet declarations in dhcpd.conf are up.

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

package/dhcp: systemd: support extra options

Add an OPTIONS configuration variable in order to make it possible to
pass custom extra options to dhcpd. This keeps the systemd support
consistent with the SysV init script.

Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
Reviewed-by: "Maxime Hadjinlian" <maxime.hadjinlian@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/dhcp/S80dhcp-server: support extra options

Add an OPTIONS configuration variable in order to make it possible to
pass custom extra options to dhcpd.

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

package/dhcp: make S80dhcp-relay read a file from /etc/default/

The S80dhcp-relay init script has configuration variables like
INTERFACES whose contents have to be passed to the daemon. These
variables are initialized as empty strings, but some of them are not
allowed to be empty and there was no means of filling them apart from
creating a root FS overlay to overwrite these scripts.

This commit adds support for reading dhcrelay under /etc/default/ to
set these configuration variables.

[Thomas: adapt to patch only S80dhcp-relay, since S80dhcp-server has
already been changed by previous commits.]

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

dhcp: misc improvements to S80dhcp-server

This commit does misc improvements to the S80dhcp-server init script:

 - Use more variables: NAME, DAEMON, CFG_FILE

 - Read the configuration file in /etc/default/ in a more usual way
   (as done in S21rngd for example)

 - Remove leftover dhcpd3 string in the stopping action.

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

dhcp: read /etc/default/dhcpd, not /etc/default/dhcpd.conf

All daemons that read a file from /etc/default/ have it named just
after the name of daemon, without any extension. This commit fixes the
dhcp package to do the same.

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

libvips: fix compile error for sparc64

Fixes following build error:
http://autobuild.buildroot.net/results/1515f070c0fd410c681fc8306700244c34eb1635/

[Thomas:
 - simplify logic.
 - add reference to submitted gcc bug.]

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

doom-wad: bump to version 1.9

Also add a hash file.

The extraction method had to be changed, since the WAD file for this
version is not distributed as a stand-alone file, but comes inside a
2-part DOS self-extracting zip archive.

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

libnspr: fix build with the musl C library

Even though libnspr uses a configure script, it does look at __GLIBC__
and other macros to determine whether certain C library features are
available or not. Such things fails completely with the musl C
library, since it doesn't define __GLIBC__.

To fix this, we borrow some logic from Alpine Linux, which consists in
explicitly specifying which features are available in the musl C
library.

Fixes:

  http://autobuild.buildroot.org/results/ddd/dddd032232126f1e476e6aea2c6a32e9cd1ddfae/

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

autofs: really fix build

The autofs configure script does a number of weird things.

First, instead of using the default search path for AC_CHECK_PROGS(),
it overrides it its own search path, ignoring the PATH environment
variable. Due to this, autofs fails to find flex even though flex is
available in the PATH. Fortunately, the authors provided a --with-path
configure option which allows to override this search path.

In addition, the configure script is checking on the build machine
that a number of programs are available, while those programs are in
fact only needed on the target. The one causing build issues is
modprobe, because it may not necessarily be installed on build
machines, so we explicitly tell autofs that modprobe is located in
/sbin/modprobe. It might be necessary to provide additional values for
other programs, as we discover other problems.

Fixes:

  http://autobuild.buildroot.org/results/447/4479382b2d4d762991cab2e93fd08ad6160c2921/

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

tpm-tools: new package

[Thomas:
 - move from Libraries/Crypto to System tools in menuconfig
 - add patch to fix build with uClibc
 - use 'select' for openssl rather than a 'depends on'
 - propagate trousers dependencies
 - add <pkg>_LICENSE and <pkg>_LICENSEFILES
 - link against libiconv when needed, fixing !locale builds
 - explicitly disable pkcs11 support, since it is not handled by
   tpm-tools.mk currently.]

Signed-off-by: Al West <al.west@v-nova.com>
Signed-off-by: Noé Rubinstein <nrubinstein@aldebaran.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

trousers: new package

[Thomas:
 - use a 'select' instead of a 'depends on' for openssl
 - select libiconv when locale support is not available, otherwise the
   build fails due to missing iconv.h
 - disable on BR2_STATIC_LIBS configurations as it doesn't build. It
   could be made to build by patching the trousers Makefile.
 - add missing dependency on thread support in the toolchain
 - add Config.in comment about the thread and dynamic library
   dependencies.
 - add missing <pkg>_LICENSE and <pkg>_LICENSE_FILES variables.]

Signed-off-by: Al West <al.west@v-nova.com>
Signed-off-by: Noé Rubinstein <nrubinstein@aldebaran.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

doom-wad: add BR2_PACKAGE_CHOCOLATE_DOOM as alternate dependency

The shareware WAD file is also supported by Chocolate Doom.

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

chocolate-doom: new package

Chocolate Doom is a set of conservative source ports for Doom, Heretic,
Hexen and Strife, with a philosophy of preserving the look, feel, and
bugs of the vanilla versions of each.

http://www.chocolate-doom.org

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

Enabling mesa3d radeon driver for all architectures

Mesa32 radeon driver doesn't depend on x86 and can be built for all architectures.

Signed-off-by: Lada Trimasova <ltrimas@synopsys.com>
Cc: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/mesa3d: virgl driver needs egl & drm

Fixes
http://autobuild.buildroot.net/results/7b3/7b38131ebb6b20daec1a6d2c49b25c5300f319fe/
http://autobuild.buildroot.net/results/61b/61b8b87f0460751257f480ef478e753635cab798/
http://autobuild.buildroot.net/results/971/97154ea0e780623d23b20b6b9fcc5df6190875c9/
http://autobuild.buildroot.net/results/dcf/dcf4e31e6b6eeadc7a2d7c87697f57d699673b9e/
http://autobuild.buildroot.net/results/399/399cca80ada5798decc87bff69a3acc432e8f3d0/
http://autobuild.buildroot.net/results/60f/60f63958b2310e00937c498ddf2a0295ff77683a/

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

package/Config.in: correct capitalization of menu item

Signed-off-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

qt5/qt5webchannel: add qwebchannel.js installation in /var/www

To use webchannel in an application qwebchannel.js is needed but this file was not
installed.

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

package/libuv: fix support for uClibc-ng

See https://bugs.busybox.net/show_bug.cgi?id=8296 for further details on
the origins of this bug.

Submitted upstream to libuv and accepted, see
https://github.com/libuv/libuv/pull/653 and
https://github.com/libuv/libuv/commit/c861972

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

package/nodejs: fix support for uClibc-ng

Fix segmentation fault in node.js when using uClibc-ng.

This patch fixes bug #8296, see
https://bugs.busybox.net/show_bug.cgi?id=8296.

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

package/nodejs: fixed comment for mips soft float

nodejs is only selected for mips hard float however the comment is
displayed for hard and soft float.  Correct this by adding the missing
!BR2_MIPS_SOFT_FLOAT to the comment.

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

package/nodejs: removed version 0.12.9

Remove 0.12.9 to rationalise the number of nodejs releases supported by
buildroot.  Going forward buildroot will only support the latest release
of nodejs and the 0.10.x branch for armv5 support.

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

package/nodejs: bump version to 5.3.0

Replace 4.2.3 with 5.3.0.  5.3.0 is the current Stable release. See
https://nodejs.org/en/blog/release/v5.3.0 for details on the release.

Copied 4.2.3 patched to 5.3.0 with the following exceptions:
 - Removed 0004-fix-arm-vfpv2.patch, committed upstream and included in
   5.3.0.  See https://github.com/nodejs/node/commit/84dea1bd0c
 - Added 0004-Fix-va_list-not-declared.patch, fix for a new bug.  This
   has already been fixed upstream but is not in 5.3.0

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

vnstat: bump to version 1.15

Also:

  - Remove no longer needed patch (merged upstream).
  - Adapt vnstat.mk to the new autotools-based build system.
  - Include patch to improve gd library usability check + add
    configure option to explicitly disable that check.

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

package/unscd: add PID file to init script

unscd creates/manages it's own PID file. Pass PID file reference to
start-stop-daemon.

Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Doug Kehn <rdkehn@yahoo.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

jquery-sidebar: new package

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

Fix fallout after generated defconfig support

Fix improper use of qstrip; use correct variables.

Fixes #8546.

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

rfkill: new package

[Thomas: use "br" as the version suffix instead of "-br" so that the
final version string is "0.5-br" and not "0.5--br".]

Signed-off-by: Viacheslav Volkov <sv99@inbox.ru>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

jQuery: introduce external plugin menu

List all jQuery plugins in a dedicated menu and remove
uneeded jQuery selection.

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

boot/uboot: handle generated defconfigs

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

boot/barebox: handle generated defconfigs

Signed-off-by: Sam Bobroff <sam.bobroff@au1.ibm.com>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

boot/at91bootstrap3: handle generated defconfigs

Signed-off-by: Sam Bobroff <sam.bobroff@au1.ibm.com>
[yann.morin.1998@free.fr: move the kconfig-package hunk to the
 corresponding patch]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/linux: handle generated defconfigs

Signed-off-by: Sam Bobroff <sam.bobroff@au1.ibm.com>
[yann.morin.1998@free.fr: move the kconfig-package hunk to the
 corresponding patch]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

docs/manual: document the new kconfig-package variable

The previous patch introduced the new FOO_KCONFIG_DEFCONFIG variable
to specify a defconfig rule rather than a (def)config file.

Add this to the manual.

Also document the pre-existing FOO_KCONFIG_FILE for which the
explanations were missing altogether.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Sam Bobroff <sam.bobroff@au1.ibm.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

core/pkg-kconfig: handle generated defconfigs

As of Linux kernel commit ea4d1a8 "powerpc/configs: Replace
pseries_le_defconfig with a Makefile target using merge_config" some
kconfig defconfigs (one so far: "pseries_le_defconfig") can be
generated using "make <defconfig>" and they do not exist on disk as a
single defconfig file.

This causes buildroot's build to fail for those configs with:
'arch/powerpc/configs/pseries_le_defconfig' for 'linux' does not exist

To handle this case and keep the makefile steps as simple as possible,
introduce a new package variable, *_KCONFIG_DEFCONFIG, that can be
used to indicate that a defconfig rule is being used, rather than a
file.

This allows the rule that generates the .config file to use either the
provided file (by copying) or a generated defconfig (by running "make
<defconfig>") as its starting point. merge_config.sh can then be run the
same way in either case, using .config as both input and output.

Note that merge_config.sh is now modifying .config in-place but this
is safe because it uses a temporary copy while making changes.

This patch introduces the new variable but does not make use of it.
Use of the new variable will be introduced in separate patches.

Signed-off-by: Sam Bobroff <sam.bobroff@au1.ibm.com>
[yann.morin.1998@free.fr:
  - include all the other kconfig-package hunks into this one patch
  - do not transform the 'echo "..."; exit 1' into $(error ...) calls
  - use a make $(if)-block instead of a shell if-block to copy the file
    or run make (like is done to check the kconfig snippets)
  - misc typoes and rephrasing in the commit log
  - do not force the _defconfig suffix in the infra  (Thomas)
]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

wine: handle all *-config scripts

We already had a patch to allow us to explicitly pass the location of
the sane-config script, but wine's configure script uses several other
*-config scripts. This commit extends the wine patch and wine.mk to
explicitly pss the location of all *-config scripts that may be used
by wine.

The wine patch has been submitted upstream.

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

ltrace: add patch to fix build on sparc

Due to a missing header inclusion, the build of ltrace of sparc fails
with the following error:

plt.c: In function 'sym2addr':
plt.c:33:12: error: dereferencing pointer to incomplete type
  return sym->enter_addr;
            ^
plt.c:34:1: warning: control reaches end of non-void function [-Wreturn-type]
 }

This commit fixes that by adding a patch to the ltrace code base.

Fixes:

  http://autobuild.buildroot.org/results/55c/55ce618dbe1b2867e5738d90afb2eb67c19200e4/

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

lockdev: fix build with the musl C library

When the musl C library is used, the __GNU_LIBRARY__ symbol is not
defined, so the lockdev code doesn't know what to do. To work around
this, we explicitly pass __GNU_LIBRARY__ when building with musl, this
musl provides what's needed for lockdev to build, as if we were
building against glibc.

Fixes:

  http://autobuild.buildroot.org/results/9db/9db4ab40955d3af0027e141245d73ee6c614fb1f/

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

autofs: add missing dependencies on flex and bison

Fixes:

  http://autobuild.buildroot.org/results/999/999d2d5fe07994642ab39b26d936c230d79947a2/

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

ubus: new package

Signed-off-by: Sergio Prado <sergio.prado@e-labworks.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/unscd: new package

A daemon which handles passwd, group and host lookups for running
programs and caches the results for the next query. This package is
intended to replace glibc nscd. Presently, glibc nscd is not installed
by Buildroot.

unscd depends on glibc because it relies on glibc function
__nss_disable_nscd.

nscd.conf is taken from glibc with unused configurations removed.

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

nfs-utils: install /etc/nfsmount.conf

Installs /etc/nfsmount.conf on the target, mainly to be able to configure the
the NFS version that mount.nfs uses by default. Without this ability you are
likely to encounter mount errors when automounting directories from NFS
version 3 servers.

Signed-off-by: Jonathan Ben Avraham <yba@tkos.co.il>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

autofs: new package

Tested on i.MX6 using NFS mounts only. Not tested with DAS mounts as
of this commit.

[Thomas:
 - remove the "# rpcbind" comments on the thread and mmu dependencies,
   since we don't select rpcbind. The thread and mmu dependencies are
   still needed though, since autofs uses pthread_*() functions and
   fork().
 - Remove "Requires a toolchain with native RPC." in the Config.in
   help text and instead add a proper Config.in comment about the
   thread and RPC dependency.
 - Add patch to fix building with uClibc, taken from OpenEmbedded.]

Signed-off-by: Jonathan Ben Avraham <yba@tkos.co.il>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/Config.in: fix alphabetical order

Signed-off-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

emlog: new package

[Thomas:
 - update to the latest github version, which has a number of
   fixes/improvements compared to the latest tag.
 - add a patch to fix the build of the kernel module with recent
   kernel versions (>= 3.19).
 - use $(INSTALL) -D with a full destination path.
 - rewrap Config.in help text, lines were too long.]

Signed-off-by: Joris Lijssens <joris.lijssens@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

strace: bump to version 4.11

Patch upstream so remove it.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

cantarell: bump to version 0.0.19

Also use CANTARELL_VERSION_MAJOR in SITE as reported by Baruch.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

rsync: security bump to version 3.1.2

Fixes:
Make sure that all transferred files use only path names from inside the
transfer.  This makes it impossible for a malicious sender to try to
make the receiver use an unsafe destination path for a transferred file,
such as a just-sent symlink.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Makefile: Add RANLIB to the list of critical variables to unexport

If RANLIB is set and we're trying to build binutils, binutils will pick
this up and potentially fail to build.

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

openbox: new package

Add support for the Openbox Windows Manager.

[Thomas:
  - don't mention MMU in the Config.in comment.
  - wchar was forgotten in the Config.in comment.
  - the dependency of the comment on thread should be "depends on
    !BR2_TOOLCHAIN_HAS_THREADS" so that the comment is shown when
    thread support is *not* available.
  - add missing select of BR2_PACKAGE_XLIB_LIBXFT and
    BR2_PACKAGE_XLIB_LIBXRENDER which are needed for pango to have xft
    support, otherwise the build fails with "configure: error: Package
    requirements (pango >= 1.8.0 pangoxft >= 1.8.0) were not met:"
  - remove useless empty newline in the hash file
  - add missing dependency on host-pkgconf, which is needed since
    the configure.ac uses PKG_CHECK_MODULES.]

Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ipmitool: Bump ipmitool to 1.8.15

New release fixes a number of bugs, including some needed for talking
to modern AMI BMC firmware as shipped on some POWER8 systems.

Signed-off-by: Stewart Smith <stewart@linux.vnet.ibm.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

directfb/legacy: add that it was removed

As pointed by Yann.

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

python-pysmb: new package

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

python-cherrypy: bump to 4.0.0

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

python-pyftpdlib: bump to 1.5.0

Change download location to pypi and add pypi MD5 checksum.

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

python-werkzeug: bump to 0.11.3

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

python-zope-interface: bump to 4.1.3

Add pypi MD5 checksum.

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

imagemagick: bump to version 6.9.2-10

Signed-off-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gst1-plugins-bad/Config.in: indentation cleanup

Signed-off-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

picocom: bump to version 2.1

Also remove no longer needed patch (merged upstream before the release
of v2.1).

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

trace-cmd: bump to version 2.6

Also remove no longer needed patch (fixed upstream).

Signed-off-by: Sergio Prado <sergio.prado@e-labworks.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

luvi: new package

Add package luvi version v2.3.5.

luvi extends LuaJIT with asynchronous I/O and several optional modules to run
Lua applications and build self-contained binaries.

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

toolchain-external: add support for musl toolchain on ARM EABIhf

Since a few releases, the pre-built musl external toolchain has added
an ARM EABIhf variant, built for ARMv5T. This commit allows this
additional external toolchain to be used.

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

gst1-plugins-bad: Add OpenAL dependency

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

package/v4l2grab: Update hash after latest version bump

Fixes
http://autobuild.buildroot.net/results/3aa/3aa8ac30de88c50539afaf7cd5ef10eb375c96f5/
http://autobuild.buildroot.net/results/927/92792f6d7aa2f9b2b8d55315369164faa42c38af/
http://autobuild.buildroot.net/results/09a/09a1931d9c117b678cc6ab9bd34f23aa84238d48/
http://autobuild.buildroot.net/results/37d/37dee4ce618c511af8bdf7876e6038e907aba7a6/
http://autobuild.buildroot.net/results/48d/48d1052240c6041db7ac7b82fafd8c02998becfc/

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

ifupdown: bump to version 0.8.2

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

start-stop-daemon: bump to version 1.18.3

Enable developer mode to skip the shared library check in configure and
avoid further patching.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

debianutils: bump to version 4.5.1

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libmicrohttpd: bump to version 0.9.48

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gnupg: bump to version 1.4.20

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

linux: don't build appended DTB image in place and support multiple images

Currently, the linux.mk logic for appended DTB image does the
appending of the DTB in place, directly at the end of the zImage using
a >> sign. This is incorrect because if you run "make linux-rebuild"
multiple times, you get the DTB appended over and over again to the
image.

Since keeping the 'zImage' or 'uImage' name for the appended DTB image
is not very clear, this commit moves to using the 'zImage.<dtb>' and
'uImage.<dtb>' format. This way, we can clearly distinguish the
original image from the appended one.

In addition, this naming scheme easily allows to generate *multiple*
appended DTB images: from one zImage, you can generate multiple
zImage.<dtb> for several DTBs, and then generate (if requested) the
corresponding uImage.<dtb>.

To achieve this, this commit:

 - Changes the definition of LINUX_APPENDED_DTB to iterate over
   $(KERNEL_DTS_NAME), and generate a zImage.<dtb> image for each of
   them.

 - Changes the addition of LINUX_APPENDED_DTB for appended uImage to
   also iterate over $(KERNEL_DTS_NAME).

 - Provide a different implementation of LINUX_INSTALL_IMAGE which
   installs all the appended DTB images (but not the bare image)

 - Remove the checks that verified that only one DT name is passed
   when appended DTB is used, since we now support generating multiple
   DT images.

Some of the tested configuration:

 - Normal uImage with several DTBs

   BR2_LINUX_KERNEL_DEFCONFIG="mvebu_v7"
   BR2_LINUX_KERNEL_UIMAGE=y
   BR2_LINUX_KERNEL_UIMAGE_LOADADDR="0x200000"
   BR2_LINUX_KERNEL_DTS_SUPPORT=y
   BR2_LINUX_KERNEL_INTREE_DTS_NAME="armada-xp-matrix armada-xp-gp armada-370-mirabox"

   Contents of output/images/:

   armada-370-mirabox.dtb  armada-xp-gp.dtb  armada-xp-matrix.dtb  uImage

 - Normal zImage with several DTBs

   BR2_LINUX_KERNEL_DEFCONFIG="mvebu_v7"
   BR2_LINUX_KERNEL_ZIMAGE=y
   BR2_LINUX_KERNEL_DTS_SUPPORT=y
   BR2_LINUX_KERNEL_INTREE_DTS_NAME="armada-xp-matrix armada-xp-gp armada-370-mirabox"

   Contents of output/images:

   armada-370-mirabox.dtb  armada-xp-gp.dtb  armada-xp-matrix.dtb  zImage

 - Appended uImage with several DTBs:

   BR2_LINUX_KERNEL_DEFCONFIG="mvebu_v7"
   BR2_LINUX_KERNEL_APPENDED_UIMAGE=y
   BR2_LINUX_KERNEL_UIMAGE_LOADADDR="0x200000"
   BR2_LINUX_KERNEL_INTREE_DTS_NAME="armada-xp-matrix armada-xp-gp armada-370-mirabox"

   Contents of output/images:

   uImage.armada-370-mirabox  uImage.armada-xp-gp  uImage.armada-xp-matrix

 - Appended zImage with several DTBs:

   BR2_LINUX_KERNEL_DEFCONFIG="mvebu_v7"
   BR2_LINUX_KERNEL_APPENDED_ZIMAGE=y
   BR2_LINUX_KERNEL_INTREE_DTS_NAME="armada-xp-matrix armada-xp-gp armada-370-mirabox"

   Contents of output/images:

   zImage.armada-370-mirabox  zImage.armada-xp-gp  zImage.armada-xp-matrix

In all configurations, the contents of output/target/boot/ was the
same if BR2_LINUX_KERNEL_INSTALL_TARGET=y.

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

linux: only install the DTBs when not in appended DTB mode

When you're using the "appended DTB" mode, the Device Tree blob gets
appended to your kernel image, so there is no point in installing both
the DTB and the kernel image to the images or target directories,
installing the kernel image itself is sufficient.

Therefore, this commit disables the definition of LINUX_INSTALL_DTB
when appended DTB is used.

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

linux: de-duplicate DTB and Linux image installation

Currently, the LINUX_INSTALL_DTB and LINUX_INSTALL_DTB_TARGET macros
are exactly the same, except for the target directory.

Similarly, LINUX_INSTALL_KERNEL_IMAGE_TO_TARGET and
LINUX_INSTALL_IMAGES_CMDS are copying the kernel image, just to a
different place (and with a different strategy).

As a preparation for future additions, this commit de-duplicate this
code:

 - LINUX_INSTALL_DTB becomes a make macro that takes one argument: the
   destination directory.

 - LINUX_INSTALL_IMAGE is a new make macro that also takes on
   argument: the destination directory.

Both macros are used by LINUX_INSTALL_KERNEL_IMAGE_TO_TARGET and
LINUX_INSTALL_IMAGES_CMDS to respectively install to the target
directory and the images directory.

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

package/wpa_supplicant: add config option to build wpa_client library

Version 2.5 of wpa_supplicant introduces the config option
CONFIG_BUILD_WPA_CLIENT_SO to enable building wpa_client as shared library.

We need to add a patch from upstream to fix a runtime error with
libwpa_client.so:
  Error relocating /usr/lib/libwpa_client.so: bin_clear_free: symbol not found

[Thomas: renumber patch.]

Cc: Martin Bark <martin@barkynet.com>
Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Reviewed-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/perl-db-file: Berkeley DB 1.85 compatibility API is not needed

This package is also compatible with Berkeley DB v2 or higher:
https://metacpan.org/source/PMQS/DB_File-1.835/Changes

1.50
    DB_File can now build with either DB 1.x or 2.x, but not both at
    the same time.

[Thomas: adjust Config.in help text accordingly.]

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

package/efl: rename libefl to efl

This allow to keep backward compatibility with older
defconfig files that used BR2_PACKAGE_EFL.

Don't add Config.in.legacy entry for libefl since it's
not part of any Buildroot release.

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

package/libefl: move to package directory

As discussed on the list move the last remaining package
in package/efl subdirectory to package/ directory.

- move the efl dependency to libefl.
- Set LIBEFL_VERSION with the version number directly.

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