barebox: bump to version 2015.10.0

Signed-off-by: Fabio Porcedda <fabio.porcedda@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

musl: bump to version 1.1.12

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

samba4: bump to version 4.3.1

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

gdb: enable host-gdb on aarch64

gdb has full support for aarch64 since version 7.6,
current minimum is 7.7.1.
Target build was re-enabled over a year ago, but host-gdb was not.

Related commits:
0e49733b306b353aad3a29a2c7a6faf25841ef0b gdb disabled for aarch64
4372449a4f8e16f763c7d5a0c125ca015a7cbef6 target-gdb re-enabled

Signed-off-by: Alex Suykov <alex.suykov@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

tstools: GitHub migration

tstools has migrated to GitHub, so let's use this opportuninty to rework
this package as well.

- Change site to GitHub.
- No releases of tags yet, so let's use the latest commit ID from the
  master branch as a version.
- There is currently no license file attributed to tstools, so state that
  in the tstools.mk file and add empty license file variable. This issue is
  being tracked upstream: https://github.com/kynesim/tstools/issues/32
- Wrap the help text in the Config.in to 72 characters, and also change
  the project URL to the GitHub one.
- Add a hash file.
- Rename the patch to use a four digit numbering (001 -> 0001).

[Peter: leave license as MPLv1.1, adjust comment to match]
Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

syslog-ng: Specify location of libesmtp

If --with-libesmtp is omitted a host tool is invoked to detect the
location of the library and headers, which is obviously the wrong thing
for a cross-compile environment. Explicitly add --with-libesmtp to
resolve this.

Fixes:
http://git.buildroot.net/buildroot/commit/?id=539acb7500588d0a277357cf831c9f25f770578d
http://git.buildroot.net/buildroot/commit/?id=01f41585a78406f6e830d5eaf4413a983812a299

Signed-off-by: Chris Packham <judge.packham@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

arch/Config.in.x86: drop trailing dot (.) from x1000 URL

And add an empty line below for clarity.

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

arch/x86: add support for Intel X1000

The Intel X1000 is the Pentium class microprocessor that ships with
Galileo Gen 1/2. This patch adds changes to arch and toolchain-wrapper
to omit the lock prefix for the X1000.

[Thomas: tweak commit log and Config.in help text.]

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

unionfs: bump to version 1.0

Switch to github homepage and site.
Drop patch, it's not required any more.

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

exfat-utils: bump to version 1.2.1

Switch to github homepage and site.
Package is now autotools-based.
libc patch no longer required.

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

exfat: bump to version 1.2.1

Switch to github homepage and site.
Package is now autotools-based.
musl tweak no longer required.

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

package/nodejs: add version 4.1.2

We add a new version, because it is not API-compatible with the previous
versions.

Also, nodejs-4.1.2 requires gcc >= 4.8.

Forward-port patches from 0.12.7:
  - 0001-Remove-dependency-on-Python-bz2-module.patch partially applied
    upstream;
  - 0002-gyp-force-link-command-to-use-CXX.patch slightly refreshed;
  - 0003-Use-a-python-variable-instead-of-hardcoding-Python.patch
    largely refreshed to address new occurences of hard-coded calls;
  - 0004-fix-build-error-without-OpenSSL-support.patch applied upstream
  - 0005-Fix-typo-for-arm-predefined-macro-in-atomicops_inte.patch
    applied upstream.

New patch:
  - 0004-fix-arm-vfpv2.patch to fix the gcc -mfpu option for VFPv2.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Jörg Krause <joerg.krause@embedded.rocks>
Cc: Martin Bark <martin@barkynet.com>
Cc: Jaap Crezee <jaap@jcz.nl>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/nodejs: fix architectural dependencies on ARM

On ARM, starting with v0.12.x, the V8 JS engine is now requires at least
an armv6 and at least a VFPv2.

Since we're about to introduce the v4.x version, which has the same
requirements, introduce an intermediate variable to hold that condition.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Jörg Krause <joerg.krause@embedded.rocks>
Cc: Martin Bark <martin@barkynet.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

linux-firmware: rtl81xx: also install rtl8188eufw firmware variant

As used by the rtl8188eu staging driver.

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

ntp: add ntptime option

Allow the `ntptime` utility to be included on a target.

[Peter: add comment why AUTORECONF is needed]
Signed-off-by: James Knight <james.knight@rockwellcollins.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

musl: Make only shared libraries conditional

External toolchain use requires a static libc (as buildroot uses
`gcc --print-file-name libc.a` to find the sysroot); thus, the static portion
of a musl build should not be conditional.

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

package/swupdate: add patch to fix link error for CONFIG_UBOOT and missing libz

CONFIG_UBOOT needs libz [1]. So CONFIG_UBOUT mustn't be available when
HAVE_ZLIB=n.

Add patch from upstream to fix a link error for CONFIG_UBOOT and missing libz.
Note, that this link error does not affect the default config provided by
Buildroot.

[1] https://github.com/sbabic/swupdate/commit/cf495e23ad117f4cae9c429e80714fe9c18f2d6a

[Thomas: use a single <pkg>_PATCH variable.]

Reported-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/swupdate: add patch to fix build issue with some toolchains

Some toolchains used in Buildroot (x86-64 and mips64) have a problem to compile
a set of object files into one object file as it is done for the
`builtin-target` declared in swupdate's Kbuild Makefile.build. This target
collects all object files from every subdirectory declared in `objs-dir` and
compiles them into a single object file named `build-in.o` using the linker ld.

For the addressed toolchains the default emulation of ld is not the correct one
leading to different kind of relocation errors:

  /usr/bin/i686-pc-linux-gnu-ld: Relocatable linking with relocations from
  format elf64-x86-64 (core/swupdate.o) to format elf32-i386 (core/built-in.o)
  is not supported.

or

  /usr/bin/mips-linux-gnu-ld: handlers/raw_handler.o: endianness incompatible
  with that of the selected emulation

or

  /usr/bin/mips64el-linux-ld: core/swupdate.o: ABI is incompatible with that
  of the selected

As there is no need to have single object file `built-in.o` to be compiled for
the subdirectories core and handlers we can easily allow swupdate to be built
with these toolchains by compiling the source files from core and handlers with
the `lib-target` target by assigning them to the `libs-y` variable as it is done
for all other subdirectories. The `lib-target` compiles a set of object files
into one archive file. With this target we avoid using the ld linker and use the
ar archiver instead.

Add a patch from upstream to fix a whole bunch of autobuild errors.

Fixes:
http://autobuild.buildroot.net/results/de9/de920298075d32f3de83a0cfb7417846eb833425/
http://autobuild.buildroot.net/results/975/975915aa33005806e78260bae385cd4b3e359ca8/
http://autobuild.buildroot.net/results/c54/c54e7a2ea353d95d41a1e966de0dffa7b6ac432e/
http://autobuild.buildroot.net/results/019/0198596ebfed05ab86552b628da2274d55bf42ae/

and many more.

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

package/swupdate: add hash file

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

feh: bump to version 2.14

Signed-off-by: James Knight <james.knight@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

xdriver_xf86-video-intel: add upstream patch fixing libdrm/kernel crash

When running this Intel driver with a newer libdrm/kernel (4.x), Xorg
can crash (segmentation fault).

References:
 http://cgit.freedesktop.org/xorg/driver/xf86-video-intel/commit/?id=7fe2b2948652443ff43d907855bd7a051d54d309
 https://bugs.gentoo.org/show_bug.cgi?id=552124
 https://bugzilla.opensuse.org/show_bug.cgi?id=931917

[Thomas:
  - add upstream link to the git commits in the patches themselves
  - add James SoB line inside the patches themselves]

Signed-off-by: James Knight <james.knight@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

xdriver_xf86-video-intel: add upstream patches fixing fstat build errors

Attempting to build xdriver_xf86-video-intel version 2.99.917 with udev
results in compilation failures. The provided patches ([1], [2])
corrects the `implicit declaration of function 'fstat'` occurrences
during the building of this driver.

[1] http://cgit.freedesktop.org/xorg/driver/xf86-video-intel/commit/?id=12af8a575d1518d40416f83195049157c3a062a5
[2] http://cgit.freedesktop.org/xorg/driver/xf86-video-intel/commit/?id=369ceec0e4910ba2c37736a59c55c0d6c26433bf

[Thomas:
 - add upstream link to the git commits in the patches themselves
 - add James SoB line inside the patches themselves
 - add a comment above AUTORECONF = YES to indicate why it was added]

Signed-off-by: James Knight <james.knight@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

nitrogen6*_defconfig: bump u-boot and kernel revisions

This update brings several improvements such as:
- switch to a generic Boundary defconfig
  - same across all the Boundary Devices platforms
- support of HDMI to MIPI daugther board
https://boundarydevices.com/product/bd_hdmi_mipi/
- enhanced display support
  - LCD ASIT500MA6F5D
  - LCD A030JN01
  - HDMI 480p

For a complete list of changes:
https://github.com/boundarydevices/linux-imx6/commits/d7a412b
https://github.com/boundarydevices/u-boot-imx6/commits/a4070a8

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

nitrogen6*_defconfig: switch to zImage

As now using zImage/bootz in the 6x_bootscript.

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

board/boundarydevices: update 6x_bootscript to use zImage

As suggested by Thomas:
http://lists.busybox.net/pipermail/buildroot/2015-September/139884.html

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

gvfs: bump to version 1.26.1.1

Lots of changes, add support for libnfs, udev, libbluray, cleanup afp
support, gtk3, and systemd.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libnfs: bump to version 1.9.8

Prerequisite for the gvfs 1.26.1.1 bump.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Use portable shebangs (/bin/bash -> /usr/bin/env bash)

"/usr/bin/env bash" is used everywhere else in host tools. These are the
only remaining offenders.

Signed-off-by: Bjørn Forsman <bjorn.forsman@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

dhcp: bump to version 4.1-ESV-R12

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

sqlite: bump to version 3.9.1

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

shared-mime-info: bump to version 1.5

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

iptables: add upstream patch fixing the message 'connlabel.conf not found'

In iptables versions 1.4.20 and 1.4.21, any call to iptables
would generate a message

    cannot open connlabel.conf, not registering 'connlabel' match: No
        such file or directory

This problem was reported at [1] and subsequently fixed with commit [2], but
no upstream release has been made since. This patch imports the fix into
Buildroot, awaiting a new release of iptables.

[1] http://marc.info/?l=netfilter-devel&m=140990442432002&w=2
[2] http://git.netfilter.org/iptables/commit/?id=825fbda5482a7d5ec5a6619c81fe07ff865c7d6e

Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

less: bump to version 481

Switch site to gnu mirror since the stable releases are mirrored there.
Also use gpg sig to calculate hash, from the gnu mirror.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Config.in.legacy: remove empty line

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

uclibc: clean-up test suite build command

Starting from uClibc 0.9.32 "test_compile" target builds tests flawlessly.
For 0.9.31 it lead to build failure, but now when AVR32 support was dropped
I think this change worth applying.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Gustavo Zacarias <gustavo@zacarias.com.ar>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Thomas De Schampheleire <patrickdepinguin@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/shairport-sync: bump to version 2.6

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

Revert "manual: Add notes about GitHub and hashes"

This note is no longer needed, so remove it.

Tarballs generated by GitHub are always the same, so hashes can be
checked.

This reverts commit 9474cc2594c8643301164500ce6eb62546da2b8f.

[Thomas: fix other remaining references to not using hashes for
github, as noticed by Maxime Hadjinlian.]

Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

system/skeleton: finalize removal of ftp user

In commit 3dde19e5f32c58ffbf7e190257b073e91e0a7e8d, the ftp user was
removed from /etc/passwd, /etc/group and /home in the skeleton, but
the corresponding entry was not removed from /etc/shadow. This commit
fixes that.

Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Cc: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

support/mkusers: allow comments in users tables

The format of the users table files is non trivial, so it is sometimes
handy to add comments explaining the syntax (or simply the reason for
the user) inline in the files.

Ignore empty lines and comment lines prefixed with '#' similar to shell
or makedevs files.

Packages that defined no user (the vast majority) would cause an empty
line to be present in the internal users table, hence the reason we
skipped empty usernames. Now that we ignore empty lines, we no longer
need to check for empty usernames.

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

gst1-plugins-base: update configure options

- always disable new gio_unix_2_0 option (only used for testing)

[Thomas: add comment in .mk file.]

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

syslog-ng: New package

syslog-ng is an enhanced log daemon, supporting a wide range of input
and output methods: syslog, unstructured text, queueing, SQL & NoSQL.

[Thomas:
 - Rewrap Config.in.help text
 - Pass --pidfile option when starting syslog-ng so that its PID file
   is created in /var/run/syslog-ng.pid, which allows
   start-stop-daemon to actually stop syslog-ng. Without this,
   S01logging was not able to stop syslog-ng.
 - Pass the executable path at stop time in S01logging, so that
   start-stop-daemon can check we're not incorrectly stopping
   something completely different.
 - Add busybox as a dependency of syslog-ng if busybox is enabled,
   since we want to override Busybox's S01logging init script.
 - Simplify the python condition, since python and python3 are
   mutually exclusive.
 - Rewrap the comment above SYSLOG_NG_FIXUP_CONFIG.]

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

dvblast: new package

DVBlast is a simple and powerful MPEG-2/TS demux and streaming
application.

[Thomas:
 - add missing hash file.
 - rewrap Config.in help text
 - remove unneeded "DVBLAST_INSTALL_TARGET = YES", since this is the
   default behavior
 - pass TARGET_MAKE_ENV in the environment when calling make.
 - use TARGET_CONFIGURE_OPTS instead of just CC/LD.]

Signed-off-by: Julian Scheel <julian@jusst.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

bitstream: new package

BiTStream is a set of headers to ease processing of mpeg ts
streams. It is used by dvblast.

[Thomas:
 - add missing hash file.
 - add missing BITSTREAM_LICENSE_FILES information.
 - rewrap Config.in help text.
 - use lowercase name for the package in the comment header in
   bitstream.mk.]

Signed-off-by: Julian Scheel <julian@jusst.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

inadyn: fix init script and default config file

This commit does a number of fixes to the inadyn package to make it
work properly "out of the box":

 * inadyn is installed in /usr/sbin, not /usr/bin, so we fix the path
   in the init script

 * Use "printf" for the Starting and Stopping messages, so that the OK
   / FAIL stay on the same line.

 * Pass the -q option to the start sequence, since it's passed in the
   stop sequence.

 * Fix the configuration file to use an existing dyndns_system and
   avoid a failure at startup.

 * Use a variable called ENABLED in /etc/default/inadyn to decide
   whether to start the service or not. By default, it is not started,
   as suggested by Gustavo, and an explicit ENABLED="yes" is needed.

 * Store the PID file in /var/run/inadyn.pid, like we do for all other
   daemons.

[Thomas: drop silly comment about uncommenting the alias line in the
configuration file, which is already uncommented. Noticed by Yann.]

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

package/ffmpeg: Add support for GnuTLS

Besides OpenSSL ffmpeg also supports GnuTLS.

If both GnuTLS and OpenSSL are enabled, the former takes precedence over
the latter:
http://git.videolan.org/?p=ffmpeg.git;a=blob;f=configure;h=7f9fed31a491ed0991aee2d354dba82e54e3291e;hb=refs/heads/release/2.8#l2719

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

gcc: simplify musl patches for SSP support

Now that we are always explicitly passing gcc_cv_libc_provides_ssp,
there is no longer any reason to modify the gcc configure/configure.ac
to take into account the musl case. When a musl toolchain is being
built, BR2_TOOLCHAIN_HAS_SSP is always 'y', and therefore
gcc_cv_libc_provides_ssp=yes is always passed when building
gcc-initial and gcc-final.

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

gcc: pass explicit gcc_cv_libc_provides_ssp also to gcc-final

During the gcc-initial build, we already pass
gcc_cv_libc_provides_ssp=yes explicitly when SSP support will be
available in the C library: at this point in time the C library is not
yet built, so gcc cannot detect if it will support SSP or not.

However, it turns out that there are some situations for which it is
also useful to tell gcc explicitly whether the SSP support is
available or not: the gcc logic to decide whether uClibc has SSP
support or not is broken since uClibc-ng bumped the glibc version it
pretends to be.

So, this commit makes sure that we explicitly pass
gcc_cv_libc_provides_ssp both to gcc-initial and gcc-final, and that
we're always passing either 'yes' or 'no'.

Fixes:

   http://autobuild.buildroot.org/results/778/778e6309ba834cc70f8243a4f6c664c0bcaeb7c5/

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

toolchain: like glibc, musl always provides SSP support

Make sure BR2_TOOLCHAIN_USES_MUSL selects BR2_TOOLCHAIN_HAS_SSP since
musl always provides SSP support (like glibc).

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

busybox: fix musl build issues

Musl doesn't implemented wmtp, and its utmp support is not sufficient
to build the latest versions of Busybox, so let's disable wtmp and
utmp support in Busybox for musl toolchains.

We re-use an existing define BUSYBOX_INTERNAL_SHADOW_PASSWORDS that
was used to do some musl-related tweaks to the Busybox configuration
around shadow password functions, and extend it to also do the
utmp/wtmp tweaks. In the process, we rename the define to
BUSYBOX_MUSL_TWEAKS.

Note that BUSYBOX_KCONFIG_FIXUP_CMDS continues to call
BUSYBOX_INTERNAL_SHADOW_PASSWORDS because there is another definition
of this macro under other conditions.

Fixes:

  http://autobuild.buildroot.org/results/b96/b96ffad13bca5e6fbf7ce79538aacbfe27b1eff3/

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

qt5base: Fix compile errors in qurl.h / qeglplatformscreen.cpp

Update the already existing fix for EGL/X11 header issue to fix
an additional problem encountered on my system where I had
compile errors in qeglplatformscreen.cpp. The problem was related
to the wrong order of includes. The X11 headers must always be
included last, as indicated in
http://lists.qt-project.org/pipermail/development/2013-March/010511.html

The fix is done in the existing 0003-xcb-egl-fixes.patch patch, since
it is an additional fix for the same problem.

[Thomas: tweak commit log, and adjust SoB details as suggested by
Arnout.]

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

ffmpeg: let the user override the --cpu value

Grab the content of BR2_PACKAGE_FFMPEG_EXTRACONF after guessing the CPU,
so the user will still have the possibility to override it.

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Reviewed-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ccache: allow for BR2_CCACHE_DIR environment override

Allow the BR2_CCACHE_DIR .config option to be overriden by the
BR2_CCACHE_DIR env variable.
This is useful for big projects where in some cases the developers home
directory might be a NFS mount (slow) and real production builds aren't.

Update documentation accordingly as well.

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

ofono: fix missing systemd and dbus files

Ofono's service file was not being installed when systemd was selected
as the init system.

Without ofono.conf in /etc/dbus-1/system.d, ofonod cannot take its
dbus name on the system bus and will fail to start with a permission
denied error.

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

connman: fix missing dbus configuration files

Without connman.conf in /etc/dbus-1/system.d/, connmand cannot take
its name on the dbus system bus and will fail to start with a
permission denied error.

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

package/ffmpeg: Add optional dependencies

Disable libcdio and libilbc since they are incompatible with ffmpeg.

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

package/wavpack: Fix iconv support

wavpack fails to link with libiconv

/bin/bash ../libtool  --tag=CC   --mode=link /home/bernd/buildroot/br3/output/host/usr/bin/i586-buildroot-linux-uclibc-gcc  -I../include -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64   -Os    -o wvunpack wvunpack-wvunpack.o wvunpack-utils.o wvunpack-md5.o  ../src/.libs/libwavpack.la -lm
libtool: link: /home/bernd/buildroot/br3/output/host/usr/bin/i586-buildroot-linux-uclibc-gcc -I../include -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -Os -o wvunpack wvunpack-wvunpack.o wvunpack-utils.o wvunpack-md5.o  ../src/.libs/libwavpack.so -lm -Wl,-rpath -Wl,/home/bernd/buildroot/br3/output/build/wavpack-4.75.2/src/.libs
/home/bernd/buildroot/br3/output/host/usr/lib/gcc/i586-buildroot-linux-uclibc/4.9.3/../../../../i586-buildroot-linux-uclibc/bin/ld: wvunpack-wvunpack.o: undefined reference to symbol 'libiconv_open'

using this defconfig

BR2_TOOLCHAIN_BUILDROOT_WCHAR=y
BR2_PACKAGE_WAVPACK=y

Patch suggested by Peter:
http://lists.busybox.net/pipermail/buildroot/2014-February/090596.html

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

package/rtmpdump: Fix makefile when GnuTLS is used

In librtmp/Makefile the variable CRYPTO defaults to OpenSSL. When
building with GnuTLS support instead, librtmp.pc includes wrong
libraries:

$ grep Req output/staging/usr/lib/pkgconfig/librtmp.pc
Requires: libssl,libcrypto

using this defconfig

BR2_TOOLCHAIN_BUILDROOT_WCHAR=y
BR2_PACKAGE_GNUTLS=y
BR2_PACKAGE_RTMPDUMP=y

Pass RTMPDUMP_CRYPTO also to INSTALL_*_CMDS to fix this problem.
Patch supposed by Thomas:
http://article.gmane.org/gmane.comp.lib.uclibc.buildroot/129525

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

package/rtmpdump: Remove optional dependency to PolarSSL

The makefile of rtmpdump has no support for PolarSSL, so lets remove
support for it, as supposed by Gustavo:
http://article.gmane.org/gmane.comp.lib.uclibc.buildroot/129531

Fixes
http://autobuild.buildroot.net/results/45d/45dd9adbac449ce0ed66af5b4655b4d9b37faa62/
http://autobuild.buildroot.net/results/ae0/ae0c4bab7975ed2ad77a9f9fd6a300d1327d56b9/
http://autobuild.buildroot.net/results/029/029695a022ba12ff177521ebc3334d7ac2373ddc/

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

systemd: bump to 227

Added back the hash as it was discussed during the last Buildroot Dev
Days.

Signed-off-by Maxime Hadjinlian <maxime.hadjinlian@gmail.com>

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

toolchain-external/CodeSourcery MIPS: available only for R2

Currently the CodeSourcery toolchains for MIPS can be selected to build
mips32 (revision level 1) targets, but the resulting binaries are built
for mips32r2 instead. This is because these toolchains don't have
library support other than mips32r2, so there is no point to allow the
selection of a mips32 variant with a CodeSourcery MIPS toolchain, since
everything will be built for mips32r2 instead.

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

package/shairport-sync: bump to version 2.4.2

Also add a hash file - github release tarballs are considered stable now.

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

binutils: fix building perf on ARC

It turned out one of the previous fixes (required to build Linux
kernel)
------------------>8-------------------
ARC Binutils: https://github.com/foss-for-synopsys-dwc-arc-processors/binutils-gdb/commit/a65b844aed9153789356e098984452df2f5d9058
Buildroot: http://git.buildroot.net/buildroot/commit/?id=2d4e2e238a0ea9395152ae71d882d79b1f35094c
------------------>8-------------------
broke building of some other software packages.

In particular perf built for ARCv2 had corrupted .plt entries that lead
to immediate crash on perf execution.

That's an example of normal .plt entries:
------------------>8-------------------
< 1-st PLT entry >:
   12c24:    30 27 8c 7f 0d 00 74 95    ld         r12,[pcl,0x000d9574]
   12c2c:    21 20 00 03                j.d        [r12]
   12c30:    0a 24 c0 1f                mov        r12,pcl

< 2-nd PLT entry >:
   12c34:    30 27 8c 7f 0d 00 68 95    ld         r12,[pcl,0x000d9568]
   12c3c:    21 20 00 03                j.d        [r12]
   12c40:    0a 24 c0 1f                mov        r12,pcl
------------------>8-------------------
Note right after jump in its delay-slot r12 gets set with current value
of program counter. This is required for the first symbol resolution,
see implementation of _dl_linux_resolve here:
http://git.uclibc.org/uClibc/tree/ldso/ldso/arc/resolve.S#n46

And that's what we got in .plt after mentioned fixes:
------------------>8-------------------
< 1-st PLT entry>:
   13384:    30 27 8c 7f 0f 00 84 75     ld         r12,[pcl,0x000f7584]
   1338c:    21 20 00 03                 j.d        [r12]

< 2-nd PLT entry>:
   13390:    30 27 8c 7f 0f 00 78 75     ld         r12,[pcl,0x000f7578]
   13398:    21 20 00 03                 j.d        [r12]
------------------>8-------------------

Note r12 setup is missing.

That happened because linker thought the size of PLT entry is 12 bytes
(which is exactly the size of PLT entry for ARCv1, read ARC750/770)
while for ARCv2 PLT entry is 16-bytes long. And erroneously trailing 4
bytes were truncated.

Current commit fixes this misbehavior and PLT gets generated properly
again.

Now we have a fix for that issue, see
https://github.com/foss-for-synopsys-dwc-arc-processors/binutils-gdb/commit/5df50c61083165455aae5504c5c3566fa5ccebb1

This fix is in arc-2.23-dev branch and will be a part of the next
release of ARC tools, so then this patch must be removed from buildroot.

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

docs/manual: add section about size graphing

Document the new graph-size target and its possibilities.

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

Makefile: implement a graph-size target

This commit implements a graph-size target that calls the script of
the same name to generate the graph and CSV files related to package
and file sizes.

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

support/scripts: add size-stats script

This new script uses the data collected by the step_pkg_size
instrumentation hook to generate a pie chart of the size contribution
of each package to the target root filesystem, and two CSV files with
statistics about the package size and file size. To achieve this, it
looks at each file in $(TARGET_DIR), and using the
packages-file-list.txt information collected by the step_pkg_size
hook, it determines to which package the file belongs. It is therefore
able to give the size installed by each package.

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

pkg-generic: add step_pkg_size global instrumentation hook

This patch adds a global instrumentation hook that collects the list
of files installed in $(TARGET_DIR) by each package, and stores this
list into a file called $(BUILD_DIR)/packages-file-list.txt. It can
later be used to determine the size contribution of each package to
the target root filesystem.

Note that in order to detect if a file installed by one package is
later overriden by another package, we calculate the md5 of installed
files and compare them at each installation of a new package.

Collecting the list of files installed by each package is done
unconditionally, as tests have shown that the performance impact of
doing this is negligible.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Ryan Barnett <ryanbarnett3@gmail.com>
Tested-by: Ryan Barnett <ryanbarnett3@gmail.com>
Tested-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libv4l: fix missing libintl linking

Add missing libintl linking to utils/dvb and utils/keytable (patch
Makefile.am and do autoreconf with host-gettext dependency).

Patch submitted upstream (see [2]).

Add proper gettext dependency for v4l-utils option.

Fixes [1]:
 dvb-format-convert.o: In function `parse_opt':
 dvb-format-convert.c:(.text+0x46): undefined reference to `libintl_gettext'

[1] http://autobuild.buildroot.net/results/40a/40a12c0d1746dcc51e4f221eefbbf31bd427cac8
[2] http://permalink.gmane.org/gmane.linux.drivers.video-input-infrastructure/95883

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

libfribidi: indentation cleanup

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

package/jq: bump to version 1.5

Remove patch applied upstream. Add hash file.

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/fdk-aac: bump to version 0.1.4

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/wavpack: bump to version 4.75.2

Autoreconf is not necessary anymore.

WavPack "autodetects" CPU type to enable ASM code. However, the assembly code
for ARM is written for ARMv7 only and building WavPack for an ARM-non-v7
architecture will fail. We explicitly enable ASM for the supported
architectures x86, x64 and ARMv7 and disable it for all others.

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

toolchain-external: bypass buildroot wrapper

The buildroot internal toolchain now adds a wrapper. When we use a
buildroot toolchain as an external toolchain, we want to bypass this
wrapper and call the compiler directly, for two reasons:

1. The options added by the wrapper are not necessarily appropriate
   when it is reused as an external toolchain. For instance, ccache
   may have been enabled while building the toolchain but not when
   using it as an external toolchain.

2. Currently, the wrapper expects to reside in .../usr/bin, but when
   used as an external toolchain it will be in .../ext-toolchain/bin.
   Therefore, the wrapper can't find the real binary and sysroot
   anymore.

To bypass the wrapper, we check for the existence of *.br_real files in
the external toolchain directory. If any such file exists, the wrapper
will add the .br_real suffix for all the wrapped files. Note that the
wrapper doesn't check if the *.br_real exists for each individual
wrapped file, it just assumes that all wrapped files have a
corresponding .br_real. This is currently true but that may change in
the future of course.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

gcc: use '.br_real' instead of '.real' suffix for the raw internal toolchain

If an externally built (non-Buildroot) toolchain also wraps the toolchain
executables, there is a risk that it will also use the '.real' extension.
To minimise this risk, use a more buildroot-specific extension instead:
'.br_real', so we can detect that the external toolchain is built using
Buildroot and get to the raw toolchain binaries.

[Peter: reword description]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/intltool: bump to version 0.51.0

Also add a stronger, locally calculated sha256 checksum.

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

vala: bump to version 0.30.0

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Reviewed-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Tested-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

harfbuzz: bump to version 1.0.6

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Reviewed-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Tested-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

polarssl: security bump to version 1.2.17

Fixes:
CVE-2015-5291 - Remote attack on clients using session tickets or SNI

Also includes countermeasures against Lenstra's RSA-CRT attach for
PKCS#1 v1.5 signatures (1.2.16) and the Logjam attack (1.2.15).

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Reviewed-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Tested-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

fping: bump to version 3.12

Fixes regression with fping6 -R.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

redis: bump to version 3.0.5

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

dillo: bump version to 3.0.5

And remove upstream applied patch.

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

evemu: bump version to 2.3.1

See [1] for details.

[1] http://lists.freedesktop.org/archives/input-tools/2015-October/001221.html

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

package/initscripts: S40network: wait for network interfaces to appear

On some machines, the network interface is slow to appear. For example,
on the Raspberry Pi, the network interface eth0 is an ethernet-over-USB,
and our standard boot process is too fast, so our network startup script
is called before the USB bus is compeltely enumerated, thus it can't
configure eth0.

Closes #8116.

[Peter: move to S40network, handle multiple interfaces]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

bridge-utils: add upstream patch fixing sysfs writes

Add upstream patch that fixes a sysfs path in br_set.
The patch is on top of the latest release 1.5, which already dates from
2011.

Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

dependencies.sh: improve the missing perl modules detection

[Thomas:
 - Check for Thread::Queue, not Thread:Queue.
 - Use 'printf' instead of 'echo -e', since printf is POSIX, but not
   'echo -e'.]

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

sconeserver: use direct imagemagick disable

Disabling sconesite image isn't enough to avoid imagemagick
autodetection, and if it's present in the host (distro) but not in the
target it gets automatically picked up. Fixes:
http://autobuild.buildroot.net/results/377/37705926f3023395c54af0532ff95f0125fbaa66/

Also add a comment about autoreconf.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

dropbear: bump to 2015.68

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Reviewed-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Tested-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

fs/axfs: fix URL indentation

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

mplayer: enable apng and tdsc only if zlib is available

apng and tdsc are enabled by default, but they need zlib, otherwise the
compilation will fail with an error like this one:

libavcodec/pngdec.c:35:18: fatal error: zlib.h: No such file or
directory
 #include <zlib.h>
                  ^
compilation terminated.

So add an autodep on BR2_PACKAGE_ZLIB to enable or disable support for
apng and tdsc accordingly.

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

wireshark: security bump to version 1.12.8

Fixes:
CVE-2015-7830 - Pcapng file parser crash.

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

sqlcipher: disable tcl support

In commit b1dcb1c2 a fix was attempted at fixing a build failure where
sqlcipher finds the distribution tcl and tries to build against it.
sqlcipher has two tcl usage scenarios: host-tcl which is required to
actually build it, and target tcl to build bindings.
Since cross-compilation of the bindings isn't clean it fails as well,
and since it wasn't a feature before either just disable it completely,
since the wrong fix makes the sqlcipher build failures worse.
Fixes:
http://autobuild.buildroot.net/results/fc0/fc0528de7fed2f34f503fe8dd8b4c21ded0d6dff/

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

configs: update cubieboard_defconfig to a 3.4 kernel

The current commit used for the kernel Git repository for Cubieboard,
274a66a7bfcbaabb88d63e4eba161965383cc416, actually points to 3.0
kernel. This is incorrect since the kernel headers version specified
by the defconfig is 3.4. And anyway, the 3.4 kernel is kind of the
official vendor kernel for Allwinner platforms.

mfld.fr@gmail.com reported in bug #7931 that commit
9a1cd034181af628d4145202289e1993c1687db6 was working for him. It is
the latest commit in the sunxi-3.4 branch. So we switch to using this
commit, which was build tested successfully.

While we're at it, remove the definition of BR2_LINUX_KERNEL_VERSION
from the defconfig: it doesn't make sense to have it defined in a
defconfig, as it's a blind option that gets defined from the value of
BR2_LINUX_KERNEL_CUSTOM_REPO_VERSION when a Git repository is used to
source the kernel.

Fixes #7931.

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

busybox: enable fully featured hush shell for nommu

As pointed by Rich Fekler on IRC the basic hush shell is pretty much
useless since it doesn't support if conditionals, loops, functions, case
or even interactive mode.
So enable the full feature-set.
Size delta: +10184 bytes uncompressed for blackfin fdpic.

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

sane-backends: specify explicit path to net-snmp-config

Otherwise if the host distro has net-snmp devel files sane-backends will
find /usr/bin/net-snmp-config first and the build will break. Fixes:
http://autobuild.buildroot.net/results/c38/c381c316907964787ebe1a0807399022f15d8158/

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

pango: remove initscript

pango-querymodules is deprecated/removed, so ditch the initscript.

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

musl: honor static vs. shared library options

[Thomas: use SHARED_STATIC_LIBS_OPTS to simplify the logic.]

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

guile: handle readline support explicitly

Otherwise the host readline (+ncurses) development files can poison the
build causing breakage. Fixes:
http://autobuild.buildroot.net/results/d0a/d0a8ad9d3f19592e06df36408d6af745f7908165/

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

sqlcipher: specify tclConfig.sh directory

Otherwise it will search around, and if the host has it then it breaks.
Fixes:
http://autobuild.buildroot.net/results/057/0572a9a1f69ee214efb10863f5d2852b25bf4f54/

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

slang: forcibly disable oniguruma support

If the host has oniguruma installed then slang can detect it and try to
(wrongly) link against it resulting in build failure. Fixes:
http://autobuild.buildroot.net/results/51f/51ff079ebbd363e0445d226ec80bc28e524fd88c/

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

axfs: add new filesystem type

The Advanced XIP File System is a Linux kernel filesystem driver that
enables files to be executed directly from flash or ROM memory rather
than being copied into RAM. It has the ability to store individual
*pages* in a file uncompressed/XIP or compressed/Demand Paged.

This commit only adds support for 'XIP all' mode, so all the files that
have the execute attribute set will be XIP'ed.

At the moment, the FS is not supported in Linux mainline (v4.3-rc5), so
the kernel has to be built with the axfs patches to be able to read it.
Patches can be found here: https://github.com/jaredeh/axfs

[Thomas: rewrap Config.in help text.]

Signed-off-by: Ariel D'Alessandro <ariel@vanguardiasur.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

axfsutils: add new package

Tools for building AXFS Filesystem.

The Advanced XIP File System is a Linux kernel filesystem driver that
enables files to be executed directly from flash or ROM memory rather
than being copied into RAM. It has the ability to store individual
*pages* in a file uncompressed/XIP or compressed/Demand Paged.

Source:
https://github.com/jaredeh/axfs

[Thomas:
 - add hash file.
 - use $(INSTALL) -D -m 0755 instead of cp for installing mkfs.axfs.]

Signed-off-by: Ariel D'Alessandro <ariel@vanguardiasur.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>