pkg-infra: introduce errors for legacy API

As discussed in the BR developer days, we want to be more strict about API
changes in buildroot. I.e., we want to make it less likely that a user's
customizations break down after upgrading buildroot.

A first step is to make sure that the user is warned about API changes.
This patch introduces Makefile.legacy and Config.in.legacy, which will
issue clear error messages for such situations.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

matchbox-lib: Append -lX11 to the linker

Fixes building with binutils-2.22
http://autobuild.buildroot.net/results/b4b26dd9c7d7bebe45638ba125fb49a5f5d0cc14/build-end.log

Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Tested-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

flex: fix build for nommu systems

The flex binary uses fork() so it breaks on !MMU builds.
Since we usually don't require flex in the target and the common
scenario is that we just want libfl in staging reverse the options so
that BR2_PACKAGE_FLEX just builds and install libfl.a and change the
LIBFL option to BR2_PACKAGE_FLEX_BINARY to install the binary in the
target.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

Add /var/www to device table

The /var/www directory is listed in /etc/passwd in the skeleton target
filesystem as the home directory of the www-data user (uid 33).

In the final target filesystem, /var/www should be owned by www-data. This
is important for the lighttpd package, for example.

Signed-off-by: Simon Dawson <spdawson@gmail.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

owl-linux: update help text about linux kernel dependencies.

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

imagemagick: add license info

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

scons: add license info

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

libmad: add license info

[Peter: license is GPLv2+ as pointed out by Arnout]
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

linux-fusion: bump to version 8.10.4

Also fix directfb build avoiding the following error to occur
(since FCEF_FOLLOW has been added in linux-fusion-8.9.0):

libtool: compile:  /opt/br/output/host/usr/bin/ccache /opt/br/output/host/usr/bin/arm-buildroot-linux-uclibcgnueabi-gcc -DHAVE_CONFIG_H -I. -I../.. -I../../include -I../../lib -I../../include -I../../lib -DDATADIR=\"/usr/share/directfb-1.4.17\" -DMODULEDIR=\"/usr/lib/directfb-1.4-6\" -D_REENTRANT -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -O3 -ffast-math -pipe -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -pipe -Os -g2 -g3 -fno-inline -Wno-inline -D_GNU_SOURCE -finstrument-functions -std=gnu99 -Werror-implicit-function-declaration -MT call.lo -MD -MP -MF .deps/call.Tpo -c call.c  -fPIC -DPIC -o .libs/call.o
call.c: In function 'fusion_call_execute3':
call.c:311:66: error 'FCEF_FOLLOW' undeclared (first use in this function)
call.c:311:66: note: each undeclared identifier is reported only once for each function it appears in
call.c: In function 'fusion_world_flush_calls':
call.c:444:54: error 'FCEF_FOLLOW' undeclared (first use in this function)
make[5]: *** [call.lo] Error 1
make[5]: Leaving directory `/opt/br/output/build/directfb-1.4.17/lib/fusion'

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Acked-by: Simon Dawson <spdawson@gmail.com>
Tested-by: Simon Dawson <spdawson@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

cramfs.mk: drop invalid -q option

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

Change Filesystem utilities heading because it has mtd utils too.

Changing it to Filesystem and flash utilities gives far better navigation
for mtd utils.

Signed-off-by: Charles Manning <cdhmanning@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

alsa-lib: fix static linking check

Closes #5732

The symbol for static linking is BR2_PREFER_STATIC_LIB.

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

CHANGES: update with recent changes

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

linux-fusion: fix build

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Acked-by: Simon Dawson <spdawson@gmail.com>
Tested-by: Simon Dawson <spdawson@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

strace: remove undef-CTL_PROC patch

This patch is no longer necessary since strace-v4.5.15 which always
defines CTL_PROC. Specifically strace git commit
35a55785ea8ff44d214af52085e3a5ea624730aa.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

linux: bump 3.6.x stable version

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

kernel-headers: bump 3.{0, 4, 6}.x stable versions

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

ccache: Fix URL

Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

gdb: properly pass the PATH when building gdb for the host

When building gdb for the host, we properly pass the PATH (through
HOST_CONFIGURE_OPTS) during the configure step, but we forget to do so
for the compilation step.

The result of this is that when the Crosstool-NG backend is used, gawk
is built and installed in $(HOST_DIR), as a dependency of the
crosstool-ng package.

Then, the host gdb configure script detects this gawk binary
($(HOST_DIR) is in the PATH), and assumes gawk is
available. Unfortunately, during the compilation step, it fails to
find the expected gawk binary, because $(HOST_DIR) is no longer in the
PATH. This causes the following build failure:

  http://autobuild.buildroot.org/results/067d0c2ea01673ba98ec11de2426f1ab92dac800/build-end.log

In order to fix this, we simply call the compilation step of gdb for
the host with $(HOST_MAKE_ENV), as it should have been done from the
beginning.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

binutils: not available for the target on AArch64

Fixes:

  http://autobuild.buildroot.org/results/74082860f607cbd857654e9504f5f7e7ae47c0ac/build-end.log

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

libpcap: fix static build

The libpcap package contained some interesting attempts to support a
static-only build, but it was not working:

 http://autobuild.buildroot.org/results/01038d3b970092d894d2bae80679247e65722785/build-end.log

In fact the configure.in of libpcap has provision to support
--enable-shared/--disable-shared, but the generated configure script
in the libpcap package has not been regenerated with the configure.in
changes.

So basically, enabling LIBPCAP_AUTORECONF=YES ensures that the
configure script gets generated, which brings us a working
--enable-shared / --disable-shared.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

libdaemon: needs mmu

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

hiawatha: needs mmu

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

manual: restructure 'Adding packages' chapter

The depends-on-vs-select part of the manual really deserves its own
section title (especially because it is referred to and the xref gets
a 'sinpara' in PDF if the section doesn't have a title). So restructure
the surrounding sections to reduce the section nesting depth.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Acked-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

manual: give example where _INSTALL_TARGET = NO

The tutorial for autotools-package and cmake-package currently gives
the bad example of setting _INSTALL_TARGET to YES, which is the default.
So change this into an example with _INSTALL_TARGET = NO, and explain in
which case this is relevant.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Acked-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

manual: more warnings to not use output/target

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Acked-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

manual: various fixes

Various consistency and correctness improvements.

Also removing some sentences that are not or no longer relevant.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Acked-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

manual: trivial fixes

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

ocf-linux: tweak kernel patch for newer kernels

Newer kernels had a slight change in the random number API so the OCF
patchset requires a slight tweak to build and run properly.

Affected kernels are, per series:
>=3.0.41, >=3.2.29, >=3.4.9, >=3.5.2 and any newer version.

The fix is applied by detecting the API change rather than guessing the
kernel version.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

lcdproc: fix autobuild errors

The autobuilders are falling over building lcdproc, with failures like the
following.

  http://autobuild.buildroot.net/results/622b7da10be751c725ba25eb40102269790b4b03/build-end.log

As Thomas Petazzoni has pointed out, the compile command lines incorrectly
contain host header and library search paths, such as the following.

  -I/home/peko/scratch/host/usr/include/freetype2
  -L/home/peko/scratch/host/usr/lib -lfreetype

This patch changes the lcdproc package to pass the freetype installation
prefixes on the configure command line.

Signed-off-by: Simon Dawson <spdawson@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

directfb: needs threads support

Fixes http://autobuild.buildroot.net/results/237aaf7b036a98867de1df47d39d01ec4137efd3

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

Revert "libhid: disable on avr32"

This reverts commit 2424cc6783bbfb542f08cf1985b8c9ef18282161.

Now that we have a timerfd fix for libusb, this can be reverted.

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

AVR32: fix libusb build

The libc of the avr32 toolchains seem to lack the timerfd-associated
functions.

Pass --disable-timerfd to libusb configure script when building for
avr32.

Fixes most of the recent avr32 breakages, such as
http://autobuild.buildroot.org/results/f418b6c0b026891c27467f61444b6eac8c105681
or
http://autobuild.buildroot.org/results/1a9e571998bc0a40822dca84267abbdd969bab5e

Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

libecore: fix build with glibc >= 2.16

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

arptables: needs mmu

arptables uses fork() and breaks on !mmu.
Fixes
http://autobuild.buildroot.net/results/fe18dd4a0b0d17404560abf3e4a332b58751f5cb/

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

linux: bump 3.6.x stable version

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

kernel-headers: bump 3.{0, 2, 4, 6}.x stable versions

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

fs/common: Create initial console device

A /dev/console node must be present in rootfs when the Linux kernel
boots otherwise the kernel will print the following warning:
"Warning: unable to open an initial console"

This is because when we use an initramfs the /dev directory is not
populated at this point. This can cause problems when a program
(e.g ldso with early debugging enabled) opens a standard file
descriptor for read/write before these descriptors are actually
created by the init process later on.

Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Tested-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

Remove unused barebox 2012.07 string

Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xtensa: fix target gdb build and 64-bit host side gdb

Fix build breakage, use the version of the ptrace header file in asm
instead of sys. Also, fix GDB running on 64 bit hosts. GDB was using
unsigned long for 32-bit registers, but unsigned long is 64 bit on
64-bit hosts.

Signed-off-by: Chris Zankel <chris@zankel.net>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xtensa: use uppercase for configurations and modified overlay structure

Except for architecture and processor names, buildroot uses capitalized
configuration names, so change the macro names for xtensa to follow that
standard.
Change the overlay file to have a subdirectory for each component
(gdb, binutils, gcc, etc.) to make it more future-prove.

Signed-off-by: Chris Zankel <chris@zankel.net>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

busybox: fix link with tirpc

Busybox does partial linking of its modules before linking everything
together into the binary.  Those partial links are done without the
standard libraries, but that also means -ltirpc can't be found.  In
addition, this probably fails horribly with static linking (untested).

The problem is that the LDFLAGS are also used in the partial links.
So instead, use CFLAGS_busybox, which is only used for the busybox
link step.  Also make sure that this is passed through the environment,
not on the command line, so the busybox Makefile can still append to
it.

Fixes e.g.
http://autobuild.buildroot.net/results/e8351e3ba86fdcdb2999548658271a6fde0526a9

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

cjson: fix copy/paste error in target installation

Signed-off-by: Danomi Manchego <danomimanchego123@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

barebox: Add BR2_TARGET_BAREBOX_2012_11

Signed-off-by: Gregory Hermant <gregory.hermant@calao-systems.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

news.html: add 2012.11-rc1 annoucement link

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

Update for 2012.11-rc1

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

CHANGES: update with recent changes

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

dbus: uses fork(), requires MMU

In order to solve
http://autobuild.buildroot.org/results/34f6843137efda20626af72714c110280ec577d7/build-end.log,
this patch makes the D-Bus package as well as all the packages that
select the D-Bus package 'depends on BR2_USE_MMU'.

In addition, for the specific case of gvfs, the missing
BR2_TOOLCHAIN_HAS_THREADS dependency is added (threads are required by
D-Bus, so they are also required by gvfs which selects D-Bus).

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

attr: fix static build

Fixes:

  http://autobuild.buildroot.org/results/58a3d4113172813183d43fddc1248b35bd0c4994/build-end.log

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

php: fix C++ link issue

The intl module is implemented in C++, but PHP fails to use g++ as the
compiler for the final link. As a workaround, tell it to link
with libstdc++.

Fixes:

  http://autobuild.buildroot.org/results/13eefdf2121fa1aea7a844bbed8b6cbce3b45996/build-end.log

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

gsl: fix build on !i386 uClibc

uClibc pretends to implement <fenv.h> as it installs the header, but
in practice, it only implements the functions for i386. This makes gsl
unhappy as it detects fenv.h, but then cannot use the fenv functions.

Fixes:

  http://autobuild.buildroot.org/results/732cc07faeca2a9098dc5106e8f654eb1323451a/build-end.log

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

gnutls: fix build failure due to non-existing gets in recent C libraries

Fixes:

  http://autobuild.buildroot.org/results/7414989c182f8fd5b3ac849b7897b8e15df0024f/build-end.log

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

libatomic_ops: bump to version 7.2

The 1.2 we were using doesn't build on ARM Thumb platforms:

  http://autobuild.buildroot.org/results/94ef6be7dcb31021462e7313724217627d4b29df/build-end.log

Moreover, the 7.2 version is the one used by Debian/Ubuntu, and it has
been tested to build fine with the two packages that depend on
libatomic_ops: libdrm and pulseaudio.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

mesa3d: requires host-libxml2 and host-python to generate some files

mesa3d now generates some C files at build time (related to the OpenGL
API) from XML files. This generation process is done using Python
scripts that require the libxml2 Python module.

Patch based on the initial work of Will Wagner (Thanks Will).

Signed-off-by: Vellemans Noel <noel.vellemans@visionbms.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

libxml2: add support for building host library with python support

Patch based on the initial work of Will Wagner (Thanks Will).
libxml2 host library with python support is required to build mesa3d (7.10.1)

Signed-off-by: Vellemans Noel <noel.vellemans@visionbms.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

python: use default configure commands to get shared build

Commit 3c90f754961c6d99059c0b6d0a66bc797a3c017b made Python use a
special ./configure command in order to avoid --enable-shared
--disable-static being passed, because it was causing issues when
building certain modules for a 64 bits system.

However, not having a shared libpython2.7 library for the host
prevents the libxml2 Python binding to get built.

So instead, we use the default configure command, but we add
--enable-static which is needed for Python to build correctly.

Note that we tested the build of Python on a 64 bits host as well as
the build of Python for a 64 bits target, and both went fine, with all
modules built properly.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

CHANGES: update with recent changes

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

news.html: announce dev days report / next dev days

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

lua: don't be fancy when creating shared library

(Possibly) fixes #5354

The lua shared library patch was creating the shared library with
-nostdlib -lgcc for some unknown reason, which most likely is
the reason for the link issue reported in #5354.

Fix it by dropping these arguments, so gcc gets to figure out itself
what dependencies are needed.

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

hiawatha: bump to version 8.6

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

polarssl: bump to version 1.2.0

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

collectd: bump to version 5.1.1

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

Allow to run severals post build scripts instead of only one

Using severals post build scripts is usefull to share
script between severals boards/projects.

[Peter: fix trailing spaces in Config.in]
Signed-off-by: Philippe Reynes <philippe.reynes@sagemcom.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

alsa-utils: bump to version 1.0.26

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

alsa-lib: bump to version 1.0.26

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

Warn the user about the usage of output/target as the root filesystem

A very common mistake done by our users is that they use
output/target/ directory as their root filesystem. Even though this is
loudly documented in our Buildroot manual, people don't read
documentation, so it is not sufficient.

This patch adds a text file named
output/target/THIS_IS_NOT_YOUR_ROOT_FILESYSTEM which explains why
output/target isn't appropriate to use as the root filesystem. The
process is:

 * At the beginning of the build, right after the skeleton has been
   copied, support/misc/target-dir-warning.txt is copied to
   output/target/THIS_IS_NOT_YOUR_ROOT_FILESYSTEM

 * In the filesystem images creation code, this file is removed before
   launching fakeroot, and restored right after that, so that this
   file is not present in the generated root filesystem images.

Note that the file has not been added to the default skeleton for two
reasons:

 * It would have annoying to have in our source tree a file named in
   capital letters inside system/skeleton/

 * The proposed way works even if the user uses a custom skeleton.

[Peter: fixed typo]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Juha Lumme <juha.lumme@gmail.com>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

lua-msgpack-native: fix fetch

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

mxs-bootloader: Barebox entry point symbol changed in 2012.08

Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

python: uses fork(), only available on MMU platforms

Fixes
http://autobuild.buildroot.org/results/1738a8b09b04d2b2fc1de6d1ff5eb0363fed6348/build-end.log

Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

Mark a number of development related packages as deprecated

automake, autoconf, libtool and make on the target are basically
useless if we don't support building a toolchain on the target. Of
course, the host variant of automake, autoconf and libtool will remain
available.

[Peter: fixup to apply after perl change]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

Deprecate the support for the toolchain on target

As discussed during the ELCE 2012 Buildroot Developers Meeting, we no
longer want to support the possibility of building a toolchain for the
target. None of the core developers have any use for this, it has been
known to be broken or cause problems for a long time without anyone
providing fixes for it.

In addition to this, Buildroot is inherently a cross-compilation tool,
so the usage of a native toolchain on the target is not really
useful. Many newcomers are tempted to use this possibility even though
it is clearly not the intended usage of Buildroot.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

libffi: add aarch64 support from upstream

Fixes build failures like:

  http://autobuild.buildroot.org/results/6b4b4c414d366c936df480747c78272050c84293/build-end.log

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

libffi: make blackfin patch a proper git patch

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xapp_xvidtune: bump version to 1.0.2 to fix Xmu related build failure

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xapp_xmh: bump version to 1.0.2 to fix Xmu related build failure

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xapp_xlogo: bump version to 1.0.4 to fix Xmu related build failure

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xapp_xload: bump version to 1.1.1 to fix Xmu related build failure

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xapp_xfd: bump version to 1.1.1 to fix Xmu related build failure

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xapp_xditview: bump version to 1.0.2 to fix Xmu related build failure

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xapp_xclipboard: bump to 1.1.2 to fix Xmu related build failure

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xapp_viewres: bump to 1.0.3 to fix Xmu related build failure

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xapp_twm: bump to 1.0.6 to fix Xmu related build failure

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xapp_editres: bump to 1.0.5 to fix Xmu related build failure

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xutil_util-macros: bump to version 1.17

Needed by other xapp versions bump, which themselves are needed to fix
the missing link against the Xmu library (build failures similar to
http://autobuild.buildroot.org/results/1a4f84ad76ad8c4e70bff6878fc232edf46ff8f1/build-end.log).

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xlib_libXt: bump version to 1.1.3

Needed by the bump of xapp_clipboard, itself needed to avoid the
following build failure:

  http://autobuild.buildroot.org/results/1a4f84ad76ad8c4e70bff6878fc232edf46ff8f1/build-end.log

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

barebox: Add 2012.11, remove 2012.07

Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

mysql_client: security bump to version 5.1.66

Fixes multiple security isssues, most of them of the server component
though.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

gdb: not available on aarch64 for the target

Fixes

  http://autobuild.buildroot.org/results/d32a0fb1584a125cccde521a3960fb87ff7e2de6/build-end.log

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

manual: minor tweaks

Minor grammatical and spelling tweaks to the manual content.

Signed-off-by: Simon Dawson <spdawson@gmail.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Acked-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

cpanminus: mark as broken

cpanminus is currently responsible for about half of the autobuilder
failures, due to the missing dependency on host-qemu. However, even
with the host-qemu proposed by Arnout, cpanminus will still not work
properly: it will try to execute on the build machine executables
built for the target. While qemu is here to emulate the instruction
set, there is still the unsolved problem of kernel headers version
mismatch between the target and the build environments.

So the whole approach that consists in using host-qemu for building
simply cannot work properly, and until it is solved, the package
should be marked as broken.

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

kexec: disable on unsupported architectures

kexec only supports the following architectures.

  i?86
  powerpc
  powerpc64
  arm*
  s390x|s390
  sh4|sh4a|sh3|sh
  mips*
  cris
  crisv32
  ia64
  x86_64
  alpha

The Buildroot kexec package is available when an unsupported target
architecture is selected, leading to autobuild failures like the following.

  http://autobuild.buildroot.net/results/e706b08abb4fd1987fc4aa970d6be50fc92dc1fa/build-end.log

This patch makes the kexec package available only for the supported
architectures.

Signed-off-by: Simon Dawson <spdawson@gmail.com>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

can-utils: add CAN FD support

Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

manual: contribute.txt: do not use --summary option with 'git format-patch'

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

manual: prerequisite.txt: add texinfo to the mandatory package list

Using the internal Buildroot toolchain backend, makeinfo, whom belongs to
the texinfo package, is required to build gcc and gdb.

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

toolchain/gdb: disable support of full gdb on target for blackfin

There is no support in gdb 6.6 for the Blackfin architecture, so
disallow this choice. This fixes the following build failure:

  http://autobuild.buildroot.org/results/e37db29810265a756833da163fda444d967d7874/build-end.log

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

libnspr: fix build with ARMv7 toolchains

Fixes the following build failure:

  http://autobuild.buildroot.org/results/d7323831372050e425a34f5104a46d8cbd6be214/build-end.log

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xlib_libXt: add patch to fix cross-compilation issue

Fixes the following build failure:

 http://autobuild.buildroot.org/results/2815885d589d75259508bcce4f272b8dddb3be77/build-end.log

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

binutils: bump 2.23 series to 2.23.1 and deprecate 2.20

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

lcdproc: remove patch rejected upstream

Remove an lcdproc patch which has been rejected upstream. It has been
decided by lcdproc developer Markus Dolze that the behaviour of the
client_add_key command when handling the minus key is "intentional to
some degree."

This patch also takes the opportunity to remove the version number from the
single remaining lcdproc patch.

Signed-off-by: Simon Dawson <spdawson@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

lcdproc: fix autobuilder failures

The lcdproc version bump to 0.5.6 has resulted in autobuilder failures such as
the following.

  http://autobuild.buildroot.org/results/9967bc20a6094e836c4c18ff2cd30edef143bb8e/build-end.log

The 0.5.6 release added a new glcd driver, which requires the freetype and zlib
libraries. This patch makes the lcdproc package depend on freetype and zlib.

Signed-off-by: Simon Dawson <spdawson@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>