logrotate: workaround wget issue with wildcard ssl certificates

Fixes http://autobuild.buildroot.net/results/815/815e2759f4961805d768efe79f0d83d13b16e706/build-end.log

Certain wget versions dislike downloading from fedorahosted.org which
provides a SSL certificate for *.fedorahosted.org. Work around it by
instead downloading from www.fedorahosted.org instead.

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

boards/rpi: bump kernel

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

package/genext2fs: add host variant

genext2fs is built only if the user selected an ext2 root filesystem.

However, some use-cases can't live with the full target/ dir on the
root filesystem, and requires separate partitions (eg. for /usr).
In this case, the user would not select an ext2 root fs in the
Buildrooot menu, and would only generate a tarball of the rootfs.
This tarball would then be used from a post-image script to build
the actual required FSes.

But then, genext2fs is not built, since the ext2 root FS was not
selected.

As for the other filesystem generators, provide a host variant of
genext2fs (genext2fs is already host-package aware, so only needs
adding a Kconfig entry).

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

package/linux-firmware: bump version for new firmware

Bump version to get a few new firmware files for some WiFi adapters.

In the rtl81xx familly, it's hard to differentiate each members: I have
a USB stick that identifies itself as TRL8188CUS, but is handled by the
rtl8192ce driver. So I moved all of rtl81xx firmwares under the same
option to avoid confusion.

It will be the responsibility of the user to eventually get rid of
unwanted firmware files in a post-build script, if needed.

Ditto for rtl87xx.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

package/tvheadend: update to a git snapshot

There are a buncha fixes that are not available in a release yet.

Needs refreshing our patches.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

git: needs libiconv on toolchains without locale support

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

git: needs zlib and host-gettext

host-gettext fixes:
http://autobuild.buildroot.net/results/a5b/a5b1bb708d62a9e296f05899f0984087fcf9e2ce/

zlib fixes:
http://autobuild.buildroot.net/results/08f/08f5ed4dd35ad1ce06dda7577aae87b9aa7069b2/

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

gnutls: wchar only needed for tools

And adjust reverse dependencies to match.

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

gnutls: bump to version 3.2.1

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

p11-kit: new package

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

aespipe: needs mmu

Fixes:
http://autobuild.buildroot.net/results/9f3/9f3f85a024791cdc6b10e3e40cf41707eb074447/

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

libbsd: correct license file reference

Fixes autobuilder error:
http://autobuild.buildroot.net/results/ccd/ccd2f3fe69a3e0321685daf95f85eca5ba2d64cf/

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

sconeserver: fix autouild failure

Fixes autobuild failures such as the following.

  http://autobuild.buildroot.net/results/1bdb5e014daf76b701afea5c82ac83ac4c97d4cb/

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

barebox: bump to version 2013.06.0

Signed-off-by: Fabio Porcedda <fabio.porcedda@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

sconeserver: requires thread support

Fixes autobuild failures such as the following.

  http://autobuild.buildroot.net/results/a290aa34bd44c0e08f7aa44a7c606420668eef6d/

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

php: bump to version 5.3.25

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

lm_sensors: bump to version 3.3.4

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

pcre: bump to version 8.33

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

ecryptfs-utils: new package

The following patch adds support for ecryptfs-utils.
http://ecryptfs.org/

[Peter: needs host-intltool, fix typo in help text]
Signed-off-by: Marcin Bis <marcin@bis.org.pl>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

intltool: bump version

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

Add support for the Calao-systems TNY-A9G20-LPW

[Peter: lock kernel/headers version to 3.9.4, use devtmpfs, add comments]
Signed-off-by: Gregory Hermant <gregory.hermant@calao-systems.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

Merge branch 'next'

Makefile: kickoff 2013.08 cycle

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

docs/news.html: add 2013.05 announcement link

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

Update for 2013.05

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

Update copyright info for 2013

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

CHANGES: update with recent changes

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

ext-toolchain-wrapper: fix paths if executable was resolved by PATH

If ext-toolchain-wrapper or any symbolic link to it was resolved by PATH,
the wrapper takes the working directory to calculate the relative paths.

Now '/proc/self/exe' is used to resolve the absolute path to the toolchain
directory if the wrapper was called neither with a relative nor an absolute
path.

[Peter: fix off-by-one, swap value == var checks around]
Signed-off-by: Patrick Ziegler <patrick.ziegler@fh-kl.de>
Tested-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

iptables: bump to version 1.4.19.1

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

gcc: bump 4.8.x version

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

qt5: mysql and sqlite plugins

Added support for mysql and sqlite plugins. Configuration is based on qt package.

[Peter: strip trailing spaces]
Signed-off-by: Wojciech Slenska <wojciech.slenska@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

package/snappy: new package

[Peter: needs C++ support]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

package/libedit: new package

[Peter: rename to libedit]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

package/pmake: force argv[0]=${0} in the pmake wrapper

This forces bmake to call the wrapper, not itself.
Also, use consistent "${0/*}" instead of "${0/pmake}".

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

pulseaudio : bump to version 3.0

Signed-off-by: Jérémy Rosen <jeremy.rosen@openwide.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

libtool : bump to version 2.4.2

Signed-off-by: Jérémy Rosen <jeremy.rosen@openwide.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

qt5: add libQt5OpenGL.so library installation

When qt5 is built with EGLFS support, libQt5OpenGL library should be copied to
target.

Signed-off-by: Wojciech Slenska <wojciech.slenska@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

gutenprint: ensure target and host variant have the same patch hooks

In preparation for the out-of-tree support, it is important that all
packages have the same patch hooks for the host variant and the target
variant (because the source tree will be shared between host and
target variants when doing out-of-tree build).

In the case of gutenprint, the patch hook executed for the target just
creates one directory which is needed for autoreconf to work, and the
host variant is not autoreconf'ed. So there is no harm in also
applying this patch hook to the host variant.

[Peter: autoreconf for host as pointed out by Arnout]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

xlib_libX11: ensure target and host variant have the same patch hooks

In preparation for the out-of-tree support, it is important that all
packages have the same patch hooks for the host variant and the target
variant (because the source tree will be shared between host and
target variants when doing out-of-tree build).

In the case of the xlib_libX11 package, the post-patch hook removes
some flags from the build of the makekeys utility, so that when
xlib_libX11 is built for the target, the build of makekeys for the
host works properly. This modification can be applied to the host
variant of xlib_libX11 without any problem.

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

libgtk2: remove useless post-patch hook

There used to be a post-patch hook to apply a patch only on the host
variant of libgtk2. However, after the reorganization of the patch
naming policy, this was no longer working properly, so
dd1848bc73ee9dfe211ca3ed3a233707e3bc5955 changed this by a patch that
works on both the target and host variants.

We can therefore get rid of the useless post-patch hook.

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

binutils: apply post-patch hook on both the host and target builds

When a Xtensa specific patch needs to be applied, it should be applied
both when doing the target and the host build.

This change is part of the effort to remove the differences between
host and target post-extract and post-patch hooks, that are
problematic for the out-of-tree support.

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

gettext: don't patch different the host and target build

Until now, the gettext source code was being patched differently for
the host and the target build, which is going to be incompatible with
out-of-tree build. The difference is that the gettext tools could be
disabled in the target build, but are always enabled on the host
build.

Therefore, this commit switches the gettext package to using proper
patches against the relevant configure.ac and Makefile.am
files. gettext is now being autoreconfigured. Configuration options
are also passed for the host variant, in order to disable the build of
a large number of things we don't care about.

Finally, the two existing patches are renamed to include a number in
their filename.

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

neard: add init script

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

libopenmax: Add libopenmax virtual package

this adds the libopenmax virtual package for hardware based video acceleration

[Peter: fix openmax-without-any-backends check]
Signed-off-by: Spenser Gilliland <spenser@gillilanding.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

libvpx: new package libvpx

adds webm multimedia support and the VP8 encoder/decoder library.

http://www.webmproject.org/

[Peter: strip trailing spaces, wrap help text]
Signed-off-by: Spenser Gilliland <spenser@gillilanding.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

system/device-table: do not set /home/default stgid

Currently, /home/default is mode 2755 which means it is setgid.
Since /home/default is not group-writable, it is useless.

Signed-off-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>

opengl: correct detection of virtual packages having no dependencies

I believe the original intent was to make it that the configure step
for the opengl virtual packages fails if there is not at least one
dependency. This patch fixes the logic so that it actually fails if
dependency list is empty

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

Revert "pulseaudio : bump to version 3.0"

This reverts commit d3bde0a64edb5a3392314d2cdd2c4e4057d15831.

This wasn't supposed to be applied to master.

libplayer: Remove unsupported QUIET variable

Fixes the following compilation problem when building with 'make -s'

Unknown option "-q".
See ./configure --help for available options.

Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

pulseaudio : bump to version 3.0

Signed-off-by: Jérémy Rosen <jeremy.rosen@openwide.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

kmod : protect against architectures without __NR_finit_module

This bug caused udev to fail to load any module, crashing with Illegal
instruction.

The patch was taken from upstream and should be integrated in the next
version of kmod (version 14)

Signed-off-by: Jérémy Rosen <jeremy.rosen@openwide.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

qt5jsbackend: only available on ARM, x86, x86-64 and mipsel

The v8 Javascript virtual machine that is the core of qt5jsbackend is
only available on certain architectures. At the moment, only ARM, x86,
x86-64 and mipsel are supported.

Fixes:

  http://autobuild.buildroot.org/results/a2e/a2e62e2deaa28ec9d4d957523a00ee0b7a634950/build-end.log

[Peter: also add dependency for qt5graphicaleffects]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

libtirpc: prevent selection when ADI Blackfin external toolchains are used

libtirpc fails to build with the Blackfin external toolchains provided
by Analog Devices, because their uClibc configuration doesn't include
AI_ADDRCONFIG. In order to prevent such build failures to happen, we
simply prevent libtirpc from being selected when such toolchains are
used. This is not a big problem, since they provide native RPC
support.

Also, since they provide native RPC, we don't have to propagate this
new dependency to the reverse dependencies of libtirpc, because they
all use native RPC when available. The exception to this rule is the
rpcbind package, which can only use the libtirpc implementation of
RPCs, and not the native one of C libraries. Therefore, the dependency
is propagated to the rpcbind package.

Fixes:

  http://autobuild.buildroot.org/results/58b/58b16449065c16afce11ba120db56839efb2b1ea/build-end.log

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

czmq: prevent selection when ADI Blackfin external toolchains are used

czmq fails to build with the Blackfin external toolchains provided by
Analog Devices, because their uClibc configuration doesn't include
AI_ADDRCONFIG. In order to prevent such build failures to happen, we
simply prevent the selection of czmq or one of its reverse
dependencies when such toolchains are used.

This fixes:

  http://autobuild.buildroot.org/results/821/82140fac4c2a2cef3f38f06cada8f17fd7f0078b/build-end.log

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

configs: add sample for olimex mx233 olinuxino

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

logrotate: Bump version to 3.8.4

The changelog for logrotate indicates some CVE's were fixed in 3.8.0.
(changelog at http://svn.fedorahosted.org/svn/logrotate/tags/r3-8-4/CHANGES)

Signed-off-by: Arne Rovet <arnerro@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

network-manager: bump to 0.9.8.0

This upstream version (the latest available at the time of this
writing) contains
http://cgit.freedesktop.org/NetworkManager/NetworkManager/commit/?id=6b64e4db2f3c9cfc0e0e240cf0bc58f3b3e90c1f
which fixes a build issue of network-manager against recent kernel
headers.

We also remove our patch against network-manager, because it has been
merged upstream as commit cbf72aeb34a6f1fd1bcd7f78ae88985154dc85af.

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

network-manager: use a <pkg>_VERSION_MAJOR variable

Just like we do for many other GNOME-hosted tarballs, use a
<pkg>_VERSION_MAJOR variable for network-manager.

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

gpu-viv-bin-mx6q: new package

This patch adds a new package for the OpenGL, OpenVG and EGL
proprietary implementations specific to the Vivante GPU of i.MX6
processors.

[Peter: show comment on uClibc toolchains, tweak help text]
Signed-off-by: Henk Fijnvandraat <h.fijnvandraat@inter.nl.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

freescale-imx: bump to 1.1.0

This commit bumps both imx-lib and firmware-imx to version 1.1.0.

Signed-off-by: Henk Fijnvandraat <h.fijnvandraat@inter.nl.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

firmware-imx: move to the freescale-imx directory

This commit moves the firmware-imx package to the newly introduced
freescale-imx/ directory, and uses the new variables provided by
freescale-imx/freescale-imx.mk.

Signed-off-by: Henk Fijnvandraat <h.fijnvandraat@inter.nl.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

imx-lib: move to the freescale-imx directory

This commit moves the imx-lib package to the newly introduced
freescale-imx/ directory, and uses the new variables provided by
freescale-imx/freescale-imx.mk.

Signed-off-by: Henk Fijnvandraat <h.fijnvandraat@inter.nl.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

freescale-imx: new package directory

The freescale-imx directory is intended to contain all freescale
i.MXyy related packages, together with their download site and version
info

Signed-off-by: Henk Fijnvandraat <h.fijnvandraat@inter.nl.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

cubieboard: add support

[Peter: adjust for _FEX_FILE]
Signed-off-by: Carlo Caione <carlo.caione@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

at91bootstrap: Disable parallel build

The at91bootstrap build sometimes fail when doing parallel builds.
Build it without parallel build.

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

fs/ext2: further fix to the UUID

Turned out that setting a nil-UUID is no better than clearing it.

What currently happens is as follows:
  - first, genext2fs does not generate a UUID
  - then we tune2fs to upgrade the filesystem
  - then we run fsck, which generates a random UUID
  - then we re-run tune2fs to set a nil-UUID

So, on the target, if the file system is improperly unmounted (eg.
with a power failure), on next boot, fsck may be run, and a new
random UUID will be generated.

*However*, fsck improperly updates the filesystem when it adds the
UUID, and there are a few group descriptor checksum errors.

Those errors will go undetected until the next fsck, which will then
block for user input (bad on embedded systems, bad).

Fix that by systematically generating a random UUID _before_ we call
to fsck.

A random UUID is not so bad, after all, since there are already so
many sources of unpredictability in the filesystem: files date and
ordering, files content (date, paths...) which renders a fixed UUID
unneeded.

And it is still possible to set the UUID in a post-image script if
needed, anyway.

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: Peter Korsgaard <jacmet@sunsite.dk>

linux: bump 3.9.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, 9}.x stable version

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

flex: propagate m4 dependency on wchar

Commit fe6a9e5e9d4046173a60235f8fd9a3b4746929c6 missed the fact that
m4 has a dependency on wide char support, so this dependency should be
propagated to flex.

Fixes:

  http://autobuild.buildroot.org/results/495/49502b2e33a346b2fcebf5e2da00af2661b54d6e/build-end.log

[Peter: Note where wchar dep comes from, show comment if not available]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

libv4l: needs thread support

Fixes:

  http://autobuild.buildroot.org/results/c74/c741ca445a63ce9ab8b0e5b87f3126d7b6051009/build-end.log

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

acpid: fix obvious syntax error in addprefix call

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

wvstreams: need largefile support

Fixes:

  http://autobuild.buildroot.org/results/bc4/bc4c617a5d7740b6eedcfc0dd24aeed5bc888f6d/build-end.log

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

Revert "sqlite: bump to version 3.7.17"

This reverts commit ddddef3e6e52b4205306190e4a3830fad06d23aa.

This was supposed to have been for next, not 2013.05.

Revert "mpg123: bump to version 1.15.4"

This reverts commit 609aa4156aa6fbda182856d3dab7df752c54eb78.

This was supposed to have been for next, not 2013.05.

Revert "gmp: bump to version 5.1.2"

This reverts commit 78927d3f92f0a397e9551dee491e26602e63a30a.

This was supposed to have been for next, not 2013.05.

sqlite: bump to version 3.7.17

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

mpg123: bump to version 1.15.4

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

gmp: bump to version 5.1.2

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

sunxi-boards: new package

[Peter: depend on arm, wrap help, change to _FEX_FILE option, error on empty]
Signed-off-by: Carlo Caione <carlo.caione@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

bzip2: improve the packaging

Instead of doing some nasty SED tricks, use a patch that modifies the
bzip2 build system to make it a little bit nicer. Those SEDs were made
differently for the host and target packages, which is contrary to the
logic of not having any difference in the source tree between the host
build and the target build.

This commit also makes sure that bzip2.mk behaves properly in a
BR2_PREFER_STATIC_LIB=y context.

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

gdb: apply post-patch hook on both the host and target builds

When a Xtensa specific patch needs to be applied, it should be applied
both when doing the target and the host build.

This change is part of the effort to remove the differences between
host and target post-extract and post-patch hooks, that are
problematic for the out-of-tree support.

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

lttng-babeltrace: fix build failure on gcc14

The gcc14 machine has a moderately old host gcc, which was causing a
recurrent build failure of host-lttng-babeltrace:

 http://autobuild.buildroot.org/results/492/49216052c161874f41738e41e6e0c89a6dd04000/build-end.log

This commit adds a patch to fix this build failure, related to the
access of the ULLONG_MAX define on compilers that did not default to
the C99 variant of the C language.

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

barebox: bump to version 2013.05.1

Signed-off-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

sqlite: bump to version 3.7.17

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

mpg123: bump to version 1.15.4

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

gmp: bump to version 5.1.2

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

jamvm: Not available for all architectures

jamvm supports arm, armeb, powerpc, i386, x86_64 and mipsel.

Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

attr: bump to version 2.4.47

The attr-0001-attr-Fix-static-build-installation.patch has been integrated
in the new upstream version.

Signed-off-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

acl: bump to version 2.2.52

Signed-off-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

flex: fix autobuilder failures

Since fe6a9e5e9d (flex: needs M4 at runtime), the autobuilders have
been producing a number of flex related build failures. They have been
hard to track down, because even on the same machine, with the same
Git commit ID and the same configuration, the failure could not be
reproduced.

However, a close inspection of flex's config.log file allowed to find
out what the problem was. In its configure script, flex uses the
host-flex to generate a minimal example, and find out the name of the
output file of flex.

When the M4 environment is passed when building the target flex, it
also affects the *execution* of the host-flex, which tries to use
/usr/bin/m4 (which doesn't exist in the autobuilder machines) instead
of the one built in $(HOST_DIR)/usr/bin/m4. So generating the minimal
example fails. And this is where what I could reproduce and what the
autobuilders script produce differ: in my case, even though host-flex
fails to run, it creates an empty lex.yy.c, which is enough to make
the configure script happy. In the context of the autobuild scripts,
this file is apparently not created at all, for an unknown reason, and
this leads to the configure script to abort.

The fix is to set ac_cv_path_M4. This will affect the default m4 used
by the target flex, but it will not affect the m4 used by the
host-flex. It allows the test made during the configure script to work
properly, and therefore should fix the issue seen in the autobuilders.

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

fs/ext2: explicitly set a NULL UUID

"tune2fs -U clear" creates an invalid filesystem, that fsck.ext2
whines about later:

    $ make rootfs-ext2
    [--SNIP--]

    $ ./host/usr/sbin/fsck.ext4 images/rootfs.ext2
    e2fsck 1.42.7 (21-Jan-2013)
    Filesystem did not have a UUID; generating one.
    images/rootfs.ext2: clean, 4616/5120 files, 53171/131072 blocks

    $ ./host/usr/sbin/fsck.ext4 -f images/rootfs.ext2
    e2fsck 1.42.7 (21-Jan-2013)
    One or more block group descriptor checksums are invalid.  Fix<y>? yes
    Group descriptor 0 checksum is 0x4131, should be 0x8bdb.  FIXED.
    [--SNIP--]

So we set an explicitly NULL UUID instead.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

sunxi-tools: new host/target package

[Peter: Fix Config.in wordwrap and trailing lines]
Signed-off-by: Carlo Caione <carlo.caione@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

toolchain-external: remove support for 'Sourcery CodeBench ARM 2011.03'

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

toolchain-external: add support for 'Sourcery CodeBench ARM 2013.05'

Signed-off-by: Fabio Porcedda <fabio.porcedda@gmail.com>
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>

Remove useless CVS related files

Fixes bug #5912.

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

flex: needs M4 at runtime

For proper runtime execution, flex requires m4 to be
installed. Passing a M4 variable at configure time is needed,
otherwise flex on the target will try to use a 'm4' binary with a
build machine path.

Fixes bug #4988.

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

toolchain: blackfin: Remove Blackfin toolchain 2011R1 release.

ADI officially supports the buildroot and related GNU toolchain for
Blackfin since ADI's 2012R1 release only. In order to avoid confusion,
it is better to remove the 2011R1 GNU toolchain for Blackfin. In
addition, the 2011R1 GNU toolchain for Blackfin doesn't support the
BF60x processors.

Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

gcc: fix gcc 4.8 build when thread support is disabled

When thread support is disabled, the libitm and libatomic libraries
from gcc should be disabled, otherwise, the build of gcc fails.

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

elf2flt: fix build when zlib is not installed on the host

For a reason that's fairly unclear to me, Peter added a '-lz' link
flag to the elf2flt.mk build in d5664ee99 ("elf2flt: fix link").

However, the zlib library may not necessarily be installed on the host
machine, so we should depend on host-zlib, and pass the appropriate
LDFLAGS. This is what this patch does.

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