package/linux: check for config file before calling kconfig-package

If we check that the user provides a config file after we call to the
kconfig-package infra, the error message we get is the one for the
kconfig-package infra, not the custom error message we want to show to
the user.

So, only call kconfig-package after we do the check. Move the check with
the existing checks for the DTS, for consistency.

[Thomas: put the checks together, but right before the kconfig-package
call, rather than in the middle of the code, were the DTS related
tests were located.]

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

package/linux: don't enforce check for DTS when not building

Currently, this is triggering the error message:
    make randconfig
    make source

Limit the checks that enforce a DTS is set and at most one DTB is
appended to when we are actually building, like is done for the
configuration-file variables.

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

core/pkg-kconfig: don't enforce check for config file when not building

Currently, this is triggering the error message:
    make randconfig
    make source

Only limit the check to enforce a config file being set to when we are
actually building, like is done in the various kconfig-using packages.

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

system: fix BR2_INIT_NONE / BR2_INIT_SYSTEMD build

Following the introduction of the initscripts package in commit
89d39fc7a392530be043bac541ade0bef746edb6 ("initscripts: new package"),
the /etc/inittab file is no longer part of the skeleton, and therefore
is not always installed.

However, system/system.mk tweaks the inittab for getty and filesystem
remount as rw, without taking the precautions of whether the init
system is Busybox or SysV. This commit fixes that by adding the
necessary conditions around the code adjusting the inittab file.

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

toolchain-external: fix support of uClibc-ng toolchains

The uClibc-ng dynamic loader is called ld-uClibc.so.1, but gcc is not
patched specifically for uClibc-ng, so it continues to generate
binaries that expect the dynamic loader to be named ld-uClibc.so.0,
like with the original uClibc.

Therefore, when a uClibc-ng toolchain is used as an external
toolchain, we need to create an additional symbolic link to make
uClibc-ng systems work properly.

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

core: fix download menu

Currently, when one enables the "Only allow downloads from primary"
option, all the other mirror options are hidden (expected!) but the
LuaRocks and CPAN mirrors (bad!).

Fix that.

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

rpcbind: add missing dependency on host-pkgconf

rpcbind depends on libtirpc, which used to depend on
host-pkgconf. However, since commit
aee57627b09d0955270a562b8adaa0ff88ef523b, the dependency of libtirpc
on host-pkgconf has been removed, because libtirpc no longer uses
PKG_CHECK_MODULES().

However, rpcbind does use PKG_CHECK_MODULES(). It used to work because
host-pkgconf was brought into the build by the dependency on libtirpc,
but that's no longer the case. Bottom line: we need to make rpcbind
depend on host-pkgconf, which is exactly what this commit does.

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

purge-locales: fix handling of X11 locale.dir

The /usr/share/X11/locale/locale.dir file is needed by libX11.
Removing it breaks locale support in X11. However, make removes
not only directories but also all files, which are not listed
in the BR2_ENABLE_LOCALE_WHITELIST.

This re-creates locale.dir database file where needed.

Signed-off-by: Valentine Barshak <gvaxon@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
[Arnout: use a separate loop, and add some explanatory comments]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

purge-locales: further refactoring after removing man paths

Since the man paths have been removed, it is no longer necessary to
grep them out and the loop can be simplified.

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

initscripts: new package

The folder init.d is currently installed by default since it's part of
our skeleton.
This patch creates a package out of it and make busybox/sysvinit depends
on it.

This way, if you chose another init, you don't end up with a useless
init.d folder.

[Thomas:
  - make the initscripts package selectable via a hidden bool
  - remove some unneeded changes in sysvinit.mk.]

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

fs/iso9660: change the name of the tmp dir to be consistent

Currently, the generated rootfs will be called rootfs.iso9660, but the
temporary directory we create while building it is called
rootfs-iso9660.tmp. They are in different directories so it's not so
obvious, but still to be consistent it's better to call the temp dir
rootfs.iso9660.tmp.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: 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>

package/dropbear: fix generating keys on RO file systems

dropbear generates its keys at the first connection, and wants to save
them in /etc/dropbear (not configurable).

Currently, our /etc/dropbear is a directory.

When the filesystem is read-only, dropbear can't save its keys, so
refuses all connections.

Fix that with:

  - at build time, create /etc/dropbear as a symlink to
    /var/run/dropbear

  - at runtime, if the filesystem is RW (we can rm /etc/dropbear),
    we replace the symlink with an actual directory; otherwise,
    when the filesystem is RO (we can't rm /etc/dropbear), we create
    /var/run/dropbear so the symlink points to an existing directory

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Acked-by: "Maxime Hadjinlian" <maxime.hadjinlian@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

purge-locales: skip /usr/man and /usr/share/man

These directories are going to be removed anyway, so no point purging
their locales.

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

netplug: systemd support

[Thomas: use relative symlink instead of absolute symlink.]

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

avahi: fix systemd service file install path

[Thomas: fix indentation.]

Signed-off-by: Aurélien Chabot <aurelien@chabot.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

supervisor: systemd support

Running supervisord under systemd sounds wrong, but it is possible
and probably makes sense in some cases.

[Thomas: use relative symbolic link instead of an absolute symbolic
link.]

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

mysql: systemd support

Service startup follows sysv initscript and includes db init.

[Thomas: use a relative symlink rather than an absolute symlink.]

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

ntfs-3g: fix incorrect comment dependency

The comment for ntfs-3g displays even when the package has met its
necessary dependencies.

Signed-off-by: Ryan Barnett <ryan.barnett@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

skeleton: apply locale settings in user shells

Add /etc/profile.d/locale.sh script from Arch Linux to /etc/profile.
This script looks for locale.conf, sources it, and exports the LANG and
LC_* variables.

[Arnout: put in /etc/profile in the skeleton rather than making it
 systemd specific.]

Signed-off-by: Maxim Mikityanskiy <maxtram95@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

atftp: fix build with gcc5

gcc5 defaults to -std=gnu11 which changes the inline semantics. This leads to
multiple definition error with 'extern inline'. See
https://gcc.gnu.org/gcc-5/porting_to.html for the details.

Fixes:
http://autobuild.buildroot.net/results/762/762bce2bbc1f994cc796f0442bfa73b7ad64e272/
http://autobuild.buildroot.net/results/28b/28bf6018dcf91ddacfb76f54a92313da8bf182e1/
http://autobuild.buildroot.net/results/fde/fdee23ecd0c5c66fda1ed38efe43048ce934b733/

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

nginx: activate the systemd service

Signed-off-by: Aurélien Chabot <aurelien@chabot.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package: Fix source path for systemd's service

Otherwise the link would appear broken in output/target it's not that
big a deal but at least it's coherent with what we already have.

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

zeromq: Remove useless space

Be coherent with all the other ifeq check where there's no space.
>From the doc of Make, it should not cause any trouble, but coherence has
its own merit.

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

board/wandboard: add a genimage config

This patch adds a genimage config file and a post-image script for the
Wandboard, to generate a medium image "sdcard.img", ready to be booted.
The image contains the layout explained in the board readme.txt file:
U-Boot, its environment, and an Ext2 rootfs partition.

The defconfig has been slightly changed to enable this feature. Also
lighten the readme file since the config file is documented and simpler.

Tested on a Wandboard Solo.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
[Arnout: change the name of the tmp dir, and remove it before using
it.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

lua-iconv: new package

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

minidlna: fix systemd service activation symlink

Signed-off-by: Aurélien Chabot <aurelien@chabot.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

lighttpd: fix systemd service activation symlink

Signed-off-by: Aurélien Chabot <aurelien@chabot.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

input-event-daemon: systemd support

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

libv4l: needs dynamic library support

Fixes:
http://autobuild.buildroot.net/results/752/752cda64e3a4a73adfd9b383d5f42b61e4a1e641/
http://autobuild.buildroot.net/results/042/0423132222f8a8030f703bb8d755f3e9a4ee73d8/
http://autobuild.buildroot.net/results/908/9087769da17e2092c3ad2eb901ff87c1d6437e6a/

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

inadyn: systemd support

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

exim: systemd support

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

package/dovecot: Add optional support for icu

Optional dependency added to fts plugin since Dovecot 2.2.17:
http://hg.dovecot.org/dovecot-2.2/diff/b179bbd226e5/configure.ac

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

package/ntp: increase pool server count in default config

To protect agains 1 falsticker NTP server, the client needs to connect
to at least 4 servers.

Source:
 http://support.ntp.org/bin/view/Support/SelectingOffsiteNTPServers
 5.3.3. Upstream Time Server Quantity

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

powerpc-utils: optionally use librtas

Now that librtas is available, optionally use it with powerpc-utils to
enable it's RTAS dependant functionality.  It is used by default, as
most distros do, but made optional because it adds several new programs
and significantly increases the size of the installed package.

[Thomas: propagate the glibc dependency from the librtas package, as
suggested by Romain Naour.]

Signed-off-by: Sam Bobroff <sam.bobroff@au1.ibm.com>
Reviewed-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

librtas: new package

[Thomas:
  - make the package depend on glibc, since it needs execinfo.h
  - use TARGET_CONFIGURE_OPTS instead of defining manually CC and LD
  - add hash file
  - remove useless empty newline.
  Thanks to Romain Naour for the review!]

Signed-off-by: Sam Bobroff <sam.bobroff@au1.ibm.com>
Reviewed-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

linux: adjust fixup logic for ktap

The ktap package requires some parts of the kernel tracing
infrastructure to be enabled, especially
CONFIG_EVENT_TRACING. However, this option is a blind option in the
kernel, so enabling it in linux.mk has no effect: we need to enable a
non-blind option that selects CONFIG_EVENT_TRACING. We've chosen to
select CONFIG_ENABLE_DEFAULT_TRACERS.

This fixes the build of ktap.

[Thomas: use CONFIG_ENABLE_DEFAULT_TRACERS as suggested by Arnout.]

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

dcron: systemd support

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

chrony: systemd support

systemd has its own NTP daemon, which must be disabled before
starting chrony. Possible (but unlikely) conflict with openntpd
is not marked in either package.

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

systemd: add option to enable kdbus support

In the future when the kernel offically supports kdbus, this option
should automatically enable the kernel kdbus config option(s).

Signed-off-by: Ryan Barnett <ryanbarnett3@gmail.com>
Tested-by: Mike Williams <mike@mikebwilliams.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

binutils: remove version 2.22

We already use 2.24 as the default, 2.25 has been around for some
time, and 2.22 is clearly very old, so let's get rid of it.

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

binutils: use binutils 2.24 by default on the target

When binutils is not built for the host, binutils.mk decides of the
version to be used when building binutils for the target. We're still
using the old 2.22 binutils version, but it's time to bump up to 2.24,
which we already use as the default version for the host.

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

toolchain-external: improve lib subdirectory matching

The toolchain from the Cavium Networks Octeon SDK provides a sysroot
with library directories lib32, lib32-fp, lib64 and lib64-fp. The -fp
variants are used for processors with hardware floating point unit, such
as the Octeon III variants.

When specifying -march=octeon3 in BR2_TARGET_OPTIMIZATION, the toolchain
will use lib32-fp, but currently Buildroot does not accept that pattern.

This patch improves the matching by accepting lib(32|64)?([^/]*)? as lib
directory name.

Signed-off-by: Bai Yingjie <byj.tea@gmail.com>
[ThomasDS: update commit message]
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
[Thomas: add comment above the function being modified to illustrate
the various cases we try to handle.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

scripts: mkmakefile: set umask before calling BR's makefile

Small optimization so we don't have another 'make' level (caused by the
umask fix) when running the generated makefile.

Signed-off-by: Guido Martínez <guido@vanguardiasur.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

pkg-generic.mk: don't depend on external package permissions

Reset permissions for rsynced packages (when using OVERRIDE_SRCDIR) to
755/644. We do this under the assumption that source files shouldn't
care about their permissions, except possibly for the exec bit.

This guarantees that if a package uses 'rsync -a' or 'cp -p' to copy
a file from its build dir to the target, it'll end up with the same
permissions on the target every time.

Signed-off-by: Guido Martínez <guido@vanguardiasur.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

toolchain/helpers.mk: use --chmod on rsync

This makes sure we don't have any weird permissions on the staging dir,
which could affect the target.

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

docs/manual: fix typo

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gnutls: bump to version 3.3.16

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

nfs-utils: Create missing directory sm.bak

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

libpcap: bump to version 1.7.4

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

nmap: add hash file

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

CHANGES: add release note for the permissions

Since the rsync of skeleton or rootfs now changes the permissions of
the copied files and directories, this may break existing setups that
rely on these permissions being kept. So add a release note explaining
how it should be done.

[Thomas: fixup commit log as suggested by Baruch.]

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

Makefile: don't depend on current skeleton/overlay permissions

We use 'rsync -a' to copy the skeleton and overlays, so the target ends
up with the exact same permissions as on the repo. The problem is we
don't track these permissions, since Git doesn't allow for that (except
for the exec bit). This means users with different umasks at the time of
cloning could end up with different target permissions.

Fix this by using --chmod on rsync calls so we don't depend on the
current permission set for the skeleton and overlays. We do depend on
the exec bit, but that's fine since that one is tracked by Git.

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

busybox: improve support for telnetd

If target has connection to the network it might be pretty useful to
have telnet connection to it instead of serial console or even in
addition to serial console.

This changes makes the busybox package automatically install an init
script, and tune the securetty file to make telnetd work when
CONFIG_FEATURE_TELNETD_STANDALONE is enabled in the Busybox
configuration.

[Thomas:
  - don't create a new Buildroot Config.in option, just test if
    CONFIG_FEATURE_TELNETD_STANDALONE is enabled or not in the Busybox
    configuration
  - move the securetty tuning in busybox.mk instead of system.mk
  - use start-stop-daemon in the init script, in order to properly
    implement the stop and restart actions
  - misc other minor improvements in the init script.]
[yann.morin.1998@free.fr:
  - don't use securetty
  - drop stray variable BUSYBOX_SET_STANDALONE_TELNETD]

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

core/skeleton: drop /etc/securetty

securetty is supposed to restrict the terminals root is allowed to
login from. As it happens, login from busybox (w/ securetty support)
is actually enforcing use of securetty, while login from util-linux
is completely ignoring securetty altogether.

Remove securetty from our skeleton altogether and stop worrying about
it.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Makefile: don't depend on the umask

Some packages and BR itself create files and directories on the target
with cp/mkdir/etc which depend on the umask at the time of building.

To fix this, use a trick inside the Makefile which wraps all rules when
the umask is not 0022. This sets the umask at the top level, and then
the building process continues as usual.

[Thomas: add --no-print-directory, as suggested by Arnout.]

Signed-off-by: Guido Martínez <guido@vanguardiasur.com.ar>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Tested-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

nfs-utils: Bump version

Refresh the patches (Thanks to Thomas Petazzoni's work)

Patches removed:
0001-build-avoid-AM_CONDITIONAL-in-conditional-execution.patch
0004-fix-build-with-uClibc.patch
0004-fix-build-with-uClibc.patch
0005-Allow-usage-of-getrpcbynumber-when-getrpcbynumber_r-.patch
0007-sockaddr-h-needs-stddef-h-for-NULL.patch
0008-tirpc-with-pkgconfig.patch

Patches modified:
0002-Patch-taken-from-Gentoo.patch
0003-Switch-legacy-index-in-favour-of-strchr.patch
0006-Let-the-configure-script-find-getrpcbynumber-in-libt.patch

Patch addedd:
0004-statd-Fix-test-for-foreground-mode.patch

Also, change source of the package to git repository (and remove the hash,
sourceforce is clearly not a platform which can be trusted nowadays)

Rework the startup script to accomodate new rpc.statd
The startup scripts now uses rcp.statd -F for a startup in foreground,
also we avoid starting it twice, it makes rcp.statd crash the whole
script.

[Thomas: add patch to fix rpc.statd foreground/daemon mode backported
from upstream, and fix the S60nfs init script to not use the -F option
of rpc.statd.]

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

Document the 72 char limit on package description.

While describing a package I though that the tab and 2 spaces was
already doing the margin for wrapping the help test (ie 70 char when
wrapping to 80) but apparently the full text need to fit 72 columns.

So this patch propose to document that fact in the two places where
the Config.in format is described.

Signed-off-by: Julien Viard de Galbert <julien@vdg.name>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/libical: bump version to v1.0.1 and switch to cmake

Project moved to github according to http://sourceforge.net/projects/freeassociation

"As of 2014-06-14, this project may now be found at http://github.com/libical."

[Thomas: propagate the C++ dependency to bluez5_utils OBEX support,
which selects libical.]

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

package/x11r7/xdriver_xf86-video-r128: bump version to 6.10.0

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

package/libpciaccess: bump version to 0.13.4

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

package/x11r7/xapp_xvinfo: bump version to 1.1.3

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

package/x11r7/xlib_libFS: bump version to 1.0.7

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

package/x11r7/xlib_libXaw: bump version to 1.0.13

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

package/x11r7/xlib_libXrender: bump version to 0.9.9

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

package/x11r7/xlib_libXt: bump version to 1.1.5

Remove part of 0001-util-don-t-link-makestrs-with-target-cflags.patch
because it was applied upstream:
http://cgit.freedesktop.org/xorg/lib/libXt/commit/util/makestrs.c?id=f9baaf55ff8cbd4bf018a34f181eda30d03b20dc

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

package/x11r7/xlib_libfontenc: bump version to 1.1.3

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

package/x11r7/xlib_libxkbfile: bump version to 1.0.9

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

package/x11r7/xproto_kbproto: bump version to 1.0.7

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

pcmanfm: bump version to 1.2.3

- Bump version to 1.2.3
- Remove unnecessary patch
- Add a hash file

[Thomas:
 - adapt patch to the latest master
 - license is GPLv2+, not GPLv2
 - replace 'select BR2_PACKAGE_XORG7' by 'depends on BR2_PACKAGE_XORG7']

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

libfm: new package

[Thomas:
  - switch from 'select BR2_PACKAGE_XORG7' to 'depends on
    BR2_PACKAGE_XORG7' as suggested by Yann E. Morin.
  - add LGPLv2.1+ in the list of licenses as well as a corresponding
    license file, as suggested by Yann E. Morin.]

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

menu-cache: new package

[Thomas:
  - Remove the MENU_CACHE_VERSION_MINOR variable, not needed. Noticed
    by Yann E. Morin.
  - Rewrap Config.in help text.]

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

libfm-extra: new package

[Thomas:
  - adjust license: it's actually under GPLv2+, plus some bits under
    LGPLv2.1+ as noticed by Yann E. Morin, and add another license
    file for LGPLv2.1+
  - add missing dependency on host-intltool.]

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

configs: add altera_socdk_defconfig for Altera Cyclone 5 Development Board

This is largely the same as altera_sockit_defconfig.

It uses a fresher Linux and u-boot than SocKit. It also speeds the
serial port up to 115200.

The post-image script is generalized by adding
BR2_ROOTFS_POST_SCRIPT_ARGS and moving it up the altera directory.
Similarly, the readme is moved up and made more generic.

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

nfs_utils: Fix for read-only rootfs

In order to work on a read-only rootfs, nfs_utils would like to write
stuff to /var/lib/nfs, since it's not a tmpfs, it won't work.

Instead of doing little dances around the filesystem, tell nfs_utils to
use /run/nfs for everyone.

Modify the startup script accordingly.

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

nfs-utils: Add rpcbind as a runtime dependency

rpcbind is necessary for nfs_utils to works as a server

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

package/dbus: fix machine-id when using systemd

When systemd is used, /var/lib/dbus becomes dangling symlink, because
nobody does mkdir /tmp/dbus, so /var/lib/dbus/machine-id could not be
written. On SysVinit systems there is init script that creates
/tmp/dbus.

This patch preserves old behavior for SysVinit systems, and introduces
new one for systemd-booted systems: /var/lib/dbus is a persistent
directory, it holds symlink /var/lib/dbus/machine-id -> /etc/machine-id
as machine-id(5) suggests, and /etc/machine-id is managed by systemd.

Signed-off-by: Maxim Mikityanskiy <maxtram95@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

qt5base: add option for custom qconfig file

Add in support for a custom qconfig file in the same way it is
already present for the Qt4 package

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

Revert "rsyslog: fix config file comment style"

This reverts commit d42f0f96aa13bc160bdfeb40ca75f53adb7af64e, which
breaks the build since the patch being added by this commit is in fact
already part of the 8.9.0 rsyslog release we are using.

Indeed, in the rsyslog repository:

$ git tag --contains fe01c9ec4979c78b6a7caf045cab17ba52d4ad08
[...]
v8.10.0
v8.10.0.ad1
v8.11.0
v8.7.0
v8.8.0
v8.8.0.ad1
v8.9.0
v8.9.0.ad1

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

flite: fix <pkg>_LICENSE_FILES

Following commit dd4bcebd7c43212650cccc2494527dc3d69024d9 ("flite: use
FLITE_STRIP_COMPONENTS and FLITE_SUBDIR"), the FLITE_LICENSE_FILES was
no longer pointing to existing directories, which was breaking
legal-info.

Fixes:

  http://autobuild.buildroot.org/results/798/7983ec41a3f136bd997ebbfc110c7fc51ad79717/

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

package/systemd: add missing dependencies

Systemd depends on mount and nologin from util-linux. Mount from busybox
is not fine, because its behavior differs from util-linux mount. Busybox
mount does not handle correctly 'mount / -o remount' (should apply
options from /etc/fstab). Nologin should be present, because
systemd-sysusers uses it as shell for newly created system users.

Signed-off-by: Maxim Mikityanskiy <maxtram95@gmail.com>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

c-icap-modules: new package

[Thomas:
  - Do not pass custom CFLAGS and remove the hacks on the Makefile:
    since the c-icap-config script is now fixed, it returns correct
    values, and the hacks are no longer needed.
  - Add optional dependency on zlib and berkeleydb.
  - Explicitly disable clamav support.]

Signed-off-by: Guillaume GARDET <guillaume.gardet@oliseo.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package: Remove shell completion at finalize stage

Instead of doing a removal of the completion file package per package,
do it all at the finalize stage so it's done once and for all.

Note: This fixes an issue with systemd where passing a --bashcompletiondir
or --zshcompletiondir would be evaluated to '.' by the autotools macro.
This would create a 'target./' directory.

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

Fix selection of bootloaders from $(BR2_EXTERNAL)

If a package is based on "generic-package", pkg-generic.mk will compute
the name of the Kconfig variable to use for checking if this package has
been selected by the user.

Unfortunately, this mechanism does not take into account the case where
a bootloader is declared in a $(BR2_EXTERNAL)/boot directory.

So, even if the bootloader has been selected, it will not be added to
$(TARGETS) and will not be built.

This patch fixes this issue.

[Arnout: use single filter line, also apply this to toolchain]

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Cc: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/rpi-userland: drop -Werror

Fixes:
    http://autobuild.buildroot.org/results/501/5012affec6c85fe7749ac86d18d3c191f4e74bdf/

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

arch: aarch64 always has a MMU

Following the addition of AArch64 big endian, the AArch64 little
endian option had lost its 'select BR2_ARCH_HAS_MMU_MANDATORY', so
let's reintroduce it.

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

aarch64: add big endian(aarch64_be) support

Add aarch64_be support. Note that CONFIG_CPU_BIG_ENDIAN should be
defined in kernel config when building a big endian kernel.

Signed-off-by: Zhang Jian(Bamvor) <bamvor.zhangjian@huawei.com>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

pkg-autotools: Remove useless space

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

flite: use FLITE_STRIP_COMPONENTS and FLITE_SUBDIR

So far, to overcome the buggy way the flite tarball is made, we had to
override the extract commands in a rather ugly way.

The newly introduced <PKG>_STRIP_COMPONENTS, along with <PKG>_SUBDIR and a
little edit to the patches, allow to remove the custom FLITE_EXTRACT_CMDS,
slightly simplifying the flite.mk code.

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Cc: Samuel Martin <s.martin49@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

icu: update icudata generator URL

It was pointing to version 51.x which wouldn't work well with 55.x.

Point it to the latest version that matches the current buildroot
version (55). When a new ICU major version is released a link for the
"old" 55 version will be available below as is it currently for
previous versions such as 54, 53 and so on.

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

keyutils: add patch to fix musl build failure

Fixes:

  http://autobuild.buildroot.org/results/253/2536b59735f33709f44fac94c81106e5ddd4fc40/

Patch has been submitted upstream.

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

kismet: mark as not available for static builds

This reverts commit a3365ebf72fc34c71e28191f107280b9d7c825a1, and
marks kismet as not available for static builds.

Commit a3365ebf72fc34c71e28191f107280b9d7c825a1 is indeed
insufficient, since kismet uses <dlfcn.h> in two places, both in
kismet_server and kismet_client. And it's fairly unlikely that kismet
will be used in pure static environments, so let's not support this
use case.

Fixes:

  http://autobuild.buildroot.net/results/c24/c24e3e325df9748e9959fe79f7026d24f118dffd/

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

systemd: fix some coding style issues

[Thomas: adjust commit title, some additioanl coding style fixes.]

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

qpid-proton: new package

[Thomas:
   - Do not define QPID_PROTON_VERSION_MAJOR, and use proto-* instead,
     as suggested by Arnout.
   - Use QPID_PROTO_VERSION in QPID_PROTON_SITE, as suggested by
     Arnout
   - Remove useless comma in the openssl test, as suggested by Arnout
   - Change the post install target hook to remove the
     /usr/share/proton-*/ directory entirely, and not just the
     examples, since it only contains some license and documentation
     files.]

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

packages: do not use TAR_STRIP_COMPONENTS, but directly --strip-components

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

package/firmware-imx: fix legal-info

Part of fimrware-imx were dropped in the new version, so we have less
legal-info to save.

Fixes:
    http://autobuild.buildroot.org/results/59b/59b09152700a5839af46edb554e53981569f8921

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

pkg-generic: fix fallout of <pkg>_STRIP_COMPONENTS introduction

The introduction of <pkg>_STRIP_COMPONENTS broke the build of the
target tar package, because support/dependencies/check-host-tar.mk
defines TAR_STRIP_COMPONENTS to --strip-components. Which leads to
have the package infrastructure do:

 $$(TAR_STRIP_COMPONENTS)=$$($(2)_STRIP_COMPONENTS)

which for the tar package evaluates to:

 $$(TAR_STRIP_COMPONENTS)=$$(TAR_STRIP_COMPONENTS)

which evalutes to:

 --strip-components=--strip-components

Which obviously doesn't work really well. And in fact the
TAR_STRIP_COMPONENTS definition in
support/dependencies/check-host-tar.mk is no longer necessary: it was
needed in the days where we were trying to support old tar versions
that did not support --strip-components. But nowadays, when such an
old tar version is encountered, we build our own host-tar which
supports --strip-components.

Fixes:

  http://autobuild.buildroot.org/results/ae2/ae20df67f99f75b1ba5d5b7316ad265d66f3aa66/

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

c-icap: disable Perl support

The Perl support in c-icap does not build properly, because it queries
the system-installed Perl for cflags/ldflags, which returns things
like -I/usr/local/include causing build failures since those are
invalid paths when cross-compiling.

Since there is no easy way to fix that without changing c-icap
configure.in, let's disable the Perl support for now, until someone is
interested enough to fix this up.

Fixes:

  http://autobuild.buildroot.org/results/f60/f6011ab8211589fe8e1e8932db9f8be81768d3e6/

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

c-icap: needs MMU support

Fixes:

  http://autobuild.buildroot.org/results/8c3/8c3e397eeecb2a6a39fbae1d8665bff92ef3e3bb/

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

package/nodejs: version bump to 0.12.7

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
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>

package/sysdig: use kernel-module helper

sysdig's Makefile was already calling the Linux buildsystem to build its
modules, so switch to using the new kernel-module infra.

This has the benefit of making sysdig actually buildable, otherwise it
fails because it tries to use the kernel headers from the host:

    make: *** /lib/modules/3.13.0-53-generic/build: No such file or directory.  Stop.
    make[5]: *** [all] Error 2
    make[4]: *** [driver/CMakeFiles/driver] Error 2
    make[3]: *** [driver/CMakeFiles/driver.dir/all] Error 2

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