psplash: install systemd services

This patch provides service files for using psplash on embedded devices
running systemd:

- psplash-start.service: start psplash.
- psplash-quit.service: kill psplash when reaching multi-user.target

The following kernel command line options should also be set:

  systemd.show_status=0 quiet splash

The option "systemd.show_status=0" is required, because, unlike Plymouth,
psplash does not have real systemd integration, i.e. it will not perform:

  kill(1, SIGRTMIN + 21);

Note that no progress messages will be printed on the splash screen.

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

systemd: fix AM_PATH_LIBGCRYPT not found

This patch installs a copy of libgcrypt.m4 from the libgcrypt source tarball
to systemd m4 directory.

Libgcrypt uses a custom m4 macro and not pkg-config to check if the
development files are available. Though libgcrypt support is optional in
systemd, this macro should be available whenever autoreconf is used,
otherwise the re-configuration will fail with:

  configure.ac:616: warning: macro 'AM_PATH_LIBGCRYPT' not found in library

The call to autoreconf is required, as it is needed by the patch which
solves the `ln --relative` issue.

As asking the user to install the development package of libgcrypt on
the host machine or adding libgcrypt as a build dependency to systemd is
not acceptable, the required file is added to the m4 directory.

Fixes: http://autobuild.buildroot.net/results/1524d346fa17749e2ae62e063b9cfdd0de95c76a/
Fixes: http://autobuild.buildroot.net/results/10bcd92437eaa27eb61f8281c93efcb53d555e35/
Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/minidlna: Bump version to 1.1.3

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/minidlna: Always depend on host-gettext

Fixes
http://autobuild.buildroot.net/results/7d1/7d10e2c13a1caf61b8bd18b85e5e5af50e5e7c48/

checking for msgfmt... no

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

cosmo: disable package until upstream site is fixed

Fixes:

  http://autobuild.buildroot.org/results/1ca/1ca98e95b106a948f1c0d0153a290ae1d9adb76d/

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

gcc: bump 4.7.x series to version 4.7.4

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

iproute2: bump version

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

faifa: Bump version

A tag has been created, so we should uses it.
This bump of version integrates the patch we had so there is no need for
it anymore.

Also, with this tag, faifa changed it's license to BSD-3c.

Signed-off-by: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

dnsmasq: static linking with lua doesn't need -ldl

Fixes http://autobuild.buildroot.net/results/77e/77e4123a3ddc934efedf4b09adc2436421ee70b3/

liblua only uses libdl when dynamic linking is used, and certain toolchains
(E.G.  bfin) doesn't provide a libdl - So only link against it if it is
needed.

At the same time change it pass the library in LIBS instead of LDFLAGS so it
ends up at the end of the linker cmdline.

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

hdparm: prevent package makefile from stripping the hdparm binary

Packages shouldn't strip executables by themselves, so that Buildroot
controls whether stripping should occur or not. This also fixes the
build on Blackfin FLAT where stripping actually doesn't work because
stripping FLAT binaries is not supported.

Fixes:

  http://autobuild.buildroot.org/results/7d4/7d4e59c96928a06db5091235bf2eacf462ba8a21/

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

infra: indent the Config.in source-ing of packages

This makes sure that a patch adding a package shows in which menu the
package is added.

Before this commit, the patch has something like this:
> diff --git a/package/Config.in b/package/Config.in
> index 7800f23..433312e 100644
> --- a/package/Config.in
> +++ b/package/Config.in
> @@ -864,6 +864,7 @@ source "package/googlefontdirectory/Config.in"
>  source "package/haveged/Config.in"
>  source "package/mcrypt/Config.in"
>  source "package/mobile-broadband-provider-info/Config.in"
> +source "package/mypackage/Config.in"
>  source "package/shared-mime-info/Config.in"
>  source "package/snowball-init/Config.in"
>  source "package/sound-theme-borealis/Config.in"

[> added to avoid git-am recognizing this as the patch]

After this commit, the function marker shows in which menu the new
package was added:
> diff --git a/package/Config.in b/package/Config.in
> index b1111c8..7e6e1a4 100644
> --- a/package/Config.in
> +++ b/package/Config.in
> @@ -864,6 +864,7 @@ menu "Miscellaneous"
>   source "package/haveged/Config.in"
>   source "package/mcrypt/Config.in"
>   source "package/mobile-broadband-provider-info/Config.in"
> + source "package/mypackage/Config.in"
>   source "package/shared-mime-info/Config.in"
>   source "package/snowball-init/Config.in"
>   source "package/sound-theme-borealis/Config.in"

To keep things consistent, this is done for Config.in.host there, even
though we don't have sub-menus there at the moment.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux-headers: bump 3.{4,10,12,14}.x series

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

package/eudev: Bump version to 1.7

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

pinentry: qt4 backend requires the Qt GUI module

Fixes:

  http://autobuild.buildroot.org/results/1d8/1d8a8ff1880549b2e2a24cc175dced1a51245cde/

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

supervisor: add runtime dependency on python-setuptools

supervisor has a runtime dependency on python-setuptools which was not
expressed in its Config.in file. When running supervisor without setuptools,
one gets:

Starting supervisord: Traceback (most recent call last):
  File "/usr/bin/supervisord", line 5, in <module>
    from pkg_resources import load_entry_point
ImportError: No module named pkg_resources

Partially fixes bug #7184 (https://bugs.busybox.net/show_bug.cgi?id=7184)

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

package/mesa3d: depend on udev

checking for LIBUDEV... no
configure: error: libudev-dev required for building DRI

Fixes
http://autobuild.buildroot.net/results/16d/16d12b986051db884eae86e9658e6c4fc0f28885/

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libinput: bump version

And remove upstreamed patches.

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

wireless tools: bump to version 30.pre9

It seems there won't be any new release soon (or ever), so update to the
latest pre version for better functionality (non-ASCII ESSIDs, bigger
scan buffer for example).
Also update the bzero deprecation patch since it missed some instances.

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

package/tvheadend: Add optional dependency to libiconv

To get reproducable builds depend on libiconv if needed.

tvheadend configure either shows

  checking for cc libiconv ...                      fail

or

  checking for cc libiconv ...                      ok

depending on the presence of libiconv, and gets linked

output/host/usr/bin/i586-buildroot-linux-uclibc-readelf -a output/target/usr/bin/tvheadend | grep NEEDED | grep iconv
 0x00000001 (NEEDED)                     Shared library: [libiconv.so.2]

Please note that libiconv is not a hard-dependency.

Tested with minimal defconfig

BR2_TOOLCHAIN_BUILDROOT_LARGEFILE=y
BR2_TOOLCHAIN_BUILDROOT_INET_IPV6=y
BR2_PACKAGE_LIBICONV=y
BR2_PACKAGE_TVHEADEND=y

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libcap-ng: security bump to version 0.7.4

Fixes CVE-2014-3215.

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

e2fsprogs: fix static build against external libblkid

Fixes:
http://autobuild.buildroot.net/results/0159246b52659508a4bbba4126ab27b28a2b2e2c/

Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

radvd: bump to version 1.12

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

p910nd: bump to version 0.97

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

faifa: force GIT_REV to a correct value

This variable is looked up by the Makefile throught a git command line
invocation.
Since we don't keep the '.git' folder in the packages when we clone
them, the command line fail to find anything.

Even worse, it can keep going up the tree until it finds buildroot's
'.git' directory.

Signed-off-by: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

faifa: remove useless CFLAGS -Wno-unused-result

Fixes
http://autobuild.buildroot.net/results/961/9614d13769bfdc908e12e3242e7e07630dd08869/

Signed-off-by: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

ti-gfx: bump to version 5_01_01_01 if EABIhf is used

SDK 5_01_01_01 only support EABIhf so 5_01_00_01 is still used if
EABIhf is not available.

Signed-off-by: Hadrien Boutteville <hadrien.boutteville@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

flann: disable with the problematic NIOS II toolchain

Fixes:

  http://autobuild.buildroot.org/results/062/0620bf8d60aa57949cede0c5b3920ef5446fe98b/

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

znc: not available on !MMU systems, uses fork()

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

znc: add patch to add missing <time.h> inclusions

With certain toolchains, znc fails to build due to missing <time.h>
inclusions. This commit fixes that by adding a patch that adds those
missing inclusions.

Fixes:

  http://autobuild.buildroot.net/results/86e/86e5e76a6ccc1396cd84107af63cdaa0ce1dfbee/

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

dhcp: bump to version 4.1-ESV-R9

Also install the SysV initscripts from the proper hook.
Since the scripts check for the proper files to be present there's no
need to exclude them according to options since they won't try to start
anything.

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

perl-cross: bump to version 0.8.5 and move to github

The perl-cross tarball allows to patch the Perl distribution (in a
_POST_EXTRACT_HOOKS).  The github helper gives a copy of the project which
allows to build this tarball.  So, the github helper can not be used here.

[Peter: mention the change to github]
Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

pkg-perl: fix for Perl XS packages configured by Makefile.PL

ExtUtils::MakeMaker adds all the header files used by the perl as
dependencies to the generated Makefile. This means that the generated
Makefile will depend on the system's header files.

Usually this is not a problem, because when building the target package,
these header files will indeed be found in $(STAGING_DIR). However, some
distro's add an extra header file to the system's perl. This header is
also included in the generated Makefile, which makes the build fail
because it doesn't exist in $(STAGING_DIR).

As a work-around, explicitly create this header file in $(STAGING_DIR).
It doesn't hurt to create it even if the system perl doesn't need it.

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

xterm: bump to version 306

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

bind: security bump to version 9.9.5

Fixes CVE-2014-0591 (a crafted query against an NSEC3-signed zone can
crash BIND).
The 9.9.x series is the new ESV vesion, 9.6.x has been retired.
Also cleanup the initscript while at it.

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

kexec: bump to 2.0.7

Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux-headers: bump 3.2.x series

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

libcgicc: bump to version 3.2.15

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

package/tvheadend: Bump version

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

package/tvheadend: Properly disable internal fetch of dvb-data

Fixes
http://autobuild.buildroot.net/results/a8a/a8a8e39d57b14f1a59c8e6322b511c761442ee3c/
http://autobuild.buildroot.net/results/227/2278a078af5cc2e9f36661ad11e70b7692481266/

Instead of patching the buildsystem just use the included option --disable-
dvbscan, therefore remove tvheadend-001-no-auto-download-dvbscans.patch

This should also point tvheadend to the right place for the dvb data according
to src/input/mpegts/scanfile.c:

  #if ENABLE_DVBSCAN
    path = "data/dvb-scan";
  #else
    path = "/usr/share/dvb";
  #endif

and the resulting binary:

strings tvheadend | grep "dvb-scan\|share\/dvb"
/usr/share/dvb

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

gettext: optimize build time

This commit significantly reduces the build time of host-gettext and
gettext, by using the capacity of gettext to handle build things in a
certain subdirectory:

 - For the host variant of gettext, we only need the gettext-tools,
   available in the directory of the same name in the gettext sources.

 - For the target variant of gettext, we only need the gettext library
   libintl, available in the gettext-runtime directory in the gettext
   sources.

So by using appropriate values of GETTEXT_SUBDIR and
HOST_GETTEXT_SUBDIR, we only build what's necessary. Moreover, by
manually patching gettext-tools/Makefile.in and
gettext-runtime/Makefile.in, we make sure to not build and install
things like examples, documentation and so on.

In addition to this, these changes avoid the need to autoreconfigure
the gettext package, which was particularly long.

Thanks to these changes, the build time of gettext goes from 1 minutes
and 37 seconds to just 24 seconds, and the build of host-gettext goes
from 2 minutes and 18 seconds to 1 minute and 13 seconds.

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

gettext: remove support for gettext-tools on target

This commit removes the BR2_PACKAGE_GETTEXT_TOOLS option, which could
be used to install gettext tools on the target. This is not needed,
because Buildroot is not designed to provide a full development
environment on the target, and gettext translation files should be
processed on the build machine, using the host gettext tools.

Remove this option will allow to optimize the build time of gettext on
the target, by only building the gettext runtime libraries.

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

minidlna: do not select BR2_PACKAGE_GETTEXT_TOOLS

Commit 9ff640ae30ca588228030faf2c6fd13af292cb66 ("minidlna: needs
gettext and host-gettext tools if locale is enabled") modified the
minidlna package to fix gettext related issues. As part of this patch,
a select of BR2_PACKAGE_GETTEXT_TOOLS was added, which according to
the commit title is used to have host-gettext tools installed.

However, this is not what this option is about: this option is about
having gettext tools installed on the target.

Since this is not what minidlna needs, and we anyway plan to remove
this BR2_PACKAGE_GETTEXT_TOOLS option, this commit removes this
incorrect select statement.

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

toolchain-external: add ADI Blackfin 2014R1 toolchain, remove 2012R1

This commit adds support for two ADI Blackfin toolchains: the 'stable'
2014R1 based on gcc 4.3, and the 'experimental' 2014R1 based on gcc
4.5.

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

toolchain-external: add CodeSourcery MIPS 2014.05, remove 2012.09

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

toolchain-external: bump Linaro AArch64 toolchain to 2014.05

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

toolchain-external: bump Linaro ARMeb toolchain to 2014.04

For some reason, there is no ARMeb toolchain available in the 2014.05
Linaro release.

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

toolchain-external: bump Linaro ARM to 2014.05

The major changes are: switch to gcc 4.9 instead of 4.8, and switch to
glibc 2.19 instead of glibc 2.18.

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

toolchain-external: support only one Linaro AArch64 toolchain at a time

Linaro toolchains are released so frequently (every month) that it
doesn't make much sense to support 3 consecutive versions. So, like we
do for ARM big-endian, let's support only one version at a time.

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

toolchain-external: support only one Linaro ARM toolchain at a time

Linaro toolchains are released so frequently (every month) that it
doesn't make much sense to support 3 consecutive versions. So, like we
do for ARM big-endian, let's support only one version at a time.

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

toolchain-external: do not pass TOOLCHAIN_EXTERNAL_CFLAGS for kernel headers check

When getting the sysroot used for the kernel headers version check,
passing TOOLCHAIN_EXTERNAL_CFLAGS causes a problem when used with
multilib toolchains, where only the main sysroot has the header files,
and the other sysroots only have the libraries.

Since the kernel headers version used is normally the same for all
sysroots, this commit solves this problem by removing the
TOOLCHAIN_EXTERNAL_CFLAGS argument when calling
toolchain_find_sysroot, so that it returns the main sysroot, in which
<linux/version.h> can be found for the kernel headers version check.

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

toolchain-external: apply kernel headers check to non-custom toolchains

Since the introduction of the kernel headers Config.in options, the
external toolchain logic had a check for custom external toolchains to
verify that the kernel headers version entered by the user matches the
one of the toolchain. However, this check was not made for non-custom
external toolchains (i.e the built-in profiles, such as Linaro,
CodeSourcery and al.), making the assumption that the Buildroot
developers will do the right selection.

However, it is quite nice when bumping external toolchains to have
this automatic kernel headers version check, to ensure we select the
appropriate kernel headers version.

Therefore, this commit makes the kernel headers version check
applicable to non-custom external toolchains.

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

pinentry: new package

[Thomas:
 - fix license version: the license is GPLv2+, not GPLv2
 - make sure the ncurses backend is selected if none of the other
   backends are selected.
 - add dependency on libiconv of the curses and gtk2 backends, needed
   in !locale configurations.]

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Reviewed-by: Markos Chandras <Markos.Chandras@imgtec.com>
Reviewed-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

lsof: do not remove WIDECHARINCL definition

WIDECHARINCL is enabled by HASWIDECHAR, so removing its definition if
BR2_USE_WCHAR is not set is useless.

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

libksba: new package

[Thomas: fix license informations.]

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

linux: bump default to version 3.15

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

linux-headers: add 3.15.x series

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

toolchain: add 3.15 choice for headers

[Thomas: fix BR2_TOOLCHAIN_HEADERS_AT_LEAST_3_15 to select
BR2_TOOLCHAIN_HEADERS_AT_LEAST_3_14 and not itself.]

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

package/libbsd: needs an (e)glibc toolchain

libbsd needs support for .init_array and checks for a
glibc >= 2.4 since .init_array was introduced at around
that time.

uClibc claims to be a glibc-compatible toolchain, but it
only impersonates a glibc-2.2.

Just disable libbsd on uClibc.

Fixes:
    http://autobuild.buildroot.net/results/e94/e949d8fabeeecc74bd1c324c516e0b4938c99dbc/
    http://autobuild.buildroot.net/results/d3e/d3e1b70fb91571efacbe32af2cd12d055508f5ac/
    http://autobuild.buildroot.net/results/b19/b19d24dbf9d05d86d839349695da45d548705b25/
    [...]

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

linux-zigbee: add missing comment for thread dependency

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

sispmctl: new package

[Thomas:
 - Add missing select of BR2_PACKAGE_LIBUSB, since
   BR2_PACKAGE_LIBUSB_COMPAT depends on it
 - Add missing thread dependency, since libusb requires threads.
 - Add missing wide-char dependency. Without wide char support,
   sispmctl doesn't build.
 - Rewrap the Config.in help text, and remove trailing whitespaces.
 - License is GPLv2+, not GPLv2.
 - Remove --disable-dependency-tracking from <pkg>_CONF_OPT. That's a
   global, standard, autoconf option, and there's no reason to pass it
   at the per-package level.]

Signed-off-by: Phil Eichinger <phil@zankapfel.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Revert "package/weston: libxkbcommon is optional"

This reverts commit cf1c2eb19d3eeddca04013cfa7b2107530abf054.

xkbcommon is still needed for the clients. There's no point in disabling
the clients, or weston is unusable (as packaged in Buildroot.)

Fixes:

  http://autobuild.buildroot.org/results/4e9/4e996c65f5b33d4518b0596d9c7076083d491a52/

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

linux-zigbee: new package

linux-zigbee is a mix of utility tools that implements a 802.15.4 stack
along with a usefull library for any program that would need to use this
stack.

Note that the name is mislead, for licensing reason, this does *NOT*
implement the ZigBee protocol.

[Thomas:
 - add dependency on threads, propagated from libnl
 - improve the Config.in description by borrowing more text from the
   upstream website
 - fix the prompt of sub-options to be more consistent, and add help
   texts where appropriate
 - fix indentation of the BR2_PACKAGE_LINUX_ZIGBEE_TESTS option help
   text
 - add missing dependencies on host-pkgconf, host-flex and host-bison
 - add missing SoB line in the patch disabling test-serial. Maxime is
   a well-known contributor, so I assumed we had his SoB.
 - fix indentation of <pkg>_CONF_OPT.
 - remove <pkg>_INSTALL_STAGING = YES since the package does not
   install any library, and the two headers it installs are available
   through the toolchain kernel headers.
 - add comment to explain <pkg>_AUTORECONF = YES.]

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

configs: bump zedboard to 2014.1

Bump zedboard defconfig to version 2014.1.  This makes it possible to use
uboot-spl instead of the xilinx specific fsbl.  However, the result is kinda
hacky due to the licensing state of ps7_init.{c,h} needed for building uboot
spl.  Directions for building a proper boot.bin using the new methodolgy is
included in the readme.txt.

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

ruby: remove rubygems

It's currently in a non-working state since it requires a couple of
extensions that don't build at the moment (they try to execute tests in
configure) and also requires a target compiler.
So remove it to avoid false expectations and reclaim target space back
of about 1.5 MiB.

[Thomas: slightly reword comments.]

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

ruby: add gdbm support

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

ruby: add psych/libyaml support

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

cloog: remove INSTALL_STAGING = YES

The package is host-only, so there's no point in specifying
INSTALL_STAGING = YES.

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

isl: remove INSTALL_STAGING = YES

The package is host-only, so there's no point in specifying
INSTALL_STAGING = YES.

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

cloog: fix license details

License is in fact LGPLv2.1+, and there is no LICENSE file around.

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

gcc: add a BR2_GCC_ENABLE_GRAPHITE option

The gcc graphite optimisations such as loop-interchange, blocking
and loop-flattening, also known as graphite are an optional feature of
gcc that is very well supported since about gcc version 4.5.

This patch adds support for graphite for the toolchain as an optional
flag for versions 4.8 onwards as an optional flag, that is disabled by
default.

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

cloog: new package

CLooG is a free software and library to generate code for scanning
Z-polyhedra.

cloog is needed for the optional graphite optimisations that are
supported by gcc since version 4.5. Therefore this package is required
for the toolchain to support graphite.

Graphite optimisations primarily involve loop blocking flattening and
interchage so are probably of mimimal use in an embedded system where
small sizes are favoured.

cloog depends on isl.

[Thomas:
 - Add patch to add missing CMake related files to the release
   tarball, preventing the build from succeeding.
 - bump to 0.18.2
 - disable libtool patch, which doesn't apply]

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

isl: new package

isl is a library for manipulating sets and relations of integer points
bounded by linear constraints.

isl is needed for the optional graphite optimisations that are supported
by gcc since version 4.5. Therefore this package is required for the
toolchain to support graphite.

Graphite optimisations primarily involve loop blocking flattening and
interchage so are probably of mimimal use in an embedded system where
small sizes are favoured.

[Thomas:
 - bump to 0.12.2 (cannot use 0.13, incompatible with cloog 0.18.2,
   comment added about this)
 - use .bz2 tarball
 - disable libtool patch, which doesn't apply]

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

filesystems: also chown symlinks

Currently, the symlinks in the generated filesystems will have the
UID of the user running the build, because 'chown' does not change
the ownership of symlinks, by default.

Although the implications are limited, some may not want that UID
to leak in the generated filesystems.

So, use 'chown -h' so even symlinks get properly chowned.

Reported-by: Angelo Dureghello <angelo@barix.com>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas De Schampheleire <patrickdepinguin@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

uclibc: add patches to fix MIPS __SIGRTMAX and SIG127

Add two uClibc 0.9.33.2 patches which fix issues relating to high signal
number handling on MIPS.

The first patch (0056) fixes _NSIG (and as a result __SIGRTMAX) to match
glibc. This fixes GDB on MIPS with uClibc, which cannot handle
__SIGRTMAX == 128 and emits the error:
GDB bug: target.c (gdb_signal_from_host): unrecognized real-time signal
This patch is from uClibc commit 2da958760f79 (MIPS: set _NSIG to
128, not 129. This matches glibc.) and applies without conflicts.

The second patch (0057) fixes the wait status macros to correctly
interpret status 0x007f on MIPS (other arches don't have signal 127).
This patch is from uClibc commit 4a96b9486871 (bits/waitstatus.h:
correctly interpret status 0x007f on MIPS) and applies without
conflicts.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

uclibc: add patch to fix arch specific eventfd defs

Add uClibc 0.9.33.2 patch to fix arch specific eventfd definitions,
particularly EFD_NONBLOCK. The definition in sys/eventfd.h was 04000
(0x800), however MIPS uses 0200 (0x80). This resulted in QEMU built for
MIPS hanging at various points until input is received due to a blocking
poll of stdin.

The patch is a backport of uClibc commit fd355bc1dbcb (eventfd.h: Use
new "bits/" scheme for arch-specific flags). The only conflict was minor
in Makefile.in.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/xbmc-pvr-addons: Bump version, use gotham branch

Suggested by Yann:
http://lists.busybox.net/pipermail/buildroot/2014-June/098698.html

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

package/sqlite: Bump version to 3.8.5

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

sunxi-mali: use only -I${includedir} as pkg-config Cflags

Using only -I${includedir} in common practice.

Settings only -I${includedir}/GLES2 breaks Qt5's CMake files,
because at configure time CMake only searchs the paths reported
by pkg-config, and not even /usr/include is used as default.

Even though pkg-config strips out standard include path, that's not
the case with pkgconf (which we are using) in cross-compilation,
which correctly reports the /usr/include dir prefixed with the
sysroot.

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

ti-gfx: use only -I${includedir} as pkg-config Cflags

Using only -I${includedir} in common practice.

Settings only -I${includedir}/GLES2 breaks Qt5's CMake files,
because at configure time CMake only searchs the paths reported
by pkg-config, and not even /usr/include is used as default.

Even though pkg-config strips out standard include path, that's not
the case with pkgconf (which we are using) in cross-compilation,
which correctly reports the /usr/include dir prefixed with the
sysroot.

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

package infra: remove duplicates in dependencies list

Currently, we just use what a package declares as its dependencies.

But some packages may declare the same depdency more than once. For
example, php has two options to add SQL support: 'mysql' or 'mysqli',
which are not exclusive. So, php.mk has mysql twice as a dependency.

Although that does not cause any grievance for make, we end up generating
dependency graphs where this duplicate dependency is visible.

Add an intermediary variable which contains the $(sort)-ed list of the
dependencies, thus eliminating any duplicates.

This has the side effect of also sorting the list, which is probably
good for reproducibility anyway.

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

package/x11r7/xlib_libXfont: Bump version to 1.4.8

Fixes CVE-2014-0209, CVE-2014-0210 & CVE-2014-0211
http://lists.x.org/archives/xorg-announce/2014-May/002431.html

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

package/trace-cmd: bump to version 2.3.2

The first patch "use-pkg-config-instead-of-python-config" has been applied
upstream.

The flag _GNU_SOURCE is now always defined in Makefile, so it can be removed
from trace-cmd.mk

CPPFLAGS are appended to CFLAGS and are used to add extra flags, but the flag
_LARGEFILE64_SOURCE in TARGET_CPPFLAGS is already defined in source files,
which causes a build error. As for CFLAGS, we fix this by filtering out our
definition of _LARGEFILE64_SOURCE from the CPPFLAGS before passing them to the
trace-cmd Makefile.

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

faifa: new package

faifa is a library and a cli used to manage PLC hardware that use
Intellon chipset.  Almost everybody use theses chips nowadays.

[Thomas: license is GPLv2+, not GPLv2. Fix indentation. Add dependency
on host-autoconf and a comment to explain what's going on. Fix
indentation of target/staging installation commands. Mark the package
as not available for static library builds as it always build a shared
library. Rewrap Config.in help text.]

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

packages: replace command install by $(INSTALL)

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

sshpass: new package

[Thomas: add MMU dependency, fix license to GPLv2+.]

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

wpa_supplicant: fix BR2_PACKAGE_WPA_SUPPLICANT_DEBUG_SYSLOG option

Signed-off-by: Sven Neumann <neumann@teufel.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

perl: removes useless files in target tree

$(TARGET_DIR)/usr/lib/perl5/$(PERL_VERSION)/$(PERL_ARCHNAME)/CORE contains include files.
*.bs & .packlist files come with perl or perl/cpan packages.

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

docs/manual: add the virtual packages list

Add the list of virtual packages as an appendix to the manual.

Also reference this list from appropriate locations elsewhere in
the manual:

  - in section 7.2.2. "Config.in file", after the existing explanations
    on dependencies on target and toolchain options, on a linux kernel,
    and on udev /dev management,

  - in section 7.2.10. "Infrastructure for virtual packages", in the
    provider Config.in and .mk explanations, to have the list of existing
    symbols to select (in Config.in) and packages to provide (in .mk).

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Samuel Martin <s.martin49@gmail.com>
Acked-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

support/scripts: generate a list of virtual packages

Generate an asciidoc table that can be included in the manual, that
lists the existing virtual packages, the corresponding symbols, and
their providers (and sub-options thereof).

The core of this change is the addition of a new formatter for virtual
packages. This formatter is a bit tricky, as it has to catter for a
bunch of corner cases:
  - provider is not a package, but is sub-options of a package
  - such a sub-option may be itself 'select'-ed by one or more
    other sub-options
  - legacy packages should not be considered as a provider

Those cases are real:
  - sub-options of mesa3d provide EGL or GLES
  - selected sub-options of mesa3d provide GL
  - udev is a legacy package, but it provides udev

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

support/scripts: introduce a symbol formatter to generate package lists

Currently, we can generate two different tables of packages:
  - a single-column table with the symbols' prompts,
  - a two-column table with the symbols' prompts and locations in the
    menuconfig.

For virtual packages, this is not enough, since we will have to display
more columns, with different content:
  - the virtual package name (but such symbols do not have a prompt)
  - the symbol name
  - the providers for the virtual package

So, instead of having a single function that knows how to generate any
table, introduce a formatter function that is passed as argument to,
and called by format_asciidoc_table(). Such formatter functions are
responsible for providing:
  - the layout of the table (number of columns, column arrangement),
  - the formatted header line,
  - a formatted line for a symbol.

What the formatter should ouput depends on its arguments:
  - if none are passed, the layout is returned,
  - if the header label is passed, it returns the formatted header line,
  - otherwise, it returns the formatted line for a symbol.

Two formatter functions are introduced in this changeset, to replace the
current 'sub_menu' feature:
  - _format_symbol_prompt() to display a one-column table with only the
    symbols' prompts,
  - _format_symbol_prompt_location() to display a two-column table with
    the symbols' prompts and locations.

This will help us to later introduce a new formatter to generate a table
for virtual packages.

[Thanks to Samuel for his pythonistic help!]

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

support/scripts: who's responsible to decide what is a package

When generating the package lists, the responsibility to decide what is
actually a package symbol is currently split between the _is_package(),
the get_symbol_subset() and the format_asciidoc_table() functions.

The two latter functions check that an item is really a symbol, and that
is has a prompt.

While this is currently correct for real packages, this will no longer
be the case when we also generate a list of virtual packages, since they
do not have a prompt.

Move the responsibility to verify that a symbol is indeed a package symbol
to _is_package(), so it's all in one place, and makes it easier to change
for virtual packages.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

support/scripts: do not display virtual packages in generated lists

If a package has both a 'real' and a 'virtual' definition, consider it
is a virtual package and do not display it in the generated package list.

This is the case for jpeg and cryptodev, that are virtual packages, but
also real (but empty) packages used to provide a prompt to enable/disable
a choice to select an implementation. In this case, we do not want to
list the virtual packages, but only their implementations.

So, consider packages that are both real and virtual as virtual packages.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Samuel Martin <s.martin49@gmail.com>
Acked-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

support/scripts: prepare handling virtual packages in generated lists

Prepare to tell apart real packages from virtual packages.

Currently, the code implicitly recognises only real packages, and
discards virtual packages, because of the heuristic used to recognise
whether a symbol is a package:

  - for real package:
    - symbols  : BR2_PACKAGE_FOO
    - .mk files: foo.mk
  - for virtual packages:
    - symbols  : BR2_PACKAGE_HAS_FOO
    - .mk files: foo.mk

The current heuristic is to check for each symbol if a corresponding .mk
file exists, by stripping 'BR2_PACKAGE_' from the beginning of the symbol,
converting the result to lowercase, and checking if a .mk file exists.

So, as a side effect, it completely misses the virtual packages [*], which
is pretty nice since we get a list with only real packages that the user
can indeed select and see in the menuconfig.

[*] Except for 'cryptodev' and 'jpeg' which are both virtual packages and
normal packages. Except they are not normal packages, they are used to
display a choice of the implementation to use. This case will be fixed in
follow-up patches.

Since we'll soon need to also output the table of virtual packages, we
need to teach the _is_package() function to recognise them as well.

This patch is the first step into that direction: it introduces a new
function _is_real_package() that is just a wrapper to _is_package(), which
gains a new parameter, being the type of packages to filter on.

No behavioural change is made in this patch, it is just a preparatory
patch.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

support/scripts: document args to _is_package() in gen-manual-lists

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Samuel Martin <s.martin49@gmail.com>
Acked-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

support/scripts: prepare expanding the packages lists

Move to a function the code generating the package name from a
symbol's name, to avoid code duplication.

This is not used currently, but will be in a subsequent patch.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Samuel Martin <s.martin49@gmail.com>
Acked-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/mesa3d: rename the prompt

Rename the prompt for mesa3d so that it matches what we usually do
for prompts of packages: just name the package in lower case.

This will help generate nicer tables in the generated packages list,
both in existing tables, and in the upcoming table of virtual packages.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/jpeg: make it behave more like other virtual packages

Currently, the virtual package jpeg is a special virtual package,
as it offers a choice for its implementation, rather than letting the
user enable them manually.

In so doing, it defines its _HAS option as a 'def_bool y' rather than
letting each implementation 'select' it.

Since we are going to generate a list of virtual packages and their
providers, this defeats the heuristic used to find providers.

Coming with an alternate heuristic that also matches the jpeg
package is quite complex, so better and easier to make it look more
like any other virtual package.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/cryptodev: make it behave more like other virtual packages

Currently, the virtual package cryptodev is a special virtual package,
as it offers a choice for its implementation, rather than letting the
user enable them manually.

In so doing, it defines its _HAS option as a 'def_bool y' rather than
letting each implementation 'select' it.

Since we are going to generate a list of virtual packages and their
providers, this defeats the heuristic used to find providers.

Coming with an alternate heuristic that also matches the cryptodev
package is quite complex, so better and easier to make it look more
like any other virtual package.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

manual: document the new graph-depends options

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Reviewed-by: Thomas De Schampheleire <patrickdepinguin@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

graph-depends: add option to pass arbitrary dot options

Kids nowaday seem to prefer a left-to-right drawing rather than the
more conventional and historical top-down drawing.

Rather than multiply the number of environment variables, just add
a single one where the user can pass arbitrary dot options, such as:

    make BR2_GRAPH_DOT_OPTS=-Grankdir=LR graph-depends

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>