package/libtirpc: add hash file

Should fix autobuild failures due to SourceForce sending garbage from
time to time:
    http://autobuild.buildroot.net/results/acc/acc086e04da324621c715962585dbdd80e04b74e/
    http://autobuild.buildroot.net/results/acc/acc086e04da324621c715962585dbdd80e04b74e/
    http://autobuild.buildroot.net/results/ca0/ca084460c06f5b69ad35080cd347554a2c53f2a8/

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/opkg: disable tests and manpages

Changeset f1d3e09 (Build shared libraries only as the default) broke the
opkg package.

opkg wants to link its testsuite statically with the internal libopkg.
But in a shared-only build, there is no static library.

Since for Buildroot we do not care about tests, just disable them.

Fixes numerous build failures;
    http://autobuild.buildroot.net/results/ddf/ddf4b38c1df73b00b76be71db3f1eb8b1d932c55/
    http://autobuild.buildroot.net/results/c1c/c1c9a21b6583c9ef3ddd32681bb13cf91a218e8c/
    http://autobuild.buildroot.net/results/284/284bf7d6f9cebf8633143ca1ce15154de941805f/
    ...

Disable manpages at the same time.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

liblo: add hash file

Fixes:
http://autobuild.buildroot.net/results/579/579df46b1fdb46fa5ec0fd28324cf28b7c06b5bb/
http://autobuild.buildroot.net/results/f20/f2023b000b3ef8d459bba04a29f58be88246cb46/
http://autobuild.buildroot.net/results/e2a/e2acf1f0355df709836e92e4382019b6e8f77978/
http://autobuild.buildroot.net/results/5b9/5b9c2a678a0950def2aeaf7e5809386c5923e974/

And many more.

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

ebtables: add hash file

Should clean up download failures like:
http://autobuild.buildroot.net/results/3c6/3c6b15a3d7ba84b7b3c257089569e2f615ec81b2/
http://autobuild.buildroot.net/results/4ee/4ee41e1616184687f56c9e3095885549ac4767d2/
http://autobuild.buildroot.net/results/9c6/9c623e3d9fde75b3dc1724909f6dc9738b12a207/

and more.

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

pcre: add a patch fixing CVE-2014-8964

Patch taken from the Debian package.

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

pcre: rename patches to new convention

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

package/berkeleydb: fix shared-only build

Changeset f1d3e09 (Build shared libraries only as the default) broke the
berkeleydb package.

For some reason, we had a very strange post-configure action in bdb:
        $(SED) 's/\.lo/.o/g' $(@D)/build_unix/Makefile

This means that it is using non-libtool object files. This breaks
building shared-only.

This code has been present almost the whole time we've had a berkeleydb
package, that is:
  - November 2003: use a sed expression instead of perl
  - March 2003   : use a perl program
  - August 2002  : introduce berkeleydb

Getting rid of this sed expression fixes the build, and allows building
the three new static/shared combinations: static only, shared only,
both static and shared.

Fixes numerous build failures:
    http://autobuild.buildroot.net/results/b41/b41e13ec6c13830667770b3731f8990062f202bd/
    http://autobuild.buildroot.net/results/925/925433ef505b190aa5ac83c23e9359afe4b21829/
    http://autobuild.buildroot.net/results/9af/9afa0dd835af079411a2ea5d94e5e4db7c993111/
    ...

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

erlang: bump to version 17.4

Erlang/OTP 17.4 is a service release on the 17 track with mostly bug
fixes, but it does contain a number of new features and characteristics
improvements as well.

Some highlights of the release are:

  - eldap: Nearly all TCP options are possible to give in the
    eldap:open/2 call.
  - ssh: Added API functions ptty_alloc/3 and ptty_alloc/4, to allocate a
    pseudo tty.
  - ssl: Handle servers that may send an empty SNI extension to the client.

Signed-off-by: Frank Hunleth <fhunleth@troodon-software.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/nettle: fix shared-only build

Changeset f1d3e09 (Build shared libraries only as the default) broke the
nettle package.

nettle really wants to statically link its tools, examples and testsuite
against its own libraries. But for a shared-only build, static libraries
are not even built, so the build breaks when trying to link any of them.

Fix that in two ways:

  - the proper way, by using whatever libraries are available, with a
    preference toward using shared libraries;

  - disabling both examples and testsuite since we do not need them.

Fixes numerous build failures:
    http://autobuild.buildroot.net/results/e93/e93da743c5b84b1bb6274aeee3fe3b52962c560c/
    http://autobuild.buildroot.net/results/51d/51d0138ff39c45603315dcedf1f654c18604e1a8/
    http://autobuild.buildroot.net/results/521/521d979658e402103f1a18fcfeab3f9705f0c000/
    ...

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

upmpdcli: use start() / stop() in init script like elsewhere

Other init scripts in Buildroot use start()/stop() instead of
do_start()/do_stop(), so change it here as well for consistency

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

transmission: use start() / stop() in init script like elsewhere

Other init scripts in Buildroot use start()/stop() instead of
do_start()/do_stop(), so change it here as well for consistency

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

rpi-userland: use start() / stop() in init script like elsewhere

Other init scripts in Buildroot use start()/stop() instead of
do_start()/do_stop(), so change it here as well for consistency.

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

Build shared libraries only as the default

Now that we have more options in terms of static/shared libraries,
switch from the existing default of building both shared and static
libraries to building shared libraries only (of course only on
platforms that support shared libraries).

Building both shared and static takes time (since the shared objects
must be built with -fPIC, while static objects are generally built
without, as -fPIC has some performance impact) and consumes a little
bit more disk space.

For example, a static+shared build of libglib2 takes 1 minutes and 59
seconds, with a final build directory of 96 MB. A shared-only build of
libglib2 takes only 1 minutes and 31 seconds (almost a 25% reduction
of the build time), and the final build directory weights 89 MB (a
reduction of almost 8%).

So, switching to a shared library only build brings some useful build
time and build size benefits.

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

ncurses: better handling for shared/static library

Now that we have clear options for the three cases of shared only,
static only and shared+static, let's use them in ncurses to pass the
appropriate --{with,without}-{shared,normal} options.

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

Turn the static lib option into a choice with more options

This commit turns the single static option into a choice, which offers
various possibilities:

 1. Build and use static libraries only;
 2. Build both shared and static libraries, but use shared libraries;
 3. Build and use shared libraries only.

On most platforms, (2) is currently the default, and kept as the
default in this commit. Of course, on certain platforms (Blackfin,
m68k), only option (1) will be available.

In addition to the introduction of the Config.in options, this commit
also:

 * Removes the 'select BR2_STATIC_LIBS' from 'BR2_BINFMT_FLAT', since
   with the use of a choice, we are guaranteed that BR2_STATIC_LIBS
   will be selected when the binary format is BR2_BINFMT_FLAT, since
   BR2_STATIC_LIBS will be the only possible solution in the choice.

 * Changes package/Makefile.in to use the proper
   --{enable,disable}-{shared,static} options for autotools packages.

[Thomas: remove useless empty newline right after 'choice'. Noticed by
Yann E. Morin.]

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

triggerhappy: new package

[Thomas:
  - add BR2_USE_MMU dependency, since fork() is used
  - rename do_start() and do_stop() to just start() and stop(), as we
    do in most init scripts in Buildroot.]

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

pkg-download: check hashes for locally cached files

In some cases, upstream just update their releases in-place, without
renaming them. When that package is updated in Buildroot, a new hash to
match the new upstream release is included in the corresponding .hash
file.

As a consequence, users who previously downloaded that package's tarball
with an older version of Buildroot, will get stuck with an old archive
for that package, and after updating their Buildroot copy, will be greeted
with a failed download, due to the local file not matching the new
hashes.

Also, an upstream would sometime serve us HTML garbage instead of the
actual tarball we requested, like SourceForge does from time for as-yet
unknown reasons.

So, to avoid this situation, check the hashes prior to doing the
download. If the hashes match, consider the locally cached file genuine,
and do not download it. However, if the locally cached file does not
match the known hashes we have for it, it is promptly removed, and a
download is re-attempted.

Note: this does not add any overhead compared to the previous situation,
because we were already checking hashes of locally cached files. It just
changes the order in which we do the checks. For the records, here is the
overhead of hashing a 231MiB file (qt-everywhere-opensource-src-4.8.6.tar.gz)
on a core-i5 @2.5GHz:

            cache-cold  cache-hot
    sha1      1.914s      0.762s
    sha256    2.109s      1.270s

But again, this overhead already existed before this patch.

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

pkg-download: verify the hashes from the download wrapper

Instead of repeating the check in our download rules, delegate the check
of the hashes to the download wrapper.

This needs three different changes:

  - add a new argument to the download wrapper, that is the full path to
    the hash file; if the hash file does not exist, that does not change
    the current behaviour, as the existence of the hash file is checked
    for in the check-hash script;

  - add a third argument to the check-hash script, to be the basename of
    the file to check; this is required because we no longer check the
    final file with the final filename, but an intermediate file with a
    temporary filename;

  - do the actual call to the check-hash script from within the download
    wrapper.

This further paves the way to doing pre-download checks of the hashes
for the locally cached files.

Note: this patch removes the check for hashes for already downloaded
files, since the wrapper script exits early. The behaviour to check
localy cached files will be restored and enhanced in the following
patch.

[Thomas: fix minor typo in comment.]

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

pkg-download: check for already downloaded file in the download wrapper

Instead of repeating the same test again and again in all our download
rules, just delegate the check for an already downloaded file to the
download wrapper.

This clears up the path for doing the hash checks on a cached file
before the download.

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

support/download: add option parsing to the download wrapper

Instead of relying on argument ordering, use actual options in the
download wrapper.

Download backends (bzr, cp, hg...) are left as-is, because it does not
make sense to complexify them, since they are almost very trivial shell
scripts, and adding option parsing would be really overkill.

This commit also renames the script to dl-wrapper so it looks better in
the traces, and it is not confused with another wrapper.

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>

ncurses: speed up host build by disabling static library building

For the host variant of packages, we normally only build the shared
libraries. However, ncurses uses non-standard options to select
between shared/static and therefore the host variant was building both
of them, even though the static libraries were unused.

By passing --without-normal, we disable the build of static
libraries. It saves a bit of disk space, and on my laptop,
host-ncurses takes 26 seconds to build instead of 40 seconds.

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

arch: remove superfluous arch dependencies on BR2_BINFMT_FLAT_*

As noted by Yann E. Morin, those dependencies are unneeded, since FLAT
can anyway only be used on m68k and Blackfin.

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

arch: introduce BR2_BINFMT_SUPPORTS_SHARED

In preparation for the refactoring of the static/shared library
support, we add a BR2_BINFMT_SUPPORTS_SHARED hidden option that binary
formats supporting shared libraries should select.

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

arch: ensure BR2_BINFMT_ELF is really used

Currently, the binary format choice is only shown for the Blackfin and
m68k architectures, since we assume that all other architectures are
using the ELF binary format. However, due to this, the BR2_BINFMT_ELF
symbol is in fact not set to 'y' for those architectures that use the
ELF format.

This will be causing problems for the refactoring of the static/shared
library support, as we will need to know if the binary format supports
shared libraries or not.

Therefore, we simply make the choice visible on all architectures,
even if it means that on many architectures no other choice than ELF
will be available.

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

arch: FDPIC is not supported on m68k

The current binfmt selection in arch/Config.in allows to select FDPIC
on m68k, which is incorrect. This commit fixes that, and makes sure
FDPIC is the default on Blackfin, while FLAT is the default on m68k.

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

Rename BR2_PREFER_STATIC_LIB to BR2_STATIC_LIBS

Since a while, the semantic of BR2_PREFER_STATIC_LIB has been changed
from "prefer static libraries when possible" to "use only static
libraries". The former semantic didn't make much sense, since the user
had absolutely no control/idea of which package would use static
libraries, and which packages would not. Therefore, for quite some
time, we have been starting to enforce that BR2_PREFER_STATIC_LIB
should really build everything with static libraries.

As a consequence, this patch renames BR2_PREFER_STATIC_LIB to
BR2_STATIC_LIBS, and adjust the Config.in option accordingly.

This also helps preparing the addition of other options to select
shared, shared+static or just static.

Note that we have verified that this commit can be reproduced by
simply doing a global rename of BR2_PREFER_STATIC_LIB to
BR2_STATIC_LIBS plus adding BR2_PREFER_STATIC_LIB to Config.in.legacy.

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

package/ffmpeg: bump to version 2.5

Bump version and update hash file.

Signed-off-by: Jörg Krause <jkrause@posteo.de>
Tested-by: Jérôme Pouiller <jezz@sysmic.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

kismet: bump version to Kismet-2014-02-R1

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

kismet: rename patches to follow the new name convention

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

libsvgtiny: add a patch to remove -Werror from Makefile

glibc-2.20 includes some changes to the include/features.h file
introduced by this commit:

  https://sourceware.org/git/?p=glibc.git;a=commit;h=ade40b10ff5fa59a318cf55b9d8414b758e8df78

Those changes make libsvgtiny fail because some warnings are thrown and
the build system is using the -Werror option. We disable this to be able
to build it, or otherwise we will see errors like this one:

GPERF: src/colors.gperf
COMPILE: build-Linux-Linux-release-lib-static/src_colors.c
In file included from
/br/output/host/usr/mipsel-buildroot-linux-gnu/sysroot/usr/include/string.h:25:0,
 from src/colors.gperf:16:
/br/output/host/usr/mipsel-buildroot-linux-gnu/sysroot/usr/include/features.h:148:3:
error: #warning "_BSD_SOURCE and _SVID_SOURCE are deprecated, use
_DEFAULT_SOURCE" [-Werror=cpp]

and this one:

In file included from src/colors.gperf:18:0:
/home/ldap/vriera/work/mips-buildroots/mips32/output/build/libsvgtiny-12121/src/svgtiny_internal.h:71:0:
error: "strndup" redefined [-Werror]

Fixes:
  http://autobuild.buildroot.net/results/3dd/3dd700405055750262738f867eb5aa08531f5781/

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

libsvgtiny: rename patches to follow the new name convention

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

systemd: bump version to v218

Trivial changes:
- Bump version to v218
- Update the hash file

Non-trivial changes:
- Remove upstreamed patch

  The patch was added by this upstream commit:
  http://cgit.freedesktop.org/systemd/systemd/commit/?id=e6c019026b8cfd27a997e6e6ed1349f8f289b7e2

- Remove non-existent configure options

  --with-firmware-path option was removed by this upstream commit:
  http://cgit.freedesktop.org/systemd/systemd/commit/?id=be2ea723b1d023b3d385d3b791ee4607cbfb20ca

  --disable-tcpwrap option was removed by this upstream commit:
  http://cgit.freedesktop.org/systemd/systemd/commit/?id=7f8aa67131cfc03ddcbd31c0420754864fc122f0

- Select a required dependency

  libmount dependency was added by this upstream commit:
  http://cgit.freedesktop.org/systemd/systemd/commit/?id=8d3ae2bd4c9bf9fc2e57f7b3776325a1c750ca30

  The error for not satisfying this required dependency looks like this:

  checking for MOUNT... no
  configure: error: *** libmount support required but libraries not
    found

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

file: bump to version 5.21

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

gnutls: bump to version 3.2.21

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

package/fdk-aac: bump to version 0.1.3

Signed-off-by: Jörg Krause <jkrause@posteo.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ipset: bump to version 6.24

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

package/libuv: bump to version 1.0.2

libuv has moved to github.com/libuv/libuv

[Thomas: fix Config.in help text, as noticed by Baruch.]

Signed-off-by: Jörg Krause <jkrause@posteo.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Add option for paranoid unsafe path checking

This commit adds a Config.in option to the "Build options" submenu to
enable paranoid checking of unsafe paths. This mechanism is added as
an option so that when we'll enable it in the autobuilders, people
trying to reproduce the build failures will be able to do so by just
downloading the configuration file. If instead we were leaving this
feature as an environment variable, everyone would have to remember to
pass this environment variable to reproduce build issues. And certain
build issues triggered by paranoid unsafe patch checking may not be
visible in the build output, for example when they happen during the
execution of configure scripts.

Since this option is fairly advanced, a new submenu inside "Build
options" is created, for Advanced options.

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

binutils: enable poison system directories option

This commit enables the poison system directories option, which is now
available thanks to the binutils patches that have been added.

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

gcc: enable poison system directories option

This commit enables the poison system directories option, which is now
available thanks to the gcc patches that have been added.

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

gcc/4.7: add patch to warn about unsafe header paths

This commit adds a patch to gcc borrowed from CodeSourcery/Yocto that
warns about unsafe include paths (i.e /usr/include,
/usr/local/include, etc.). The patch was adapted to gcc 4.7.4, and
modified to support the BR_COMPILER_PARANOID_UNSAFE_PATH environment
variable to error out instead of just warn when unsafe paths are
used. Even though erroring out can be chosen by passing
-Werror=poison-system-directories, we are not sure this option in
CFLAGS will always be passed, so having an environment variable
guarantees it will always be passed, and also allows to have an
identical behavior to the external toolchain wrapper.

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

gcc/arc-2014.08: add patch to warn about unsafe header paths

This commit adds a patch to gcc borrowed from CodeSourcery/Yocto that
warns about unsafe include paths (i.e /usr/include,
/usr/local/include, etc.). The patch was adapted to gcc arc-2014.08,
and modified to support the BR_COMPILER_PARANOID_UNSAFE_PATH
environment variable to error out instead of just warn when unsafe
paths are used. Even though erroring out can be chosen by passing
-Werror=poison-system-directories, we are not sure this option in
CFLAGS will always be passed, so having an environment variable
guarantees it will always be passed, and also allows to have an
identical behavior to the external toolchain wrapper.

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

gcc/4.8: add patch to warn about unsafe header paths

This commit adds a patch to gcc borrowed from CodeSourcery/Yocto that
warns about unsafe include paths (i.e /usr/include,
/usr/local/include, etc.). The patch was adapted to gcc 4.8.3, and
modified to support the BR_COMPILER_PARANOID_UNSAFE_PATH environment
variable to error out instead of just warn when unsafe paths are
used. Even though erroring out can be chosen by passing
-Werror=poison-system-directories, we are not sure this option in
CFLAGS will always be passed, so having an environment variable
guarantees it will always be passed, and also allows to have an
identical behavior to the external toolchain wrapper.

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

gcc/4.9: add patch to warn about unsafe header paths

This commit adds a patch to gcc borrowed from CodeSourcery/Yocto that
warns about unsafe include paths (i.e /usr/include,
/usr/local/include, etc.). The patch was adapted to gcc 4.9.1, and
modified to support the BR_COMPILER_PARANOID_UNSAFE_PATH environment
variable to error out instead of just warn when unsafe paths are
used. Even though erroring out can be chosen by passing
-Werror=poison-system-directories, we are not sure this option in
CFLAGS will always be passed, so having an environment variable
guarantees it will always be passed, and also allows to have an
identical behavior to the external toolchain wrapper.

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

binutils/arc-2014.08: add patch to warn about unsafe library paths

This commit adds a patch to binutils borrowed from CodeSourcery/Yocto
that warns about unsafe library paths (i.e /usr/lib, /usr/local/lib,
etc.). The patch was adapted to binutils arc-2014.08, and modified to
support the BR_COMPILER_PARANOID_UNSAFE_PATH environment variable to
error out instead of just warn when unsafe paths are used. Even though
erroring out can be chosen by passing
--error-poison-system-directories, we are not sure this option in
LDFLAGS will always be passed, so having an environment variable
guarantees it will always be passed, and also allows to have an
identical behavior to the external toolchain wrapper.

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

binutils/2.22: add patch to warn about unsafe library paths

This commit adds a patch to binutils borrowed from CodeSourcery/Yocto
that warns about unsafe library paths (i.e /usr/lib, /usr/local/lib,
etc.). The patch was adapted to binutils 2.22, and modified to support
the BR_COMPILER_PARANOID_UNSAFE_PATH environment variable to error out
instead of just warn when unsafe paths are used. Even though erroring
out can be chosen by passing --error-poison-system-directories, we are
not sure this option in LDFLAGS will always be passed, so having an
environment variable guarantees it will always be passed, and also
allows to have an identical behavior to the external toolchain
wrapper.

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

binutils/2.23: add patch to warn about unsafe library paths

This commit adds a patch to binutils borrowed from CodeSourcery/Yocto
that warns about unsafe library paths (i.e /usr/lib, /usr/local/lib,
etc.). The patch was adapted to binutils 2.23, and modified to support
the BR_COMPILER_PARANOID_UNSAFE_PATH environment variable to error out
instead of just warn when unsafe paths are used. Even though erroring
out can be chosen by passing --error-poison-system-directories, we are
not sure this option in LDFLAGS will always be passed, so having an
environment variable guarantees it will always be passed, and also
allows to have an identical behavior to the external toolchain
wrapper.

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

binutils/2.24: add patch to warn about unsafe library paths

This commit adds a patch to binutils borrowed from CodeSourcery/Yocto
that warns about unsafe library paths (i.e /usr/lib, /usr/local/lib,
etc.). The patch was adapted to binutils 2.24, and modified to support
the BR_COMPILER_PARANOID_UNSAFE_PATH environment variable to error out
instead of just warn when unsafe paths are used. Even though erroring
out can be chosen by passing --error-poison-system-directories, we are
not sure this option in LDFLAGS will always be passed, so having an
environment variable guarantees it will always be passed, and also
allows to have an identical behavior to the external toolchain
wrapper.

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

toolchain-external: instrument wrapper to warn about unsafe paths

The CodeSourcery toolchains have a very interesting feature: they warn
the user when an unsafe header or library path is used, i.e a path
that will lead host headers or libraries to leak into the build.

This commit adds a similar functionality into our external toolchain
wrapper, so that it can be used with all external toolchains, and can
also be tuned as needed. By default, the external toolchain wrapper
now gives warnings such as:

  arm-linux-gcc: WARNING: unsafe header/library path used in cross-compilation: '-I /usr/foo'
  arm-linux-gcc: WARNING: unsafe header/library path used in cross-compilation: '-L /usr/bleh'

but the compilation continues successfully. One can then easily grep
in his build log to search for occurences of this message.

Optionally, if BR_COMPILER_PARANOID_UNSAFE_PATH is defined in the
environment to a non empty value, the external wrapper will instead
error out and abort the compilation.

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

package/openpowerlink: enable dynamic build

Also remove the install hook, all static libraries are
removed from TARGET_DIR/usr/lib by target-finalize target

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

package/openpowerlink: do not override CMAKE_SYSTEM_PROCESSOR

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

package/openpowerlink: bump to version 1.8.5

Remove upstream commit

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

util-linux: add an option to build host programs

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

blktrace: new package

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

package: indentation cleanup

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

netsnmp: bump to version 5.7.3

Security patch now upstream so remove.
Add hash file.

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

mtr: bump to version 0.86 and use github call

[Thomas: remove useless --without-glib option, as noticed by Vincent.]

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

usbutils: drop double needs from comment

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

wget: add host-pkgconf dependency

Uses pkgconfig since 1.16.1+ to find libraries.

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

jasper: add a patch fixing CVE-2014-9029

See http://www.ocert.org/advisories/ocert-2014-009.html for the details.

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

util-linux: add security patch for CVE-2014-9114

Fixes CVE-2014-9114 - command injection flaw in blkid.
See https://bugzilla.redhat.com/show_bug.cgi?id=1168485
Patch upstream.

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

bwm-ng: add a hash file

It will clarify errors like this one:
  http://autobuild.buildroot.net/results/f22/f22888d87659dda8ff75f25490eab0a68f02bc9b/

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

dhcpcd; bump to version 6.6.5

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

squid: bump to version 3.4.10

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

barebox: bump to version 2014.12.0

Signed-off-by: Fabio Porcedda <fabio.porcedda@gmail.com>
Reviewed-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Tested-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

poppler: Install files into staging

Install poppler's header files and libraries into staging, since it
provides a library that can be used by other packages.

[Thomas: slightly extended the commit message.]

Signed-off-by: Alan Ott <alan@signal11.us>
Suggested-by: Jeremie Scheer <jeremie.scheer@armadeus.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

poppler: Add option for Qt support

Add an option to build the libpoppler-qt4 library.

[Thomas: slightly adjust the Config.in file, by:
  - Using an if BR2_PACKAGE_POPPLER ... endif block instead of a depends
    on BR2_PACKAGE_POPPLER for the new option.
  - Rewording the option prompt, since there's no need to indicate that
    the Qt support is for poppler since we're already a sub-option of
    poppler.]

Signed-off-by: Alan Ott <alan@signal11.us>
Suggested-by: Jeremie Scheer <jeremie.scheer@armadeus.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
merge

uclibc: enable UCLIBC_HAS_STRING_GENERIC_OPT in all configs

Description in uClibc for this optoin is:
--->---
bool "Use faster (but larger) generic string functions"
default y
help
  Answer Y to use the (tweaked) glibc generic string functions.

  In general, they are faster (but 3-5K larger) than the base
  uClibc string functions which are optimized solely for size.

  Many people will answer Y.
--->---

Additional 3-5K of size is not that important if resulting libuClibc.so is
of size about 300k.

But benefits in terms of performance could be pretty significant.

For example on ARC in LMbench we see more than 3 times bump in some tests
--->---
libc bcopy unaligned
...
8.39 - 6.76  (UCLIBC_HAS_STRING_GENERIC_OPT=no)
8.39 - 23.86 (UCLIBC_HAS_STRING_GENERIC_OPT=yes)
--->---

Also since the option in question is default in uClibc I don't expect if
enabled back (it was silently disabled in http://git.buildroot.net/buildroot/commit/?id=cfbf8abc33d86a0cf5c1bb3e0817a22009b7f301 on introduction of NPTL in uClibc
by Khem Raj) it to introduce problems.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Anton Kolesov <anton.kolesov@synopsys.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

pkg-cmake.mk: Set CMAKE_SYSTEM_PROCESSOR.

This is rarely needed by packages, but convenient to have when it is.

[Thomas:
  - don't define ARM_VARIANT as this name is too global, use
    CMAKE_SYSTEM_PROCESSOR_ARM_VARIANT instead.
  - don't use ifndef, but a more traditional else clause, for the
    non-ARM cases.]

Signed-off-by: Volker Krause <volker.krause@kdab.com>
Reviewed-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Reviewed-by: Romain Naour <romain.naour@openwide.fr>
Acked-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

support/script/scancpan: remove duplicated dependency

Dependencies from metacpan comes as a list of modules which is
transformed in a list of distribution for BR.  Different modules could
be included in the same distribution, so duplication is possible.

This can for example be seen with the HTTP-Daemon module, which would
get two times the dependencies on HTTP-Message without this commit.

[Thomas: slightly extend commit log.]

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

website: Javascript code cleanup

By inversing the "if (!result.error)" we gain one level of indent.
Also moving variable declarations into a more correct block of code.

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

website: Fix indentation

[Thomas: fix commit title.]

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

website: Remove debug message

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

sqlite: bump to version 3.8.7.4

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

usbutils: bump to version 008

Now uses hwdb from libudev for usb information.
Drop all redundant/obsolete dependencies and cleanup hooks.
Don't install to staging, it's pointless, it provides no libraries at
all and it doesn't bundle usb.ids any more.

[Thomas: change 'comment' in Config.in to match what's recommended in
the Buildroot manual.]

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

slang: use specific build and install commands for static builds

Doing this we ensure that only the static library and a static version
of slsh are built.

Fixes:
  http://autobuild.buildroot.net/results/f12/f124ca3737baf5aca32029226805133d3544715c/

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

slang: add adapted upstream patches for static builds

These are adapted upstream patches to make them apply on the version we
are using in Buildroot, and also taking into account that we already
have another patch which modifies the same file, so these patches would
not apply as they are. Unnecessary parts of the upstream patches have
been ingnored, such as changelogs or version changes.
The purpose of these patches are allowing the static library and a
static version of slsh be built without building any dynamic bit.
These patches will be included in the next release, so we can remove
them when we do the next version bump.

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

slang: rename patches to follow the new name convention

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

thrift: disable tutorial(s)

The tutorials don't pass pedantic mode hence causing failures.
And we don't care about tutorials anyway, disable them to save time as
well. Fixes.
http://autobuild.buildroot.net/results/f68/f68d537e623f565000ca966372a7c0277f4e3888/

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

package/mpd: add option to enable or disable OSS support

By default, the OSS support is enabled, which may not necessarily be
useful on most Linux systems that use ALSA.

[Thomas: tweak commit title and log.]

Signed-off-by: Jörg Krause <jkrause@posteo.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/mpd: enable neighbor discovery support

Enable support for neighbor discovery useful in Samba and UPnP networks.

[Thomas: add dependency on smbclient or upnp plugin, without which the
neighbor discovery support isn't very useful.]

Signed-off-by: Jörg Krause <jkrause@posteo.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/mpd: fix line length for help text

Make help text lines not longer as 80 characters (one tab weights
eight chars).

[Thomas: rewrap more help text.]

Signed-off-by: Jörg Krause <jkrause@posteo.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/mpd: make all package names in help text uppercase

Make the package name uppercase to be consistent in all the help texts.

Signed-off-by: Jörg Krause <jkrause@posteo.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/mpd: fix typo

Fix little type: it's UPnP, not UPnp.

Signed-off-by: Jörg Krause <jkrause@posteo.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/mpd: bump to version 0.19.6

Also update hash file.

Signed-off-by: Jörg Krause <jkrause@posteo.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

python-docopt: new package

[Thomas:
  - Remove dependency on python || python3, as we did for all other
    python modules that work with both Python 2 and Python 3.
  - Rewrap help text.
  - Add <pkg>_LICENSE_FILES value.]

Signed-off-by: Alvaro G. M <alvaro.gamez@hazent.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

iptables: rename patches to new convention

Signed-off-by: Ryan Coe <bluemrp9@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

configs/qemu: update to the latest kernel/headers versions

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

package/musl: remove obsolete setting of prefix

Using the config option '--libdir=/lib' makes the setting of prefix to an
empty path obsolete in MUSL_INSTALL_TARGET_CMDS.

Signed-off-by: Jörg Krause <jkrause@posteo.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/musl: fix C library installation directory

Set the installation path for the libraries to /lib instead of /usr/lib.

This fixes an issue when building a toolchain with the musl library by
the internal toolchain backend of Buildroot in the first step and import this
toolchain later as a custom external toolchain in a second step. For this use
case check-musl in toolchain/helpers.mk failed because it did not find the
libc or libm in sysroot/lib.

This patch superseeds: [PATCH 1/1] toolchain/helpers.mk: fix check-musl
http://patchwork.ozlabs.org/patch/417587/

Signed-off-by: Jörg Krause <jkrause@posteo.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

bind: security bump to version 9.9.6-P1

Fixes CVE-2014-8500 - A flaw in delegation handling could be exploited
to put named into an infinite loop, in which each lookup of a name
server triggered additional lookups of more name servers.

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

lighttpd: disable static build

lighttpd modules support code requires shared libraries support.

The scons based build system seems to be support static link of modules, but
we currently don't use scons for lighttpd.

Fixes:
http://autobuild.buildroot.net/results/39c/39c027bf6b76a0f9c15a184f449dadacfaefe980/

[Peter: move to top of file to fix indentation of sub options]
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libsoxr: add a hash file

It will clarify errors like this:
  http://autobuild.buildroot.net/results/923/923d4ba849839beca89f427687bbd903d6037b15/

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

ltrace: disable elfutils unwinding when libunwind is used

ltrace can't use both. Use --with-elfutils=no to disable use of elfutils
(libdwfl) for unwinding.

Note that elfutils (libelf) is still a mandatory dependency regardless of
--with-elfutils.

Fixes:
http://autobuild.buildroot.net/results/a3c/a3c1647578cdeda3aaf672084a9a893f06dbad26/
http://autobuild.buildroot.net/results/fdb/fdb7e760faeda77b263a6dac3e322a25b53e67ff/
http://autobuild.buildroot.net/results/70a/70a0614ca7baf6a46898834a87cd46865f6e3385/

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

libcgicc: bump to version 3.2.16

Also add hash file.

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

python-thrift: bump to version 0.9.2

Following the bump of the thrift package, it makes sense to bump the
python-thrift package as well. Python 3 is still not supported
apparently, so we keep this package a Python 2 only one.

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

thrift: bump to version 0.9.2

Drop patches that are upstream.
Add hash file.
Tweak autoreconf patch since README is now missing as well.

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

support/scripts/dependencies.sh: cleanup the output of the 32 bits gcc test

The screen is cluttered when we build for 32 bit target and 32 bit gcc
is missing.

  ~/buildroot$ make
  [...]
  /usr/bin/ld: cannot find crt1.o: No such file or directory
  /usr/bin/ld: cannot find crti.o: No such file or directory
  /usr/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-linux-gnu/4.9/libgcc.a when searching for -lgcc
  /usr/bin/ld: cannot find -lgcc
  /usr/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-linux-gnu/4.9/libgcc_s.so when searching for -lgcc_s
  /usr/bin/ld: cannot find -lgcc_s
  /usr/bin/ld: cannot find -lc
  /usr/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-linux-gnu/4.9/libgcc.a when searching for -lgcc
  /usr/bin/ld: cannot find -lgcc
  /usr/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-linux-gnu/4.9/libgcc_s.so when searching for -lgcc_s
  /usr/bin/ld: cannot find -lgcc_s
  /usr/bin/ld: cannot find crtn.o: No such file or directory
  collect2: error: ld returned 1 exit status
  [...]
  Your Buildroot configuration needs a compiler capable of building 32 bits binaries.

The final note is enough, and adding 2>/dev/null to the gcc test
invocation is also more consistent with the rest of the script. The
patch makes the '/usr/bin/ld:' and 'collect2:' lines go away.

Signed-off-by: Jens Stimpfle <debian@jstimpfle.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

support/scripts/dependencies.sh: improve g++ test to avoid misleading output

When g++ is not installed, a misleading error message turns up because
of a bad combination of an unquoted shell variable and control flow.

 ~/buildroot$ make
 You may have to install 'g++' on your build machine
 /home/testuser/buildroot/support/dependencies/dependencies.sh: 136: [: -lt: unexpected operator

[Thomas:
  - fixed commit log, as per the suggestion of Yann E. Morin.
  - don't change existing empty new lines, suggested by Yann.
  - use positive logic in the newly added test, suggested by Yann.]

Signed-off-by: Jens Stimpfle <debian@jstimpfle.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ola: bump to 0.9.3

This update requires a few changes. First, good news, the patches that you
guys have submitted to OLA have been merged, so we can drop those:

 ola-0001-fix-build-warning.patch -> ea375582b0bfee93d66608ffc807078ffc48e961
 ola-0002-move-python-sub-check-to-configure.ac.patch -> 673a7602a6bf7a6aa8a8461ebd9362d59f6e21df
 ola-0003-fix-check-for-python-module-for-cross-compilation.patch -> b51b48be81ec38bc7d1229be0c7d3189c5ddbafa

Less good news: OLA now builds a small protoc wrapper (ola_protoc) that is
built and executed at compile-time on the host. If we don't change anything,
ola_protoc is built with the target toolchain and therefore can't run on the
host. Explanation for ola_protoc is here

To solve this, I created a package host-ola, which builds and provides an
ola_protoc for the host. It tries to disable as much as possible of things
that we won't need at configure time. Only ola_protoc is built and installed
so it's not that long.

The change has been built-tested only.

[1] https://github.com/OpenLightingProject/ola/blob/master/protoc/ola-protoc.cpp#L20

[Thomas:
 - add HOST_OLA_DEPENDENCIES variable, so that host-ola doesn't get
   all the dependencies of the target ola package: certain target ola
   dependencies do not have corresponding host packages (and it
   doesn't make sense to have them as dependencies just to build
   ola_protoc)
 - improve the commit log to mention which upstream commits correspond
   to our patches.]

Signed-off-by: Simon Marchi <simon.marchi@polymtl.ca>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>