configs/arm_foundationv8: fix build and update documentation

The arm_foundationv8_defconfig currently doesn't build because it uses
an external AArch64 toolchain that has been updated (Linaro toolchain)
and is now based on gcc 5.x. Unfortunately, Linux 3.13 is too old to
be built with gcc 5.x.

As a minimal fix to make this defconfig build again, we switch to use
the internal toolchain backend. At the time this defconfig was
introduced, there was no support for AArch64 in the internal toolchain
backend, but now that it is available, there is no reason to not use
it. This makes the defconfig also more consistent with the other
defconfigs.

Obviously, the kernel headers used to build the toolchain are set to
the fixed version 3.13, so that they match the kernel being used.

While we're at it, the readme.txt file corresponding to this defconfig
is updated. Runtime testing has been performed with the latest version
of the ARM Foundation model.

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

package/zmqpp: add patch to fix build issue

Add a patch from upstream to fix build error:
  error: no match for 'operator<<'

Fixes:
http://autobuild.buildroot.net/results/ed4508a00663b03632a4eb9411ba05852d02fd88/

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Tested-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

apitrace: do not use bundled version of libpng

In case libpng is not found apitrace its own version of libpng. However
current bundled version (1.1.16) is impacted by
https://sourceforge.net/p/libpng/bugs/230/. We simplify problem by
forcing use of external libpng.

Notice this patch also impact use of zlib. Apitrace also use its own
version of zlib if none was found. Since we select libpng, external
zlib is also selected.

Fix:
  http://autobuild.buildroot.net/results/ac790191127cf4d8b19d0c7c640981264772f503/

Signed-off-by: Jérôme Pouiller <jezz@sysmic.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

bullet: disable on microblaze

gcc doesn't finish compiling btSoftBody.cpp when optimization is
enabled.

Fixes:
http://autobuild.buildroot.net/results/6e4a9ede477b031f47f7e271c38f67f432a3573c
http://autobuild.buildroot.net/results/7dc4ab759dd3b1e3e03fe52c78e31327340bf8c9
http://autobuild.buildroot.net/results/637bae8b2be6c4c64f4f7c661cc8d4606b553538

[Thomas: add reference to gcc bug.]

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

package/mpd: fix static build issue with alsa

mpd forgets to link the archive libmixer_plugins.a against libasound leading
to undefined references in a static context.

Add a patch sent upstream to fix:
http://autobuild.buildroot.net/results/f10/f10c6ea2d29bbcef8f33fc9ba1430e99edca895d/
http://autobuild.buildroot.net/results/312/31212c484f452fe4ea6ae084d3edefc4580830d8/

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gpsd: add work-around and re-enable for microblaze

Fix (works around) bug #6872
https://bugs.busybox.net/show_bug.cgi?id=6872

The gpsd package was disabled due to an internal compiler error for
the microblaze architecture in the following commit:
http://git.buildroot.net/buildroot/commit/?id=d517415bcda380336f4a27bf248cef4d81791c9e

The problem is due to an specific GCC optimization flag
(-fexpensive-optimizations), so disabling it we can work around this
problem and compile gpsd for the microblaze architecture.

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

libxml2: security bump to version 2.9.3

- Fixes:
  - CVE-2015-5312 - Another entity expansion issue
  - CVE-2015-7497 - Avoid an heap buffer overflow in xmlDictComputeFastQKey
  - CVE-2015-7500 - Fix memory access error due to incorrect entities boundaries
  - CVE-2015-8242 - Buffer overead with HTML parser in push mode

- Incorporates upstreamed patches as well, which also fixed:
  - CVE-2015-1819 - The xmlreader in libxml allows remote attackers to cause
    a denial of service (memory consumption) via crafted XML data, related
    to an XML Entity Expansion (XEE) attack.
  - CVE-2015-7941 - out-of-bounds memory access.
  - CVE-2015-7942 - heap-buffer-overflow in xmlParseConditionalSections.
  - CVE-2015-8035 - DoS via crafted xz file.

Signed-off-by: Danomi Manchego <danomimanchego123@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/yajl: add patch to fix static build issue

mpd's plugin soundcloud requires yajl. In static build context yajl library is
named yajl_s so linking against yajl fails:
  /bin/ld: cannot find -lyajl

Use a pull request patch [1] to let the shared and the static library
have the same name. They will be differentiated by the extension (.so or .a).

Fixes:
http://autobuild.buildroot.net/results/5ac/5acba9b0ed852512fc88e83973a3b5389e6d54be/
http://autobuild.buildroot.net/results/739/73934ad67efa121fedeaa03fdad290aa63f1485e/

[1] https://github.com/lloyd/yajl/pull/174

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/yajl: add hash

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ustr: fix build issue with GCC5

This patch fixes the following error messages:

ustr-cmp-code-so-dbg.o: In function `ustr_pool_make_subpool':
ustr-cmp-dbg-code.c:(.text+0x0): multiple definition of `ustr_pool_make_subpool'
ustr-b-code-so-dbg.o:ustr-b-dbg-code.c:(.text+0x0): first defined here
ustr-cmp-code-so-dbg.o: In function `ustr_pool_clear':
ustr-cmp-dbg-code.c:(.text+0x44): multiple definition of `ustr_pool_clear'
ustr-b-code-so-dbg.o:ustr-b-dbg-code.c:(.text+0x44): first defined here
ustr-cmp-code-so-dbg.o: In function `ustr_pool_free':
...

That's because the application assumes gnu89 behaviour of "inline" and
"extern inline", but gcc5 defaults to gnu11. So we need to force gnu89
in the compilation flags.

Fixes:
http://autobuild.buildroot.org/results/fde02ae2883a058b98e57ba087466d820ab1a5eb/
http://autobuild.buildroot.org/results/e04849c31e2a2528282ce6cdcf69aa10076d721d/
http://autobuild.buildroot.org/results/7b3295e711fc51907c311adb63f0b7d06a90a7b9/
...and many others.

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

package/zmqpp: add hash file

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

mongrel2: add upstream patch to fix symbol conflict

The symbol 'max_align_t', defined in the mongrel2 sources
(src/mem/align.h), is also defined in one of the standard headers in
some toolchains (the ones based on GCC 5+ apparently).

This conflict has been solved upstream, and a new version (1.10.0)
containing the fix has been released already, so this backported patch
may be dropped in the future when mongrel2 gets a version bump.

Fixes:

  http://autobuild.buildroot.net/results/b7f/b7f86bdda094de96633e9921e0e5895eadc89217/
  http://autobuild.buildroot.net/results/f62/f623703aa6d7992b67d305331734e4ed0c7b2be4/
  http://autobuild.buildroot.net/results/eb3/eb385cb4f87078122028e85cbe31caa01d794991/
  ...

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

cryptodev-linux: add build fix patch for 4.3+ kernels

Patch status: sent upstream
https://mail.gna.org/public/cryptodev-linux-devel/2015-11/msg00003.html

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

configs/qemu: enable dhcp on network interfaces

Enable dhcp networking for qemu configs considered good enough for
testing purposes.

This excludes:
arm_nuri - emulation doesn't seem 100% correct for networking.
ppc_virtex_ml507 - doesn't emulate networking.
sh4eb - emulation doesn't seem 100% correct for the NIC.

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

configs/qemu: cleanup and capitalization

Capitalize comments in a consistent way.
Cleanup redundant entries.
Drop global patch dir for sparc ss10 since it's not required any more.

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

sdl: overwrite libpth autodetection

The sdl libpth autodetection gets confused in case a host
version is present. Overwrite libpth autodection for the
target build dependent on BR2_PACKAGE_LIBPTHSEM_COMPAT and
use pth-config from staging directory.

Fixes [1]:

checking for pthreads... no
checking for pth-config... /usr/bin/pth-config
checking pth... yes
[...]
./src/SDL.c:33:17: fatal error: pth.h: No such file or directory
 #include <pth.h>

[1] http://autobuild.buildroot.net/results/733/73330e51b3c099d179004db0b9522ac13358b8f7

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

iproute2: fix build with musl

This commit adds two patches to the iproute2 package that are
necessary to build it against the musl C library. Both patches have
been submitted upstream.

Fixes:

  http://autobuild.buildroot.org/results/5d3/5d3f8c2242e9a89bb545166fc7d70429d2cddd42/

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

xapp_sessreg: fix build issue with gcc 5

Backport an upstream commit that fixes the build of xapp_sessreg with
gcc 5.

Fixes:

  http://autobuild.buildroot.org/results/ec0/ec01564854f0738df05b8d995d9f0db396a0e5a8/

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

Update for 2015.11-rc2

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

perl-file-util: restore host variant and fix dependencies

The host variant was added by us (DATACOM) because it is useful to
build some third-party packages that we use in our firmware. We humbly
request you to keep it alive.

Declare that it requires host-perl-module-build.

Signed-off-by: Carlos Santos <casantos@datacom.ind.br>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libserial: add patch to fix build failure on sparc/sparc64

Like c-periphery and lua-periphery, libserial fails to build because
it tries to use some baud rate definitions that are not available on
sparc and sparc64 (the highest baud rates).

This commit fixes that by introducing a patch to libserial to make the
use of these high baud rates conditional on their availability.

Fixes:

  http://autobuild.buildroot.org/results/f9b/f9bbb8a6636cd3e3203b059f627aac7b1d511eb2/

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

libuecc: not available in static builds

Fixes:
http://autobuild.buildroot.net/results/c0d/c0d91fc0fa28fda0ffd1c9a236bde5435222bcc0/

Unconditionally builds a shared library.

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

package/minidlna: fix static build

libvorbis.pc does not exist, it is named vorbis.pc.
Also, sqlite3 was missing when querying pkg-config.

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

slang: fix build on musl

Add patch from Alpine Linux to fix the build of slang with the musl C
library.

Fixes:

  http://autobuild.buildroot.org/results/bc3/bc3cca6e6a906afa4c57dcddfb4a9ecd0cb44398/

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

directfb: disable for musl toolchains

Fixes:
http://autobuild.buildroot.net/results/548/548bf7b800a946e97135c0ba66363f388e517d6f/

Directfb doesn't build with musl, and upstream seems dead - So just disable
it.

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

qt5base: use 'depends on' for directfb

This is the only package where we select directfb. Change it to use depends
on to match what we do for the X11 backend.

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

core/pkg-infrastructures: remove LD_LIBRARY_PATH from the environment

If system tools are selected, the host's lib/ directory may shadow
libraries from the system which are configured differently and do not
have all of the symbols required by the system tool.

Since buildroot now uses rpath everywhere, LD_LIBRARY_PATH should not
be necessary anyways.

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

core: check host executables have appropriate RPATH

When we build our host programs, and they depend on a host library we
also build, we want to ensure that program actually uses that library at
runtime, and not the one from the system.

We currently ensure that in two ways:
  - we add a RPATH tag that points to our host library directory,
  - we export LD_LIBRARY_PATH to point to that same directory.

With these two in place, we're pretty much confident that our host
libraries will be used by our host programs.

However, it turns our that not all the host programs we build end up
with an RPATH tag:
  - some packages do not use our $(HOST_LDFLAGS)
  - some packages' build system are oblivious to those LDFLAGS

In this case, there are two situations:
  - the program is not linked to one of our host libraries: it in fact
    does not need an RPATH tag [0]
  - the program actually uses one of our host libraries: in that case it
    should have had an RPATH tag pointing to the host directory.

For libraries, they only need an RPATH if they depend on another library
that is not installed in the standard library path. However, any system
library will already be in the standard library path, and any library we
install ourselves is in $(HOST_DIR)/usr/lib so already in RPATH.

We add a new support script that checks that all ELF executables have
a proper DT_RPATH (or DT_RUNPATH) tag when they link to our host
libraries, and reports those file that are missing an RPATH. If a file
missing an RPATH is an executable, the script aborts; if only libraries
are are missing an RPATH, the script does not abort.

[0] Except if it were to dlopen() it, of course, but the only program
I'm aware of that does that is openssl, and it has a correct RPATH tag.

[Peter: reworded as suggested by Arnout, fix HOT_DIR typo in comment]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/mplayer: Fix linking with libvorbis/libdvdread

Fixes
http://autobuild.buildroot.net/results/7f6/7f6e5396e95a3c3b676ca59bedc5317aa88e5f4c/
http://autobuild.buildroot.net/results/6ac/6acd9cf25c3088d313247510db47a214fff30c67/
http://autobuild.buildroot.net/results/a0d/a0da41bb734dbcdb17e13d852cfbdba5e069429e/
http://autobuild.buildroot.net/results/d0c/d0caf60c1ddf4e05086c93f646e32b5ddc75b6ce/

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
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>

libxml2: add several security patches

Fixes:
CVE-2015-1819 - The xmlreader in libxml allows remote attackers to cause
a denial of service (memory consumption) via crafted XML data, related
to an XML Entity Expansion (XEE) attack.
CVE-2015-7941 - out-of-bounds memory access.
CVE-2015-7942 - heap-buffer-overflow in xmlParseConditionalSections.
CVE-2015-8035 - DoS via crafted xz file.

All patches upstream.

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

qemu: add patch to fix SSP support detection

The QEMU configure script incorrectly assumes SSP is supported by the
toolchain in some cases where the compiler accepts -fstack-protector*
flags but the C library does not provide the necessary __stack_chk_*()
functions.

Even though a full compile and link test is performed by the script,
this is done with a code fragment which does not actually meet any of
the conditions required to cause the compiler to emit canary code when
the -fstack-protector-strong variant is used. As no compile or link
failure occurs in this case, a false positive is generated and a
subsequent error is seen when the probe for pthreads is performed.

The fix consists in patching the configure script to use a more
appropriate test program for the SSP support checks.

Fixes:

  http://autobuild.buildroot.net/results/efb/efbb4e940543894b8745bb405478a096c90a5ae2/
  http://autobuild.buildroot.net/results/32d/32d6d984febad2dee1f0d31c5fa0aea823297096/
  http://autobuild.buildroot.net/results/aa6/aa6e71c957fb6f07e7bded35a8e47be4dadd042c/
  ...and many others.

Signed-off-by: Rodrigo Rebello <rprebello@gmail.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
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>

package/gauche: fix build error on BR2_STATIC_LIBS

Building gauche depends on dynamic link, so add dependency on
!BR2_STATIC_LIBS.

Fix:
http://autobuild.buildroot.org/results/da5/da5b9605552d4914c5e6f0d890367b92536419c1

Signed-off-by: Hiroshi Kawashima <kei-k@ca2.so-net.ne.jp>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

linux-headers: bump 3.2.x series

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

configs/qemu: enable tmpfs for microblaze

Otherwise the tmpfs-backed filesystems fail:
mount: mounting tmpfs on /dev/shm failed: Invalid argument
mount: mounting tmpfs on /tmp failed: Invalid argument
mount: mounting tmpfs on /run failed: Invalid argument

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

configs/qemu: fix networking for sparc64

It contained a typo that prevented the driver from being built.

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

configs/qemu: bump x86 to 4.3 kernel

Now that uclibc-ng was bumped to version 1.0.9 nothing prevents x86
from working correctly.

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

liquid-dsp: blacklist CS powerpc toolchains

They're unable to build liquid-dsp successfully resulting in an internal
compiler error. Fixes:
http://autobuild.buildroot.net/results/2ac/2acd745f4e5ec6edfd09afcd57b9ece216724c1b/

[Thomas: propagate the dependencies to the comment.]

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

qt5base: Patch Qt config to detect sunxi-mali

Currently Qt 5.5 only detects and build the eglfs_mali device
integration if the commercial Mali driver package from ARM is used.
This patch makes sure the Qt configure script also test for the
sunxi-mali driver package. It also removes the dependency of
the proprietary fbdev_window.h.

This issue is set to be fixed in upcoming Qt 5.6:
https://codereview.qt-project.org/#/c/125837/

[Thomas: renumber patch from 0010 to 0009.]

Signed-off-by: Daniel Nyström <daniel.nystrom@timeterminal.se>
Tested-by: Ariel D'Alessandro <ariel@vanguardiasur.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Tested-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
[Thomas: build tested before the patch, verified that indeed the
eglfs-mali plugin doesn't get built, and that after the patch it gets
built as expected.]

pulseaudio: fix build issue with old gcc versions

This commit adds a patch to pulseaudio (already submitted upstream)
that avoids a structure redefinition build failure that occurs with
old gcc versions.

Fixes:

  http://autobuild.buildroot.org/results/a1e/a1e7e59f9c65a6ce38800439c78b7808048ad708/

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

package/linux-headers: add licensing information

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

core/legal-info: fix missing double-$

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

package/libecore: do not build anything at patch time

Trying to build anything at patch time will result in a broken
legal-info, as the needed host dependencies are not yet built.

Make that hook a pre-configure hook rather than a post-extract
hook.

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

package/libcurl: carefully override LD_LIBRARY_PATH

To build libcurl, we need to override LD_LIBRARY and force it to a sane
value, otherwise libcurl is confused when target == host (see a51ce319,
libcurl: fix configure with openssl when target == host).

That is currently OK, since we always set LD_LIBRARY_PATH to a non-empty
value.

However, we're soon to stop setting it at all.

So, if the user has an empty (or no) LD_LIBRARY_PATH in his envirnment,
we'd end up adding the current working directory to LD_LIBRARY_PATH (as
an empty entry in a colon-separated list is most probably interpreted as
meaning the current working directory, which we do know can cause issue,
and which we expressely check against in support/dependencies/dependencies.sh

Fix that by only using an existing LD_LIBRARY_PATH if it is not empty.
Also use a Makefile construct as it is easier to read than a shell one
(we can do that, as all variables from the environment are available as
make variables).

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/perl-file-util: remove host variant

It does not build, and no one depends on it.

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

package/mysql: unconditionally define host variables

To build the host-mysql, we only build parts of the source, just the
strictly minimum required to then cross-compile it.

However, the host variables (conf opts, build and install cmds) are only
defined when the mysql server is enabled in the configuration.

So, this breaks:
    make defconfig; make host-mysql

Even though it is not much use to have that partial host-mysql on its
own, it is still very interesting to be able to build it, if at least
for testing changes in the core package infrastructures (like new step
hooks or the likes...)

Move the definitions of the host variant out of the server conditional
block.

[Peter: add comment about what we build and why as suggested by Arnout]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Marcelo Gutiérrez(UTN/FRH) <kuyurix@gmail.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/axfsutils: fix Makefile

mkfs.axfs uses zlib, but does not have an rpath to our host dir.

That's because:
  - we're not passing our host CFLAGS or LDFLAGS
  - it is forcibly setting CFLAGS in the Makefile, overriding anything
    specified by the user
  - it is not using LDFLAGS at all

Add two patches so that CFLAGS and LDFLAGS from the environment are
used if present.

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

apply-patches.sh: use everything from the series file

When a series file exists, we should use every file mentioned in it,
not just the ones ending with .patch or .diff. Also, there's no need
to uncompress anything if it's mentioned in a series file (the tools
that manipulate series files don't support compressed patches).

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Doug Kehn <rdkehn@yahoo.com>
Tested-by: Doug Kehn <rdkehn@yahoo.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

arch: add tune options for powerpc e5500 and e6500

These were forgotten when the subarches were added in cd88e49.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Alvaro Gamez <alvaro.gamez@hazent.com>
Cc: Gustavo Zacarias <gustavo@zacarias.com.ar>
Tested-by: Alvaro G. M <alvaro.gamez@hazent.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

gcc: add patches for powerpc e6500 64-bit support

Building with -mtune=e6500 led to build failures in glibc (probably in
uclibc as well) because gcc was built for a 32-bit target even though
the target tuple is powerpc64-*. This lead to a mix of 32-bit and
64-bit support and build errors like:

  fatal error: gnu/lib-names-32.h: No such file or directory

The root cause is that the configure script is not handling e6500
correctly, because of stupid typo in the condition.

Change has been submitted upstream.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Alvaro Gamez <alvaro.gamez@hazent.com>
Cc: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

uclibc: cleanup unused ARC stuff

ARC is supported by uClibc-ng, cleanup old code.

Signed-off-by: Waldemar Brodkorb <wbx@uclibc-ng.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

ushare: fix linking issue with gcc 5.x

Fixes:
http://autobuild.buildroot.net/results/1d4/1d4d621f3bd7862a75bd7e95e275ccb6694b54c9/
http://autobuild.buildroot.net/results/140/140767a7437b60a62f553ab49c104c37a4949c6c/
http://autobuild.buildroot.net/results/a88/a88ae349d4d7cdff4f15c8ad9e3a1eac74dc6c4b/
http://autobuild.buildroot.net/results/599/59946c728593502e4a17cb7e2d75eabac54db537/

And many more.

GCC5 defaults to -std=gnu11, which has different semantics for inline than
previous versions:

https://gcc.gnu.org/gcc-5/porting_to.html

Which causes linker issues when display_headers() and start_log() are
referenced from other files.  There's no real reason why these needs to be
inline, so just drop the keyword.

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

uclibc: update to 1.0.9

Fixes a fanotify include bug where old kernels (2.6.x) fail to build.
Fixes recent failures on x86 with Linux 4.3.x kernel regarding
network sockets. Includes the removed patch.

Signed-off-by: Waldemar Brodkorb <wbx@uclibc-ng.org>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

lua-periphery: fix build on Sparc/Sparc64

Add a patch similar to the one for c-periphery fixing the build on
Sparc and Sparc64 caused by unsupported high baudrates on this
architecture.

Fixes:

  http://autobuild.buildroot.org/results/c2f/c2fd64ed2f17c53a4704284a8281305f97a48169/

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

package/x11r7/xdriver_xf86-video-siliconmotion: Fix compile warnings being treated as error

Fixes
http://autobuild.buildroot.net/results/059/059d771cb39532dee73a5f50e24b3441005ed38b/
http://autobuild.buildroot.net/results/542/5429dd3f07430cf1487a8cfa82ba55e66a640473/
http://autobuild.buildroot.net/results/073/073105a692033665d31b2dc034eff2e0fdca7f10/
http://autobuild.buildroot.net/results/018/01805d60b8b046122d357e2559268e8a45dcf9d3/
and many others

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

package/gauche: force to use internal libatomic_ops

Force to use internal libatomic_ops of gauche, since the external
libatomic_ops packaged in Buildroot is not compatible with gauche.

Fixes:

  http://autobuild.buildroot.org/results/a97/a979f6a58f0fbb1896830461d6c384f4a65c9429/

[Thomas: tweak commit log.]

Signed-off-by: Hiroshi Kawashima <kei-k@ca2.so-net.ne.jp>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/gauche: fix getcontext build issue at link time

This patch is ported from bdwgc package. Since Gauche uses an internal
copy of the boehm gc code, it is affected by the same problem.

Both configure and configure.ac are modified because autoreconf fails
due to an incompatibility with the version of the autotools used by
Buildroot.

Fixes:

   http://autobuild.buildroot.org/results/8c6/8c6ce526735a36b8c8b4c6047e3c69039c4527cb/
   http://autobuild.buildroot.org/results/894/89460829a737a4dab19103f7a46905636420b0a8/

[Thomas: adjust commit log and patch description.]

Signed-off-by: Hiroshi Kawashima <kei-k@ca2.so-net.ne.jp>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Tested-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/gauche: exclude build on archs not boehm gc ported

This patch limits gauch build only on architectures boehm gc is ported.

Will fix:
http://autobuild.buildroot.org/results/368/3681948013fcd39cc80b690b880924421e429c80//
http://autobuild.buildroot.org/results/ba7/ba7eb5a878ec25033596dc7c9b0f4678d4a83de5//
http://autobuild.buildroot.org/results/1cb/1cbf1ee5bde3d50b935ae8ba931c8d0cc1052acc//
http://autobuild.buildroot.org/results/8af/8afb36606eabdc6ae6e65a28a11877af39f9a7e0//

[Thomas:
 - add comment explaining where the architecture dependency is coming
   from, as suggested by Arnout.
 - move the definition of BR2_PACKAGE_GAUCHE_ARCH_SUPPORTS to the
   beginning of the Config.in file, to be consistent with what we do
   in other packages.]

Signed-off-by: Hiroshi Kawashima <kei-k@ca2.so-net.ne.jp>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

zedboard: uboot: load env earlier

By default, with the original patch bootcmd runs sdboot before loading uEnv.txt.
Consequently, if the user change modeboot, the user's defined content is updated
too late.
By loading uEnv.txt before 'run $modeboot' the correct boot mode is used instead
of default mode.

Signed-off-by: Gwenhael Goavec-Merou <gwenhael.goavec-merou@trabucayre.com>
Acked-By: Jan Viktorin <viktorin@rehivetech.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

boost: do not allow context module on sparc

The boost context module does not build on sparc, even if sparc is
advertised as a supported architecture by the boost documentation. So
let's disallow this module for the time being.

Fixes:

  http://autobuild.buildroot.org/results/206/2060e6e31c4d739947097faf6587e0a06681fee2/

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

uboot: fix ARCH for i386 and x86_64

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

strongswan: security bump to version 5.3.4

Fixes:
CVE-2015-8023 - authentication bypass vulnerability in the eap-mschapv2
plugin that was caused by insufficient verification of the internal
state when handling EAP-MSCHAPv2 Success messages received by the
client.

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

libgudev: add a hash file

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

package/fastd: propagate libcap dependency

libcap neds headers >= 3.0, so propagate that down to fastd which
selects libcap.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Alexander Dahl <post@lespocky.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/ebtables: disable for musl toolchains

ebtabes fail to build with a musl based toolchain due to several errors.
Most of them has been fixed by [1] and [2] but it's not enought because
useful_functions.c fail to build due to mixed usespace and kernel headers
usage [3].

In file included from [...]/sysroot/usr/include/netinet/ether.h:8:0,
                 from useful_functions.c:28:
[...]/sysroot/usr/include/netinet/if_ether.h:96:8: error: redefinition of ‘struct ethhdr’
 struct ethhdr {
        ^
In file included from include/linux/netfilter_bridge.h:8:0,
                 from include/linux/netfilter_bridge/ebtables.h:17,
                 from include/ebtables_u.h:27,
                 from useful_functions.c:25:
include/linux/if_ether.h:119:8: note: originally defined here
 struct ethhdr {
        ^
Makefile:78: recipe for target 'useful_functions.o' failed

Fixes:
http://autobuild.buildroot.net/results/75c/75cc79ae2a2d05ad27198f1b745c8bef37c3a87d/build-end.log

[1] http://git.netfilter.org/ebtables/commit/?id=f8079671326e9fd079391d24911a9a8a77f1d6fd
[2] http://git.netfilter.org/ebtables/commit/?id=4488ba7aa2bb612a624c68957f464fc0ee786e05
[3] http://wiki.musl-libc.org/wiki/FAQ#Q:_why_am_i_getting_.22error:_redefinition_of_struct_ethhdr.2Ftcphdr.2Fetc.22_.3F

Signed-off-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libpng: security bump to version 1.6.19

Fixes:
png_set_PLTE/png_get_PLTE functions failed to check for
an out-of-range palette when reading or writing PNG files with a bit_depth
less than 8.

CVE not yet assigned.

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

zxing-cpp: needs dynamic library

Fixes:
http://autobuild.buildroot.net/results/982/982bd00de9da8cedd47209ece4fac92bbcccb04b/

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

x265: needs dynamic library

Fixes:
http://autobuild.buildroot.net/results/60f/60fc8d54ec61c96e9beaa84154fe8d7e61e55d4e/

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

fastd: needs mmu

Fixes:
http://autobuild.buildroot.net/results/87b/87bf97966cb802fc307ec9710725a3947ec33c39/

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

busybox: add upstream unzip segv patch

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

ranger: propagate python3 wchar dependency

Fixes:
http://autobuild.buildroot.net/results/da1/da1a4cac055ab4b1a301cf719233ff00d1e793a2/
http://autobuild.buildroot.net/results/8bb/8bbfd91d1d972b5d382e7355a8a36ab532510745/
http://autobuild.buildroot.net/results/05c/05c7691c0cc1ef4cfa842643ab564a3958e66542/
http://autobuild.buildroot.net/results/dfd/dfdbb574708393b627b1244a627444478155c70f/

Python3 needs wchar, so ranger needs to depend on that as well before we can
safely select python3.

While we're at it, also add a comment about the toolchain dependencies.

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

ibrdtnd: don't build (latex) documentation

Fixes:
http://autobuild.buildroot.net/results/05d/05d2d80050061e3fe1bd95a9644f4178ef73ef62

The build system builds api documentation if it finds pdflatex, which we
don't need to spend time on and which may fail if not all the pdflatex
dependencies are available, so disable it.

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

ibrdtnd: needs host-pkgconf

The configure script uses pkg-config to find the optional dependencies, so
make sure it is present:

checking pkg-config is at least version 0.9.0... yes

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

polarssl: security bump to version 1.2.18

Fixes a potential heap corruption on Windows when
mbedtls_x509_crt_parse_path() is passed a path longer than 2GB. This
cannot be triggered remotely. Found by Guido Vranken, Intelworks.

Fixes a potential buffer overflow in some asn1_write_xxx() functions.
This cannot be triggered remotely unless you create X.509 certificates
based on untrusted input or write keys of untrusted origin. Found by
Guido Vranken, Intelworks.

The X509 max_pathlen constraint was not enforced on intermediate
certificates. Found by Nicholas Wilson, and fix and tests provided by
Janos Follath.

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

ltris: add upstream patch to fix buld with GCC 5

LTris was failing to build with GCC 5 due to a change in semantics
regarding inline functions compared to previous GCC versions.

The issue has already been solved upstream, but a new version hasn't
been released yet.

Fixes:

  http://autobuild.buildroot.net/results/d5e/d5e832c3aa7ee7af128e2ba1835fe7d5d69d7ce8/
  http://autobuild.buildroot.net/results/e73/e7306093e0ae1a1dcb69f015b10d56e4c95e689e/
  http://autobuild.buildroot.net/results/781/781e891e8a59dab29b0ec37d43cd03107f6cce5b/
  ...

Signed-off-by: Rodrigo Rebello <rprebello@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

configs/qemu: bump to the latest linux version

ppc-mpc8544ds: switch to custom kernel config since the new 4.3-shipped
mpc85xx_basic_defconfig doesn't work with Qemu.
Incidentally while cleaning it up it now seems to work fine with newer
qemu versions.

sparc64_sun4u: ditch ne2k driver since it's useless, and add the e1000
driver which works fine.

x86: stick to 4.2.x kernels since 4.3 doesn't work right with uclibc
(any variant) based toolchains.

Results table:

Defconfig               Kernel          Qemu    Network Status
--------------------------------------------------------------
aarch64_virt            4.3             2.3.0   YES     OK
arm_nuri                3.10.93         2.3.0   NO(1)   OK(2)
arm_versatile           4.3             2.3.0   YES     OK
arm_vexpress            4.3             2.3.0   YES     OK
microblazebe            4.3             2.3.0   YES     OK
microblazeel            4.3             2.3.0   YES     OK
mips64el_malta          4.3             2.3.0   YES     OK
mips64_malta            4.3             2.3.0   YES     OK
mipsel_malta            4.3             2.3.0   YES     OK
mips_malta              4.3             2.3.0   YES     OK
ppc_g3beige             4.3             2.3.0   YES     OK
ppc_mpc8544ds           4.3             2.3.0   YES     OK
ppc_virtex_ml507        4.3             2.3.0   NO      OK
ppc64_pseries           4.3             2.3.0   YES     OK
sh4                     4.3             2.3.0   YES     OK
sh4eb                   4.3             2.3.0   NO(1)   OK
sparc_ss10              4.3             2.3.0   YES     OK
sparc64_sun4u           4.3             2.3.0   YES     OK
sparc_sun4u             4.3             2.3.0   YES     OK
x86                     4.2.6           2.3.0   YES     OK(3)
x86_64                  4.3             2.3.0   YES     OK
xtensa_lx60             4.3             2.3.0   YES     OK

(1) - Supposed to work but broken, probably Qemu emulation issue
(2) - In like 1/3 runs it fails to start properly
(3) - linux 4.3.0 doesn't like uclibc-based toolchains (net issues)

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

wine: Update download location

Signed-off-by: André Hentschel <nerv@dawncrow.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/python-protobuf: eliminate unneeded build-time dependencies

This was causing unexpected HTTP requests by the setup.py script for
python-protobuf. These "dependencies" aren't actually required for a successful
build, and are not staged into the target install directory.

Fixes:

  http://autobuild.buildroot.org/results/e44/e44694f30e39266491a1040e284e504d6d37ef9e/

Signed-off-by: Steven Noonan <steven@uplinklabs.net>
Tested-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

docs/manual: fix bash export syntax in example code

To export a variable in .bashrc we shouldn't put a '$' at the
beginning of the line, and there must be a '=' between key and value.

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

ffmpeg: mark as not available on nios2

ffmpeg was already marked as not available for the NIOS2 Sourcery
toolchains, but it could still be built with the internal toolchain
backend or a custom external toolchain.

However, an inspection of the latest glibc source code indicates that
FE_INVALID, FE_OVERFLOW and FE_UNDERFLOW and indeed not available in
the nios2 variant of <fenv.h>.

Consequently, this patch makes ffmpeg not available on nios2, which
allows to simplify a bit the dependencies.

It propagates this dependency to:

 - minidlna (and at the same time makes sure the minidlna comment is
   not displayed on nios2, which wasn't properly taken into account
   until now)
 - mpd
 - opencv
 - opencv3
 - squeezelite
 - tovid

Even if it selects ffmpeg, Kodi does not need an update since Kodi is
only available on a limited number of architectures (which don't
include nios2, obviously). Other packages only make use of ffmpeg when
available.

Fixes:

  http://autobuild.buildroot.org/results/921/9212f5a6432c5e695ac0630695405cea05e28610/

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

libssh2: disable on broken nios2 toolchains

Fixes:

  http://autobuild.buildroot.org/results/bcf/bcfd8287a8d3dec2d66aa8191910f4dab0d7be36/

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

package/mplayer: needs host-yasm for mmx support

Fixes
http://autobuild.buildroot.net/results/6a8/6a8c3a70e33cb5b3fc40d036fb0716d915185f33/
http://autobuild.buildroot.net/results/cef/cef30d4e14d39ab64bb27a6514a0390755f4a9d2/
http://autobuild.buildroot.net/results/03b/03b1b42688b07789971a99337708b4dbe4278665/
http://autobuild.buildroot.net/results/5e3/5e3320dfc6793a9ff0494161cdc9a3be795462c4/

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

hostapd: add security patches

Fixes:
CVE-2015-5314 - hostapd: EAP-pwd missing last fragment length validation

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

wpa_supplicant: add security patches

Fixes:
CVE-2015-5310 - wpa_supplicant unauthorized WNM Sleep Mode GTK control
CVE-2015-5315 - wpa_supplicant: EAP-pwd missing last fragment length
validation
CVE-2015-5316 - EAP-pwd peer error path failure on unexpected Confirm
message

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

package/skeleton: ensure custom skeleton uses merged /usr if needed

When using systemd, the policy in Buildroot is to use a merged /usr
(see c5bd8af6, "system: add options for /bin /sbin and /lib to be
symlinks into /usr" for more info). So, we apply a few tricks in some
packages to account for the merged /usr case.

However, when using a custom skeleton, we have no say in how that
skeleton is organised, so it may well have a split /usr. In that case,
our little tricks might not work as expected.

So, when the user uses a custom skeleton and wants systemd as an init
system, we must check that the custom skeleton is setup with a merged
/usr.

We do that by checking that each pair of {/lib,/usr/lib} {/bin,/usr/bin}
and {/sbin,/usr/sbin} have the same inode numbers, i.e. /lib must have
the same inode number as /usr/lib (and so on...). When a pair does not
share the same inode number, this is not a merged /usr and we abort.

We implement that check with make constructs, so it is done very early
in the build process, and we can abort early if need be.

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

package/skeleton: don't handle merged /usr for custom skeleton

In case the user wants to use a custom skeleton, we should not try to
handle the symlinks (resp. mkdir) to handle merged (resp. split) /usr.

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

package/skeleton: qstrip path before using it

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux-headers: bump 3.{10, 14}.x and 4.{1, 2}.x series

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

liquid-dsp: fix powerpc/altivec build failures

When a powerpc target is detected the build system tries to force
altivec support but that's wrong:

Not every powerpc processor has altivec instruction support.

It's using the removed/deprecated -faltivec compiler directive when it
should be using -maltivec instead.

Even if the above conditions are corrected the altivec codepath fails
to build properly.

So just disable altivec in general using the standard
--enable-simdoverride configure option. Fixes:

http://autobuild.buildroot.net/results/e1b/e1b6cf76a7a4ae754a928d211e779e9381198f48/

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

gettext: fix legal info

The license is wrong, it's GPLv3+ rather than GPLv2+ now.
Also there's a specific clause for libintl that makes it LGPLv2.1+, see
gettext-runtime/COPYING.

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

ruby: explicitly disable SSP support

The SSP support checks of the Ruby configure script do not do a full
link, and therefore don't properly detect if SSP support is fully
functional or not, which causes some build failures if the compiler
supports -fstack-protector but the C library does not provide the SSP
implementation.

To avoid this, we explicitly tell ruby if SSP support is available or
not.

Fixes:

  http://autobuild.buildroot.net/results/001/00154b0142f51c197d5d53ae36f7adc4f8250d9d/

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

ruby: do not overwrite RUBY_CONF_ENV

In commit 53bf889cdca77979814bc6b74170e2f104fc3b70 ("ruby: should not
use dl_iterate_phdr() on Blackfin") we added
ac_cv_func_dl_iterate_phdr=no to RUBY_CONF_ENV when building on the
Blackfin architecture.

However, while doing so, we overwrote the previous value of
RUBY_CONF_ENV, which was defined to have custom CFLAGS:

RUBY_CONF_ENV = CFLAGS="$(RUBY_CFLAGS)"

This has no visible effect because the custom CFLAGS are only used on
SuperH/SuperH 64, while the overwrite of RUBY_CONF_ENV takes place on
Blackfin.

However, it doesn't look very pretty, and future additions to
RUBY_CONF_ENV may break this subtle behavior.

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

dhcp: S80dhcp-server: also source /etc/default/dhcpd.conf for INTERFACES

Similar to how we do it for dhcpd.service. The file should really have been
named /etc/default/dhcpd instead, but changing it now would break existing
systemd setups so leave it as it is.

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

dhcp: dhcpd.service: /etc/default/dhcpd.conf is optional

The environment file is for optional customization, so don't error out if it
isn't present.

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

dropbear: dropbear.service: /etc/default/dropbear is optional

The environment file is for optional customization, so don't error out if it
isn't present.

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

libnss: security bump to version 3.20.1

Fixes:
CVE-2015-7181 - A use-after-poison flaw was found in the way NSS parsed
certain ASN.1 structures. An attacker could use this flaw to cause NSS
to crash or execute arbitrary code with the permissions of the user
running an application compiled against the NSS library.
CVE-2015-7182 - A heap-based buffer overflow flaw was found in the way
NSS parsed certain ASN.1 structures. An attacker could use this flaw to
cause NSS to crash or execute arbitrary code with the permissions of the
user running an application compiled against the NSS library.

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

system: Fix warning when selecting systemd and ensure merged /usr workarounds are enabled

With systemd and a custom skeleton, we end up with the following warning:

warning: (BR2_INIT_SYSTEMD) selects BR2_ROOTFS_MERGED_USR which has
unmet direct dependencies (BR2_ROOTFS_SKELETON_DEFAULT)

Which makes sense as the logic to install the symlinks to /usr is part of
the default skeleton handling, but the BR2_ROOTFS_MERGED_USR symbol is ALSO
used by packages to activate various workarounds for a merged /usr, so it
should be always active when using systemd even with a custom rootfs
skeleton (that by definition must use a merged /usr as required by systemd).

So introduce a dummy symbol with the same name for the custom skeleton case,
similar to how we handle the BR2_PACKAGE_BUSYBOX_SHOW_OTHERS option.

This is a hidden option so it can only be activated by enabling systemd (and
not if people just want to use a merged /usr with a custom skeleton but not
systemd), but as this is really a legacy option that is probably good enough
(if not, patches are welcome!).

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

Revert "system: Fix warning when selecting systemd"

This reverts commit b4718c3a28fc3b178253b24bd6fbb8540a724c7e.

The BR2_ROOTFS_MERGED_USR symbol is used to activate various workarounds for
a merged /usr by packages, so it should be active even with a custom rootfs
skeleton.

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

mongrel2: uses dlopen(), needs dynamic library support

Fixes:

  http://autobuild.buildroot.net/results/c5e/c5ecfe4a5ad3712d65293de517a24e76c4376c57/

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

liquid-dsp: work-around compiler bug on SuperH

On SuperH, the BR2_PACKAGE_LIQUID_DSP_FAST option of the liquid-dsp
package causes triggers a compiler bug with the Sourcery CodeBench
toolchains:

  src/filter/src/firdes.c:406:1: internal compiler error: in gen_reg_rtx, at emit-rtl.c:859

This commit works around this problem by simply making this option
unavailable with the problematic toolchains.

Fixes:

  http://autobuild.buildroot.net/results/d43/d43574b8f487648bc440fbe0b63b2a08b309dfc7/

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

boost: fix sparc64 compile, set the correct instruction-set

The default for sparc is v7, which doesn't work for sparc64.
Use ultrasparc for sparc64.
Fixes following autobuild errors:
http://autobuild.buildroot.net/results/0cad0a6eddc153098bb6d1501afd1cf4a57a82d1/

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>