proftpd: Add ftp user creation

In order to remove '/home/ftp' and the ftp user from the 'skeleton'
package, we need to add the creation of the ftp users to the package so
it still work out of the box (with an anonymous user).

Signed-off-by Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Reviewed-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libdrm: update driver options

- add AMD GPU driver (available since 2.4.63)
- freedreno is no longer experimental (since 2.4.59)
- add NVIDIA Tegra driver (available since 2.4.59)
- update help texts (install vs. installs, change to complete
  sentences)

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/efl: remove shared version number

This is necessary to move efl packages to package directory
and prepare the efl version bump to the latest release.

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

qemu: add new board support for qemu-system-sparc64

SPARC64 builds can be tested with Qemu.

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Tested-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

toolchain: add sparc64 architecture support

Introduce sparc64 architecture to buildroot.

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Tested-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

qt: fix build for arm with BR2_STATIC_LIBS

Some Buildroot-generated static toolchains for ARM do not have
dlfcn.h header, which breaks builds of qt-4.8.7 because its
QtCore's plugin loader expects this header to be present (and
dynamic libraries to be supported). For certain platforms without
dynamic libraries, there is a QT_NO_DYNAMIC_LIBRARY constant
defined in plugin loader's source. This patch puts that constant
into Qt's build environment if Buildroot is configured without
shared library support.

Fixes:
- http://autobuild.buildroot.net/results/11a4c19638ca211bd2a424fd92602166235c9433
- http://autobuild.buildroot.net/results/c056e8781558267f6971a5f017f47e90dcf24e07
- http://autobuild.buildroot.net/results/b8f34afa8686e890a1a3882a52081bb6a3b9c13b
and several others.

Signed-off-by: Alexander Lukichev <alexander.lukichev@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

media-ctl: remove package

Package media-ctl is out-dated, source (and developement) have been moved
to v4l-utils since June 2014. Up-to-date version is available in
the libv4l package (BR2_PACKAGE_LIBV4L/BR2_PACKAGE_LIBV4L_UTILS).

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libv4l: update v4l-utils help text (add more listed tools)

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

fbterm: fix musl compile

- add missing include, fixes:

  fbio.cpp:33:8: error: ‘fd_set’ does not name a type
   static fd_set fds;

  improxy.cpp:439:3: error: ‘fd_set’ was not declared in this scope

- add missing WAIT_ANY define, fixes:

  fbterm.cpp: In member function ‘void FbTerm::processSignal(u32)’:
  fbterm.cpp:212:22: error: ‘WAIT_ANY’ was not declared in this scope
      s32 pid = waitpid(WAIT_ANY, 0, WNOHANG);

Fixes:
http://autobuild.buildroot.net/results/ab2/ab2334700b44bac233994b76f2acf1795f0

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

configs/riotboard: bump kernel to version 4.2.3

Signed-off-by: Nikolay Dimitrov <picmaster@mail.bg>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/Config.in: fix alphabetical order

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

package/Config.in: correct capitalization of menu items

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

configs/qemu: add sh4eb (big endian) sample

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

gnuradio: break dependency chain by using 'depends on'

This commit fixes the following recursive dependency:

package/python/Config.in:5:error: recursive dependency detected!
package/python/Config.in:5: symbol BR2_PACKAGE_PYTHON is selected by BR2_PACKAGE_GNURADIO_PYTHON
package/gnuradio/Config.in:46: symbol BR2_PACKAGE_GNURADIO_PYTHON is selected by BR2_PACKAGE_GNURADIO_UTILS
package/gnuradio/Config.in:55: symbol BR2_PACKAGE_GNURADIO_UTILS depends on BR2_PACKAGE_PYTHON_NUMPY_ARCH_SUPPORTS
package/python-numpy/Config.in:1: symbol BR2_PACKAGE_PYTHON_NUMPY_ARCH_SUPPORTS depends on BR2_PACKAGE_PYTHON

by replacing the "select BR2_PACKAGE_PYTHON" by a "depends on
BR2_PACKAGE_PYTHON", and ditto for the BR2_PACKAGE_GNURADIO_PYTHON
selection.

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

arch/mips: remove some remaining mips1/2/3/4 bits

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

gnuradio: add missing dependency to python-numpy for python support

Some gnuradio python blocks need python-numpy to be present. Without
python-numpy, gnuradio-python based apps fails, on runtime, with
traceback like:

Traceback (most recent call last):
  File "./top_block.py", line 18, in <module>
    from gnuradio import analog
  File "/usr/lib/python2.7/site-packages/gnuradio/analog/__init__.py", line 35, in <module>
    from am_demod import *
  File "/usr/lib/python2.7/site-packages/gnuradio/analog/am_demod.py", line 22, in <module>
    from gnuradio import gr
  File "/usr/lib/python2.7/site-packages/gnuradio/gr/__init__.py", line 44, in <module>
    from top_block import *
  File "/usr/lib/python2.7/site-packages/gnuradio/gr/top_block.py", line 30, in <module>
    from hier_block2 import hier_block2
  File "/usr/lib/python2.7/site-packages/gnuradio/gr/hier_block2.py", line 26, in <module>
    import pmt
  File "/usr/lib/python2.7/site-packages/pmt/__init__.py", line 58, in <module>
    from pmt_to_python import pmt_to_python as to_python
  File "/usr/lib/python2.7/site-packages/pmt/pmt_to_python.py", line 22, in <module>
    import numpy
ImportError: No module named numpy

Signed-off-by: Gwenhael Goavec-Merou <gwenhael.goavec-merou@trabucayre.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

opencv3: use BR2_PACKAGE_PYTHON_NUMPY_ARCH_SUPPORTS

Opencv3 select python-numpy. To avoid duplicating architecture
dependencies, an hidden variable has been added to python-numpy
package. This patch propagate this modification to the opencv3
package.

Signed-off-by: Gwenhael Goavec-Merou <gwenhael.goavec-merou@trabucayre.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

python-numpy: introduce ARCH_SUPPORTS hidden variable

Some packages selects python-numpy. This package has a some CPU specific
code. To simplify and avoid duplicate dependencies this patch introduce an
hidden ARCH_SUPPORTS variable.

[Thomas: fix the definition of the new option to make sure its value
is actually 'y' when a supported architecture is used.]

Signed-off-by: Gwenhael Goavec-Merou <gwenhael.goavec-merou@trabucayre.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Config.in.legacy: remove incorrect BR2_LEGACY selects

We shouldn't select BR2_LEGACY for string options.

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

Config.in.legacy: add missing select BR2_LEGACY

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

dbus: bump to version 1.10.0

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Tested-by: Nicolas Cavallari <nicolas.cavallari@green-communications.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

picocom: bump to version 2.0

- Bump to version 2.0
- Switch the download site to GitHub
- Apply an upstream patch to fix a missing define
- Add a hash file
- Rewrap the help text to 72 characters length
- Change the official website

[Thomas: fix s/MAKE_ENV/TARGET_MAKE_ENV/.]

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

collectd: drop rrdtool selects

Drop indirect rrdtool selects, they're already selected by rrdtool
itself so there's no need.

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

postgresql: security bump to version 9.4.5

Fixes:

CVE-2015-5289: json or jsonb input values constructed from arbitrary
user input can crash the PostgreSQL server and cause a denial of
service.

CVE-2015-5288: The crypt() function included with the optional pgCrypto
extension could be exploited to read a few additional bytes of memory.
No working exploit for this issue has been developed.

sparc build fix patch upstream so drop it.

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>

stunnel: bump to version 5.24

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>

bind: bump to version 9.9.8

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>

webkitgtk24: use the correct Config.in option for NEON

In order to enable JIT support on ARM, webkitgtk24 currently looks at
BR2_ARM_ENABLE_NEON, which is not correct: BR2_ARM_CPU_HAS_NEON should
be used instead.

The BR2_ARM_ENABLE_NEON is only visible for cores that select
BR2_ARM_CPU_MAYBE_HAS_NEON, in order to allow to specify if his
particular SoC has chosen to integrate NEON or not. And if so,
BR2_ARM_CPU_HAS_NEON gets selected.

BR2_ARM_CPU_HAS_NEON is in fact selected in two different ways:

 - Either directly by the CPU core selection, if NEON is mandatory in
   this CPU core.

 - Or by BR2_ARM_ENABLE_NEON, for CPU cores where NEON support is
   optional.

So really, BR2_ARM_CPU_HAS_NEON is what should be used by packages.

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

ethtool: bump to version 4.2

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>

swupdate: fix description typo

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

dos2unix: bump to version 7.3.1

Drop upstream patch.

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

mplayer: add license info

Most code is GPLv2+ or compatible except for one file that is
GPLv2-only, hence as a whole is GPLv2 (see Copyright).
Also kill some whitespace and make the only hard dependency
(host-pkgconf) a hard dependency, not an addition, since it's not
conditional.

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

linux-headers: bump 3.12.x series

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

libcurl: bump to version 7.45.0

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

ccache: bump to version 3.2.3

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

xterm: bump to version 320

Also switch to http site.

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

sane-backends: bump to version 1.0.25

- Bump version to 1.0.25
- Update hash file. The hash value provided by the alioth.debian.org URL
  is wrong, so use the one provided by www.sane-project.org. I have
  reported this upstream.
- Remove unnecessary patches:
  0002-remove-unsafe-headers-path.patch
  0003-remove-cups-and-systemd-support.patch

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

freetype: bump to version 2.6.1

- Bump to version 2.6.1.
- Update hash file.
- Remove includes hooks since they are no longed needed.

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

mplayer: bump to version 1.2

- Bump to version 1.2
- Update hash file.
- Remove no longer existent configure options:
  --disable-dvdread-internal, --disable-tremor-internal.
- Tweak 0001-disable-install-strip.patch.
- Remove 0002-fix-cc-parsing.patch. Upstreamed:
  https://github.com/pigoz/mplayer-svn/commit/0f3ddfe146b8b2c6fe9302017375ef6b30ffbe71
- Remove 0004-add-arc-support.patch. Upstreamed:
  https://github.com/pigoz/mplayer-svn/commit/3b8c5f22313a416b24dcc4e7facde0a886352a75
- Remove 0005-Support-newer-GIFLIB-versions.patch. Upstreamed:
  https://github.com/pigoz/mplayer-svn/commit/a0ddaef5457e222dade386901bf448c5e3ac7b89
- Remove 0006-Support-newer-GIFLIB-versions-part2.patch. Upstreamed:
  https://github.com/pigoz/mplayer-svn/commit/a43d259602723d92db1463b166739fb8d22c2d51
- Remove 0007-mplayer-enable-aarch64.patch. Upstreamed:
  https://github.com/pigoz/mplayer-svn/commit/1cceebf1fa0cc1b27429e937f1cfa61c406f9159
- Rename 0003-mpdemux-live555-async-interface.patch to
  0002-mpdemux-live555-async-interface.patch.

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

git: bump to version 2.6.1

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

glib-networking: bump to version 2.46.0

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

libglib2: bump to version 2.46.0

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

dbus-cpp: not available on musl

Fixes:
http://autobuild.buildroot.net/results/9b2/9b29ff3aac5cd543dbe2a1a632b61e27ca25085e/
http://autobuild.buildroot.net/results/304/3041ed6d7ff648f805cf2e29762e2b57cd0540e4/

musl doesn't provide the initializer for recursive pthread mutexes.

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

wandboard_defconfig: bump u-boot to 2015.07 and kernel to 3.14.28_1.0.0

Bump U-boot version to 2015.07 from the mainline repository. Since there
is SPL support for the wandboard in the U-boot 2015.07, it is now
possible to boot the same U-boot image for all wandboard variants
(solo/dual/quad).

Bump kernel version to 3.14.28_1.0.0 from the wandboard git repository.

Tested on Wandboard Solo, Wandboard Dual and Wandboard Quad.

Signed-off-by: Sergio Prado <sergio.prado@e-labworks.com>
Tested-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

glib-networking: specify GIO_MODULE_DIR for target

Since the upgrade to pkgconf 0.9.12 all directories and variables with
directories are sysroot-prefixed, whereas with the old patch only some
variables were (includedir, mapdir, sdkdir, libdir).

libglib2 uses a giomoduledir=${libdir}/gio/modules variable in
gio-2.0.pc, which when expanded via sysroot by pkgconf at configure time
points to the staging directory (via libdir expansion).
When target install gets called the gio module gets installed to staging
rather than target, with the obvious consequences this has.

Work-around this pkgconfig abuse by overriding GIO_MODULE_DIR at target
install time.

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

gdk-pixbuf: security bump to version 2.32.1

CVE-2015-4491 already fixed in the release so drop patch.

Fixes a heap overflow with a gif file and a heap overflow and DoS with a
tga file. CVEs not assigned yet.

Tests disabled since they fail to build properly on uClibc toolchains.

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

python-numpy: uClibc patch to avoid ldexpl and frexpl

It turns out that uClibc only enables the long double math functions
for some architectures (god know why, so what if long double is the
same as double, these functions should still be defined). Since
python-numpy links with those functions, the module will fail to load
on ARM, MIPS and SH.

However, python-numpy actually checks for each function if it is really
available. Only, it overrides that check for ldexpl and frexpl when
long double is the same as double (i.e. in exactly the case the uClibc
doesn't have these functions).

So add another exception for this.

Upstream-status: not applicable (code has changed too much)

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Tested-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

fs/romfs: remove redunant ROMFS_TARGET definition

Probably a leftover from our last fs infra reworking a couple of years
ago.

Incidentally, also added a newline at the end of the file.

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

pkgconf: bump to version 0.9.12

Drop patches, they aren't required any more.
It fixes a bug where inline comments cause issues, like:

Libs: -lblah # -lbleh

...it would make pkgconf search for the '#' package

This is required for the nettle 3.1.1 upgrade since howgeed.pc adjusts
Libs according to static/dynamic builds using the comment feature.

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

gcc: backport xtensa libgcc stack unwinding fixes

These backported patches fix the following failing uClibc-ng tests:
  nptl/tst-cancelx4,
  nptl/tst-cancelx16,
  nptl/tst-cancelx18,
  nptl/tst-cancelx20,
  nptl/tst-cancelx21,
  nptl/tst-cleanupx1,
  nptl/tst-cleanupx3,
  nptl/tst-oncex3,
  nptl/tst-oncex4.

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

nettle: bump to version 3.1.1

API is now compatible with nettle 2.7.x

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

wireless-regdb: bump to version 2015.09.25

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

arm_foundationv8_defconfig: rely on default Image name for kernel

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

qemu_aarch64_virt_defconfig: rely on default Image name for kernel

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

linux: add 'Image' as the image name for aarch64

On aarch64, the image name is always Image, so let's add support for
that.

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

microzed: u-boot needs host-dtc

defconfigs that use U-Boot 2015.07 need host-dtc to build
properly.

Signed-off-by: Davide Viti <d.viti@infosolution.it>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libtorrent: zlib is mandatory

zlib isn't optional even though it seems like it, fixes build failure
such as:

log.cc:51:18: fatal error: zlib.h: No such file or directory

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

rtorrent: drop sigc dep & switch homepage

libsigc isn't used/needed any longer so drop it.
And switch to the proper homepage rather than the tarball repository.

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

libtorrent: drop sigc dep & switch homepage

libsigc isn't used/needed any longer so drop it.
And switch to the proper homepage rather than the tarball repository.

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

toolchain-wrapper.c: unbreak BR_CROSS_PATH_ABS handling

Fixes #8386

We should check if BR_CROSS_PATH_ABS is defined, not if it evalutates to
true for the pre processor.

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

rtorrent: bump to version 0.9.6

- Bump to version 0.9.6

- Add a hash file.

- Change the website URL to the new official one.

- Tweak the 0001-cross_compile.patch for the new version.

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

libtorrent: bump to version 0.13.6

- Bump to version 0.13.6.

- Add a hash file.

- Change the website URL to the new official one.

- Help the build system to find the zlib headers, otherwise the
  configure script will fail like this:

    checking if zlib is wanted... yes
    checking for inflateEnd in -lz... no
    checking zlib.h usability... no
    checking zlib.h presence... no
    checking for zlib.h... no
    configure: error: either specify a valid zlib installation with
    --with-zlib=DIR or disable zlib usage with --without-zlib

- Disable instrumentation support by default since it avoids runtime
  failures on rtorrent for some architectures like MIPS, PowerPC or ARM:

    https://github.com/rakshasa/rtorrent/issues/156

- Tweak the 0001-cross_compile.patch for this new version.

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

linux-headers: bump 3.18.x series

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

gcc: sort the patches before they are hashed

$(wildcard ...) in make doesn't sort the files, so the order of the
hashed files is not predictable. Therefore, the ccache hash could
change from one build to another. We don't want that, so sort the
files explicitly.

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

configs: fix configurations that need host-dtc

olimex_a20_olinuxino_lime_defconfig and zedboard_defconfig use U-Boot
2015.07, and they need host-dtc to build properly.

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

u-boot: add an option to indicate that DTC is needed

Some U-Boot configurations require the Device Tree compiler to be
available, so we need to depend on host-dtc (example configuration:
zynq_zed).

However, we don't want to build it unconditionally, since the vast
majority of U-Boot configurations don't need it (and host-dtc itself
has a bunch of dependencies).

So, we simply add a Config.in option that allows users to indicate
whether their U-Boot needs DTC or not, and depend on host-dtc if this
option is enabled.

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

dtc: add host build

Having a host variant of dtc is needed for example for U-Boot, which
uses the Device Tree for a number of platforms.

In addition, now that we have a proper host-dtc package, it is no
longer needed for the linux package to install the host dtc compiler:
users interested in having the host dtc compiler can simply enable
this package.

A Config.in.host option is added to build host-dtc, because the
initial reason why the host DTC built by the kernel was installed in
$(HOST_DIR)/usr/bin (commit 707d44d0a28906ebda49584dd5f55985406f0bde
from Thomas DS) was:

    Having dtc as a host tool can be useful for users that have a
    custom boot scenario where the device tree is not embedded in the
    kernel.

[Thomas:
  - rework the commit log.]
[Peter: keep the (renamed) linux version]
Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux: install dtc binary as linux-dtc

So it doesn't conflict with host-dtc. The Linux kernel version may be a
patched version supporting E.G. overlays.

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

portaudio: fix dependency on alsa-lib

BR2_PACKAGE_PORTAUDIO_ALSA is misspelled in one place, resulting in
alsa-lib not being listed in the dependencies.

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

webkitgtk24: fix compilation failure due to unicode quotes

As reported in this upstream bug...

https://bugs.webkit.org/show_bug.cgi?id=128971

...the pretty unicode double quotes break Python stdin. Backporting the
patch that was attached to that bug report and tweak it minimally to
make it apply on Buildroot.

Fixes:

  http://autobuild.buildroot.net/results/c8e/c8ea1cab0a1ee081f9bca994ff5309ea0117d692/

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

toolchain/external: commonalise comments about Linaro toolchains

Those two comments:
  - are exactly the same
  - have the same dependencies (except for arm/armeb)

So, make it a common comment. It will be useful to have that comment
when we introduce new Linaro toolchain versions.

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

util-linux: update tool select options

Add missing app config options:
  - line
  - tunelp

Drop app config options no longer supported by util-linux:
  - arch
  - ddate

Signed-off-by: Danomi Manchego <danomimanchego123@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>

parted: add patch to fix musl build

Fixes:

  http://autobuild.buildroot.org/results/e1b/e1b6de7b234f2ec154d772ee93e33275dff38f3e/

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

configs: fix build of calao_qil_a9260_defconfig

The Barebox, AT91Bootstrap and Linux patches for the Calao QIL A9260
board are all located in the same directory, board/calao/qil-a9260/,
with only a prefix to indicate to which component they apply.

Unfortunately, since commit 0eba4759fab9d093afacac77a0ac31f1f0b3cceb
("packages: apply custom patches using *.patch instead of
<pkg>-*.patch"), the logic to apply custom patches in Barebox and
AT91Bootstrap has been changed from applying <package>-*.patch to
applying *.patch (in order to be consistent with what we do for all
packages in Buildroot).

However, this had the effect of trying to apply both the AT91Bootstrap
and Barebox patches on both packages, which obviously cannot work.

This commit fixes that by moving patches around, and switching to use
BR2_GLOBAL_PATCH_DIR for all of Barebox, AT91Bootstrap and Linux
patches.

In addition, the defconfig was not specifying explicitly which Barebox
version to use, so right now it's trying to use Barebox 2015.09, on
which the board-specific patch does not apply. So we've forced to use
Barebox 2012.08, which was the lastest release available at the time
the defconfig was initially contributed.

Cc: Gregory Hermant <gregory.hermant@calao-systems.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

configs: fix build of calao_usb_a9g20_lpw_defconfig

Both the Barebox and AT91Bootstrap patches for the Calao USB A9G20
board are located in the same directory, board/calao/usb-a9g20-lpw/,
with only a prefix making a difference between whether they apply to
Barebox or AT91Bootstrap.

Unfortunately, since commit 0eba4759fab9d093afacac77a0ac31f1f0b3cceb
("packages: apply custom patches using *.patch instead of
<pkg>-*.patch"), the logic to apply custom patches in Barebox and
AT91Bootstrap has been changed from applying <package>-*.patch to
applying *.patch (in order to be consistent with what we do for all
packages in Buildroot).

However, this had the effect of trying to apply both the AT91Bootstrap
and Barebox patches on both packages, which obviously cannot work.

This commit fixes that by moving patches around, and switching to use
BR2_GLOBAL_PATCH_DIR.

Cc: Gregory Hermant <gregory.hermant@calao-systems.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

configs: remove gnublin_defconfig

This defconfig tries to fetch its kernel source code and U-Boot source
code from gitorious.org, which is not longer available. Therefore,
this defconfig has been failing to build since quite some time. Since
there's no obvious other place to grab the kernel and u-boot source
code, and the defconfig was not updated since a long time, let's
simply get rid of it.

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

micropython: needs dynamic library

Fixes:

   http://autobuild.buildroot.org/results/06f/06fc79dad3d4c0c5f997f4c2e4b52014727764ac/

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

python-numpy: re-enable for uClibc

In commit b0f025386f6861fdc21e38e3a33b65669276f2c2 a patch was added
to make sure that python-numpy does work in uClibc. This patch was
needed for the x86 architecture, because uClibc in fact does support
fenv for x86. However, we don't turn on this support in our uClibc
config. Because the Config.in depends still allowed uClibc for x86,
this lead to a built failure, which lead to the fix in commit b0f02538.
But since this fix also fixes it for all other (supported)
architectures, we can just remove the !uClibc dependency completely.

This is also propagated to the reverse dependency in opencv3.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Gwenhael Goavec-Merou <gwenhael.goavec-merou@trabucayre.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

nfs-utils: systemd support

nfs-utils has been patched to install only the needed systemd unit files
supported.
Also tmpfiles was needed because the folder /run/nfs is a tmpfs
and nfs-utils need some files to be there to work correctly.

Signed-off-by: Aurélien Chabot <aurelien@chabot.fr>
[Maxime: Remove patches included originally in the patch as they have
have been merged upstream in v1.3.3]
Signed-off-by Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
[Thomas:
  - add comment in nfs-utils_env.sh to explain why an empty script is
    needed.
  - remove incorrect indentation.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

nfs-utils: Bump version

Remove patch as it was backported from upstream:
http://git.linux-nfs.org/?p=steved/nfs-utils.git;a=commit;h=17a3e5bffb7110d46de1bf42b64b90713ff5ea50

Signed-off-by Maxime Hadjinlian <maxime.hadjinlian@gmail.com>

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

package/schifra: remove package

Signed-off-by: Romain Naour <romain.naour@openwide.fr>
Cc: Arnaud Rébillout <rebillout@syscom.ch>
Cc: Simon Dawson <spdawson@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

devmem2: add a hash file

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

squashfs: Add lz4 hc compression

lz4 offers a high compression mode. A minimalistic file system shrinks by about 15%.

[Peter: Unconditionally use hc mode for lz4 instead of introducing another option]
Signed-off-by: Matthias Weisser <m.weisser.m@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

mpdecimal: fix comment about fenv.h

We're disabling fenv.h usage on i386 because it is not available in
musl/uclibc (and not musl/glibc as was written).

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

mpdecimal: use official upstream site

The official upstream site is available again, so let's switch to
using it.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Tested-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Revert "opencv: mark as not available on Xtensa"

Now that xtensa toolchain is able to compile huge functions this
workaround is no longer needed.

This reverts commit d21932508e69d4ab849cba9958f953ae9342cb8c.

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

xtensa: switch from text-section-literals to auto-litpools

Now that both binutils and gcc support auto-litpools use that option
instead of text-section-literals to be able to compile huge functions.

Fixes:
  http://autobuild.buildroot.net/results/dd384fe0ef02a4205bea66a4a16ca2062afe53b4/
  http://autobuild.buildroot.net/results/87dd357a4b883ea3cd75546b3d63c4c28245beee/
  http://autobuild.buildroot.net/results/b5bca00dec1ecb118c7fb9c10dee74c94809c831/
and many others.

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

gcc: backport mauto-litpools xtensa option

With support from assembler this option allows compiling huge functions,
where single literal pool at the beginning of a function may not be
reachable by L32R instructions at its end.

Currently assembler --auto-litpools option cannot deal with literals
used from multiple locations separated by more than 256 KBytes of code.
Don't turn constants into literals, instead use MOVI instruction to load
them into registers and let the assembler turn them into literals as
necessary.

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

binutils: backport auto-litpools xtensa gas option

Auto-litpools is the automated version of text-section-literals: literal
pool candidate frags are planted every N frags and during relaxation
they are turned into actual literal pools where literals are moved to
become reachable for their first reference by L32R instruction.

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

gcc: remove the --with-pkgversion option from the ccache hash

One of the gcc configure options that we hash for ccache is
--with-pkgversion which is set to something like Buildroot
2015.11-git-00426-ge7e7e4f - i.e., it will change with every buildroot
commit. That's obviously not wanted, so substitute this away.

Also add a \n to the printf so the output is a bit more readable.

[Peter: update documentation to match]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

ccache: support changing the output directory

When building in a different output directory than the original build,
there will currently be a lot of ccache misses because in many cases
there is some -I/... absolute path in the compilation. Ccache has an
option CCACHE_BASEDIR to substitute absolute paths with relative paths,
so they wil be the same in the hash (and in the output).

Since there are some disadvantages to this path rewriting, it is made
optional as BR2_CCACHE_USE_BASEDIR. It defaults to y because the
usefulness of ccache is severely reduced without this option.

In addition to CCACHE_BASEDIR, we also substitute away the occurences
of $(HOST_DIR) in the calculation of the compiler hash. This is done
regardless of the setting of BR2_CCACHE_USE_BASEDIR because it's
quite harmless.

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

ccache: use mtime for external toolchain, CONF_OPTS for internal toolchain

Our current ccache disables hashing of the compiler executable itself,
because using the default 'mtime' doesn't work in buildroot: we always
rebuild the compiler, so the mtime is always different, so the cache
always misses.

However, in the current situation, if a user changes the compiler
configuration (which would result in the compiler generating different
object files than before) and does 'make clean all', ccache may in fact
reuse object files from the previous run. This rarely gives problems,
because
(1) the cache expires quite quickly (it's only 1GB by default),
(2) radically changing compiler options will cause cache misses because
    different header files are used,
(3) many compiler changes (e.g. changing -mtune) have little practical
    effect because the resulting code is usually still compatible,
(4) we currently don't use CCACHE_BASEDIR, and almost all object files
    will contain an absolute path (e.g. in debug info), so when
    building in a different directory, most of it will miss,
(5) we do mostly build test, and many of the potential problems only
    appear at runtime.
Still, when ccache _does_ use the wrong cached object files, the
effects are really weird and hard to debug. Also, we want reproducible
builds and obviously the above makes builds non-reproducible. So we
have a FAQ entry that warns against using ccache and tells the user to
clear the cache in case of problems.

Now that ccache is called from the toolchain wrapper, it is in fact
possible to at least use the 'mtime' compiler hash for the external
toolchain and for the host-gcc. Indeed, in this case, the compiler
executable comes from a tarball so the mtime will be a good reference
for its state. Therefore, the patch (sed script) that changes the
default from 'mtime' to 'none' is removed.

For the internal toolchain, we can do better by providing a hash of
the relevant toolchain options. We are only interested in things that
affect the compiler itself, because ccache also processes the header
files and it doesn't look at libraries because it doesn't cache the
link step, just compilation. Everything that affects the compiler
itself can nicely be summarised in $(HOST_GCC_FINAL_CONF_OPTS). Of
course, also the compiler source itself is relevant, so the source
tarball and all the patches are included in the hash. For this purpose,
a new HOST_GCC_XTENSA_OVERLAY_TAR is introduced.

The following procedure tests the ccache behaviour:

 Use this defconfig:
BR2_arm=y
BR2_CCACHE=y

 make
 readelf -A output/build/uclibc-1.0.6/libc/signal/signal.os
-> Tag_CPU_name: "ARM926EJ-S"

 Now make menuconfig, change variant into BR2_cortex_a9

 make clean; make
 readelf -A output/build/uclibc-1.0.6/libc/signal/signal.os
-> Tag_CPU_name: "ARM926EJ-S"
 should be "Cortex-A9"

After this commit, it is "Cortex-A9".

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Danomi Manchego <danomimanchego123@gmail.com>
Cc: Károly Kasza <kaszak@gmail.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Cc: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

toolchain-wrapper: support change of BR2_CCACHE

By moving the ccache call to the toolchain wrapper, the following
scenario no longer works:

make foo-dirclean all BR2_CCACHE=

That's a sometimes useful call to check if some failure is perhaps
caused by ccache.

We can enable this scenario again by exporting BR_NO_CCACHE when
BR2_CCACHE is not set, and by handling this in the toolchain wrapper.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Tested-by: Romain Naour <romain.naour@openwide.fr>
Reviewed-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Makefile.in: remove now-unused TARGET_CC/CXX_NOCCACHE

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

qpid-proton: remove now-redundant ccache handling patch

Although this patch is still relevant for upstream, we don't need it
anymore in buildoort because we no longer have _ARG1.

Note that it would be relevant again if we would add host-qpid-proton.
However, the 0.10 release already has the fix.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package-cmake: remove now-redundant target ccache support

All the complexity with the different ways that CMAKE_C_COMPILER and
CMAKE_C_COMPILER_ARG1 can be set are no longer needed, it's all handled
by the toolchain wrapper now.

Note that it is still necessary to handle this for the host build.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

qt5base: remove now-redundant ccache handling

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

cryptodev-linux: remove now-redundant fix-ccache-compile patch

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

barebox: remove now-redundant ccache handling

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

uboot: remove now-redundant ccache handling

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux: remove now-redundant ccache handling

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>