linux: bump default to version 4.2

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

linux-headers: add 4.2.x series

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

toolchain: add 4.2.x choice for headers

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

Merge branch 'next'

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

Kickoff 2015.11 cycle

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

docs/website/news.html: add 2015.08 announcement link

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

Update for 2015.08

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

CHANGES: Update with recent changes

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

exfat: fix build with musl

Fixes:
http://autobuild.buildroot.net/results/e1d/e1db07f0ea1e70c62f3294016c1b3a094de71d12/

The endianness handling functions in platform.h are protected behind ifdef
__GLIBC__ which musl doesn't define even though it does provide the
endianness handling interface.  Work around it by ensuring __GLIBC__ is
defined.

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

packages/lm-sensors: fix compilation with musl

Fixes:
http://autobuild.buildroot.net/results/f72/f72ae17cea910a1dbd3d5d4d09cfbc90d9ba8dc0/

Imports a patch from Alpine Linux to remove __GLIBC__ conditional
compilation. Retested with both musl, glibc and uclibc.

Patch accepted upstream.

Signed-off-by: Brendan Heading <brendanheading@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

CHANGES: Update with recent changes

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

x264: unbreak x86 build after version bump

Fixes:
http://autobuild.buildroot.net/results/3ec/3ec54f722d6008fc422540d3a5462b306d16e84c/

The recent x264 version bump broke the configure step on x86/x86-64 as x264
ends up using gas instead of yasm as assembler.  The reason for this is the
recent upstream commit to optionally use nasm instead of yasm if AS= is
passed:

commit b568a256b9bc6c500d7b1ffe4b9c3311ee5ff337
Author: Henrik Gramner <henrik@gramner.com>
Date:   Sat May 23 19:44:16 2015 +0200

    x86: Experimental nasm support

    Enables the use of nasm as an alternative to yasm.

    Note that nasm cannot assemble x264 with PIC enabled since it currently doesn't
    support [symbol-$$] addressing which is used extensively by x264's PIC code.
    This includes all 64-bit Windows and 64-bit OS X builds, even non-shared.

    For the above reason nasm is currently intentionally not auto-detected, instead
    the assembler must be explicitly specified using "AS=nasm ./configure".

    Also drop -O2 from ASFLAGS since it's simply ignored anyway.

But as we pass AS=$(TARGET_AS) it ends up using gas instead. Fix it by
explicitly passing AS=yasm instead.

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

package/bootutils: fix musl compilation issue

Fixes:
http://autobuild.buildroot.net/results/389/38914a0a7ff0aa79acbbbe6c46ab701325dfceb8/
http://autobuild.buildroot.net/results/c0d/c0d09f5dd0e26c0d2454ebb2ea5317db1135db89/

musl requires strict inclusion of <sys/types.h> for makedev(3).

note - upstream project appears dead since 2009.

Signed-off-by: Brendan Heading <brendanheading@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/x11r7/xserver_xorg-server: dri needs arm >= v6

Fixes

  CC       dri.lo
/tmp/ccc6IbbW.s: Assembler messages:
/tmp/ccc6IbbW.s:3114: Error: selected processor does not support ARM mode `ldrex r2,[r1]'
/tmp/ccc6IbbW.s:3117: Error: selected processor does not support ARM mode `strexeq r2,r0,[r1]'
/tmp/ccc6IbbW.s:3273: Error: selected processor does not support ARM mode `ldrex r2,[r1]'
/tmp/ccc6IbbW.s:3276: Error: selected processor does not support ARM mode `strexeq r2,r3,[r1]'
/tmp/ccc6IbbW.s:3352: Error: selected processor does not support ARM mode `ldrex r1,[r2]'
/tmp/ccc6IbbW.s:3355: Error: selected processor does not support ARM mode `strexeq r1,r0,[r2]'
/tmp/ccc6IbbW.s:3451: Error: selected processor does not support ARM mode `ldrex r3,[r2]'
/tmp/ccc6IbbW.s:3454: Error: selected processor does not support ARM mode `strexeq r3,ip,[r2]'
/tmp/ccc6IbbW.s:3522: Error: selected processor does not support ARM mode `ldrex r3,[r0]'
/tmp/ccc6IbbW.s:3525: Error: selected processor does not support ARM mode `strexeq r3,r1,[r0]'
Makefile:653: recipe for target 'dri.lo' failed
make[5]: *** [dri.lo] Error 1
make[5]: Leaving directory '/home/buildroot/buildroot/output/build/xserver_xorg-server-1.17.2/hw/xfree86/dri'

using this defconfig
BR2_arm=y
BR2_TOOLCHAIN_BUILDROOT_WCHAR=y
BR2_TOOLCHAIN_BUILDROOT_CXX=y
BR2_ROOTFS_DEVICE_CREATION_DYNAMIC_EUDEV=y
BR2_PACKAGE_MESA3D=y
BR2_PACKAGE_MESA3D_DRI_DRIVER_SWRAST=y
BR2_PACKAGE_XORG7=y
BR2_PACKAGE_XSERVER_XORG_SERVER=y

[Peter: fix conditional, add comment explaining issue]
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/x264: Fix aarch64 support, bump version

The current x264 revision fails to include asm code from common/aarch64
which is needed by packages linking to x264, like ffmpeg:

Quoting http://autobuild.buildroot.net/results/189/189f215bcbc510cd292c0f337fe6b39979f01813//ffmpeg-2.7.2/config.log

/home/test/autobuild/instance-3/output/host/usr/aarch64-buildroot-linux-gnu/sysroot/usr/lib/libx264.so:
 undefined reference to `x264_deblock_h_chroma_422_intra_neon'

Function "chroma_422_intra_neon" is defined in
x264/common/aarch64/deblock-a.S. The code in
x264/common/aarch64/* is only compiled if an
assembler was found:
http://git.videolan.org/?p=x264.git;a=blob;f=Makefile;h=6193c5936939fda36f9a2e00c734223318ccda14;hb=HEAD#l134

To allow this add BR2_aarch64 as an assembler-enabled arch.

This however breaks compilation due to broken NEON detection:
"no NEON support, try adding -mfpu=neon to CFLAGS
If you really want to run on such a CPU, configure with --disable-asm."

The bug was fixed upstream Jul, 25th, 2015:
http://git.videolan.org/?p=x264.git;a=commit;h=73ae2d11d472d0eb3b7c218dc1659db32f649b14

When we already have to do a version bump let's take the latest one ;)

Fixes
http://autobuild.buildroot.net/results/90a/90ae4b2fa8a6cff0f18abea2b1c45282afe99e4b/
http://autobuild.buildroot.net/results/109/109686cbc8327e62f757db576cfff49c490b52ae/
http://autobuild.buildroot.net/results/189/189f215bcbc510cd292c0f337fe6b39979f01813/

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

libssh: new package

libssh2 and libssh both implement SSH and provide a library API for apps.
Both support SSH, SFTP, auth, channels etc. Both are 25K-30K lines of code.

[libssh2 vs libssh - A comparison]
http://www.libssh2.org/libssh2-vs-libssh.html

[Peter: add sha256 hash, fix typo in help text]
Signed-off-by: Scott Fan <fancp2007@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

u-boot: 2015.07 - fix creation of .config

Due to recent changes in U-Boot, see
http://git.denx.de/?p=u-boot.git;a=commit;h=a26cd04920dc069fd6e91abb785426cf6c29f45f
re-creation of .config from defconfig by "make oldconfig" happened
incorrectly.

Default prompt for target selection was set as No that lead to missing platform
selection by automated scripts like that:
------------------>8------------------
"yes "" | make oldconfig" on defconfig
------------------>8------------------

And that lead to build failure:
------------------>8------------------
$ make
scripts/kconfig/conf --silentoldconfig Kconfig
  CHK     include/config.h
  UPD     include/config.h
  GEN     include/autoconf.mk
In file included from ./include/common.h:18:0:
include/config.h:5:22: fatal error: configs/.h: No such file or directory
 #include <configs/.h>
                      ^
compilation terminated.
scripts/Makefile.autoconf:72: recipe for target 'include/autoconf.mk' failed
make[1]: *** [include/autoconf.mk] Error 1
------------------>8------------------

This patch reverts mentioned change making possible to build U-Boot
2015.07 in Buildroot.

Note there's a probability that this particular patch won't make its way
in upstream U-Boot and better solution will be found there. But for now
we need to fix U-Boot building in Buildroot anyways.

Hopefully for the next release this patch won't be necessary.

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

package/dropbear: add warning about volatile keys

When the rootfs is read-only, keys will be generated in a volatile
location, which is inherently bad as host keys will change on each boot,
rendering them virtually useless.

Add a warning so the user is at least aware of the issue.

Hide the rm output to avoid noisy output, now that we have a proper warning.

Move the starting message after the symlink-block, to avoid messages
collision. Move the umask as well, since /etc/dropbear/ may be world
readable; just the private host keys should be ?00 (and dropbear handles
that by itself).

[Peter: minor tweaks to commit message]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/Config.in: fix alphabetic order

Commit ebf3a6117c58 (package/libyuv: new package, 2015-08-11) added a
reference to the libyuv Config.in file at the wrong place. Fix that.

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

gnuradio: sys/types.h is needed for mode_t

fix :
http://autobuild.buildroot.net/results/8e3/8e3afd5f22312f0660f631fdb82bbb64eda913fb/build-end.log

error: 'mode_t' has not been declared

Signed-off-by: Gwenhael Goavec-Merou <gwenhael.goavec-merou@trabucayre.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/tvheadend: needs dynamic libraries

Fixes
http://autobuild.buildroot.net/results/f1c/f1c40ac9dda5ceeb5665d021333058eb29828d62/
http://autobuild.buildroot.net/results/bd0/bd0e8619df7cd8f21da96ae470691b1ebb2f156c/
http://autobuild.buildroot.net/results/f1c/f1c40ac9dda5ceeb5665d021333058eb29828d62/

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

package/x11r7/xserver_xorg-server: Fix glamor dependency

X.org xserver depends on libepoxy for glamor support, which depends on
EGL support. Mesa3d is not the only possible EGL provider therefore
change the dependency check.

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

package/sp-oops-extract: new package

A tool for extracting OOPS/panic logs from MTD.

Tested using arm-buildroot-linux-gnueabihf toolchain.

[Thomas:
 - use sp-oops-extract instead of sp_oops_extract as the Config.in
   prompt and in the .mk file comment
 - remove @ in front of the build and install commands.
 - use $(TARGET_CONFIGURE_OPTS) instead of manually passing CC, AR,
   LD, CFLAGS and LDFLAGS.]

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

ibrdtn-tools: new package

[Thomas:
  - include package/ibrdtn-tools/Config.in from package/Config.in
  - fix the Config.in option name: we don't use - in option names, it
    should be replaced by _. Also do the same change in the .mk file.
  - from Config.in, don't select BR2_PACKAGE_IBRDTND (it's not a
    dependency of ibrdtn-tools), but instead select
    BR2_PACKAGE_IBRCOMMON and BR2_PACKAGE_IBRDTN.
  - in the Config.in file, fix the comment dependency logic (we want
    the comment to be shown when either thread support *or* C++
    support is missing)
  - add hash file.
  - do not install to staging, since the package does not install any
    library
  - remove IBRDTN_TOOLS_SOURCE variable, since it's equal to the
    default value
  - add ibrcommon, ibrdtn and host-pkgconf as mandatory dependencies,
    and handle libdaemon and libarchive as optional dependencies.
  - remove custom INSTALL_STAGING_OPTS and INSTALL_TARGET_OPTS:
    automake is used, so there's no need for any hack.]

Signed-off-by: Tom Sparks <tom_a_sparks@yahoo.com.au>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ibrdtnd: new package

[Thomas:
  - add package to package/Config.in
  - remove "select" on libdaemon, libcurl, sqlite and openssl from
    Config.in, since those are optional dependency.
  - add "select" on ibrcommon, since it is a mandatory dependency.
  - remove bogus includes of package/ibrcommon/Config.in and
    package/ibrdtn/Config.in, those are directly from
    package/Config.in.
  - fix Config.in comment dependency and indentation.
  - remove INSTALL_STAGING = YES, since ibrdtnd does not install a
    library.
  - remove libdaemon, libcurl, sqlite and openssl as mandatory
    dependencies, and handle them as optional dependencies.
  - add ibrcommon in the dependencies (even if ibrdtn already depends
    on it)
  - remove custom INSTALL_STAGING_OPTS and INSTALL_TARGET_OPTS, they
    are not needed, since the Makefiles are properly generated by
    automake.
  - explicitly disable features for which Buildroot doesn't have the
    necessary dependencies (dtndht, wifip2p, vmime)
  - add hash file.]

Signed-off-by: Tom Sparks <tom_a_sparks@yahoo.com.au>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ibrdtn: fix bogus dependency on zlib

Due to a copy/paste mistake, zlib was being added to
<pkg>_DEPENDENCIES even when not enabled in the configuration. This
commit fixes that.

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

ibrcommon: add missing dependency on host-pkgconf

pkg-config is used by ibrcommon to detect the availability of openssl,
libnl and libxml2, so we must depend on host-pkgconf.

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

ibrdtn: new package

[Thomas:
  - Add missing include in package/Config.in to make the package
    appear in menuconfig.
  - Fix the comment dependencies: the comment should be shown either
    if C++ is not available *or* if threads are not supported.
  - Add a hash file.
  - Add missing dependency on host-pkgconf, since pkg-config is used
    to detect the availability of ibrcommon.
  - Remove the custom INSTALL_STAGING_OPTS and INSTALL_TARGET_OPTS:
    they are not needed since automake is used.
  - Make --with-compression conditional on whether zlib is available.
  - Add optional dependency on libglib2.
  - Add missing final newlines in .mk and Config.in files.]

Signed-off-by: Tom Sparks <tom_a_sparks@yahoo.com.au>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

ibrcommon: add hash file

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

ibrcommon: fix dependencies of comment in Config.in

We want the comment to be displayed either if C++ is not supported
*or* if threads is not supported.

Also, fix the indentation to use a tab.

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

gdb: use 7.9 by default

Now that we have added 7.10, let's switch to 7.9 as the default gdb
version.

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

gdb: add support for gdb 7.10

This commit adds support for gdb 7.10.

gdb 7.9.1 carries the following patches:

 - 0001-gdbserver-fix-uClibc-whithout-MMU.patch
   merged as 6282837972a5c7b89968319caf821fcbd2a166bb, and part of 7.10

 - 0002-gdbserver-xtensa-drop-xtensa_usrregs_info.patch
   merged as deb44829ecc1dd38275af0fcf91acd319e227a89, and part of 7.10

 - 0003-gdbserver-xtensa-fix-typo-in-XCHAL_HAVE_LOOPS.patch
   merged as a2d5a9d76f2366ed93095fc5a63eafa06b22f808, and part of 7.10

Consequently, none of the three patches we have for 7.9.1 are useful
for 7.10.

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

ibrcommon: new package

[Thomas:
  - add package in package/Config.in, to make it visible in menuconfig
  - make the openssl, libnl and libxml2 dependencies optional, since
    they are definitely not mandatory
  - add README to the license files, since it contains useful
    licensing related information.
  - remove custom INSTALL_STAGING_OPTS and INSTALL_TARGET_OPTS since
    the package uses automake.
  - add missing final newline in Config.in and .mk file.]

Signed-off-by: Tom Sparks <tom_a_sparks@yahoo.com.au>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/libldns: new package

Needed by freeswitch.

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

package/libsoundtouch: new package

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

package/libopenh264: new package

[Thomas:
  - make the license info more specific: it's BSD-2c, not just BSD.
  - use a single conditional block to test the architecture, instead
    of multiple separate conditions.
  - add missing arm/armeb handling in the architecture condition.
  - make the host-nasm dependency only used on x86/x86-64, since it's
    not used for other architectures
  - group the ARCH= and ENABLE64BIT= make variable definitions in a
    LIBOPENH264_MAKE_OPTS variable, and use them at install time (in
    addition to build time) to avoid build issues.
  - wrap too long lines in the .mk file.]

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

package/libilbc: new package

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

package/libg7221: new package

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

package/setools: needs wchar

Even though setools itself does not require wide chars, the configure
script needs them when testing for C99 support.

Since:
  - it is not clear whether wchar are mandatory for C99 or not,
  - only uClibc may miss support for wide chars,
  - using SELinux would probably involve rather big systems (i.e. glibc
    would not be such a huge overhead),
just make setools depend on the toolchain having widechar.

Fixes:
    http://autobuild.buildroot.org/results/051/051a7f72ebadd56407700c22b57e942f08eb1c09/
    http://autobuild.buildroot.org/results/a4b/a4b2a6f1920430af43c23239de17200e70951b3b/
    ...

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Matt Weber <matthew.weber@rockwellcollins.com>
Cc: Clayton Shotwell <clayton.shotwell@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/libcodec2: new package

[Thomas:
 - replace patches from Bernd by patches that are Git formatted and
   have a chance of being upstream.
 - remove the no longer necessary post configure hook to build
   generate_cookbook for the host
 - pass --disable-unittests to disable unit tests.]

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

package/libbroadvoice: new package

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

eigen: use mercurial tag

Now that our Mercurial download mechanism has been fixed to also
support tags, let's use this possibility in the eigen package.

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

support/download: fix the Hg backend for tags

When the version of a package is a Mercurial tag, the download fails,
with:
    abort: unknown revision 'X.Y.Z'!

This is because, in Mercurial, tags are commits like the others, and
when we clone, we actively request a tag. But then, the server
"dereferences" that tag and sends us the revision pointed to by that
tag. Of course, since the tag is a commit after the revision we got,
we do not have the revision adding the tag.

So, we just have to download the full repository to be sure we have
the tags in our local clone.

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

ipkg: add patch to fix musl build

Like Romain Naour did for util-linux, this commit adds a patch for the
ipkg package to remove the use of the __P, which is useless for modern
compilers, and not defined by the musl C library.

Fixes:

  http://autobuild.buildroot.net/results/c27d1d1e6a9bb52fead7b48625bb456f42b2f056/

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

curlftpfs: fix musl build

Add a patch to use off_t instead of __off_t.

Fixes:

  http://autobuild.buildroot.net/results/ed5/ed5b1a49304f115d793da332de6eef8cdaffbc8f/

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

openssh: fix static compilation

PIE and static doesn't work on Linux.

Fixes:
http://autobuild.buildroot.net/results/dce/dce0202e039f4636d68532c4aab8738938b76650/

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

docs/website: remove broken symlink for top-level manual

Recently, we've tried to have the manual lie at the top-level of the
site, to get a shorter URL.

However, that is broken: we do not need only a link the the HTML, but
also to the CSS and to any resource linked from the html (images...).

The server configuration has now been updated to add a redirect, so we
no longer need that symlink anyway. Remove it.

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

docs/manual: makedevs only accepts numerical values

Because of a limitation in how makedevs resolves names [0], usernames
and groupnames would be resolved with UIDs and GIDs from the host, not
the target.

Fixing makedevs would involve us rewriting a /etc/passwd and /etc/group
parser (not too complex, still not trivial).

Document that the user and group fields should only be numbers, not
names. Also, the manual is currently misleading, as an example indeed
uses names for the user and group fields. Fix that and expand the
explanations.

The mode field should also be numbers, not a rwx combo.

Slight typographical eye-candy about major/minor fields.

[0] it uses getpwnma(3) and getgrnam(3) to resolve names to IDs, and
those only resolve names on the host (i.e. there is no way to specify
a chroot-like feature)

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

package/cppcms: bump to 1.0.5

Bump cppcms since it seems to fixe an issue with gcc5+.
The new release has been build tested against a failing Buildroot config.

Fixes:
http://autobuild.buildroot.net/results/d9d/d9d3f40bf0a203e052d2a43ac7a6a1dfa60379b3/build-end.log

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

package/libsilk: new package

[Thomas: fix license, it's BSD-3c.]

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

package/libyuv: new package

[Thomas: fix license, it's BSD-3c, not BSD-4c.]

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

mtd: add mtdpart option

As pointed by Jörg Krause one of the novelties of the 1.5.2 release is
the new mtdpart tool, so add an option for it.

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

sysvinit: update website URL

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

package/libhttpparser: New package

http-parser provides a simple C API to parse HTTP responses and
requests.

This packaging uses the official Joyent repository.

[Thomas:
 - since we're always building the shared library, depend on
   !BR2_STATIC_LIBS in Config.in. We could build only the static
   library, but then we would have to do all the lib installation
   manually.
 - indent using tabs in Config.in
 - use 'make install' instead of handcoding the library installation.
 - use $(TARGET_CONFIGURE_OPTS) instead of just CC= and LD=.
 - use tabs for indentation in commands instead of spaces
 - add patch to fix reinstallation.]

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

perl-http-negotiate: fix dependencies

regenerated by support/scripts/scancpan

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

perl-file-listing: fix dependencies

regenerated by support/scripts/scancpan

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

package/qt5/qt5base: fix build with ccache

Building with ccache failed with:

    Running configuration tests...
    Failed to process makespec for platform 'devices/linux-buildroot-g++'
    Project ERROR: Compiler <path_to_output_dir>/host/usr/bin/ccache <path_to_output_dir>/host/usr/bin/<cross_compile>-g++ not found. Check the value of CROSS_COMPILE -device-option
    Could not read qmake configuration file <path_to_output_dir>/build/qt5base-5.5.0/mkspecs/devices/linux-buildroot-g++/qmake.conf.
    Error processing project file: /dev/null

This was caused by Buildroot setting this in
qt5base-5.5.0/mkspecs/devices/linux-buildroot-g++/qmake.conf:

    QMAKE_CXX               = $${BR_CCACHE} $${CROSS_COMPILE}g++

But qt5base-5.5.0/mkspecs/features/device_config.prf expects QMAKE_CXX
to be a single valid (absolute or QMAKE_PATH_ENV-relative) path to an
existing file, which is not possible if using ccache as above.

Add a patch fixing this by testing only the first value in QMAKE_CXX.

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

python-can: remove incorrect dependency

When using the python-package infrastructure, there is no need for
packages to declare a dependency on the Python interpreter: the
package infrastructure does it automatically.

Moreover, this is actually broken in the case of python-can, which can
be selected either with Python 2.x or Python 3.x. If the latter is
chosen, python-can will still trigger the build of Python 2.x, which
is incorrect.

This was discovered by Vicente Olivert Riera during the analysis of
the following build failure:

  http://autobuild.buildroot.net/results/aff/affb1d4a328c479be73b7818364db5914bf3d376/

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

scrypt: commit packaging improvements

The changes brought by this commit should have been part of the
previous commit, but were not due to a mistake. This commit implements
the following changes to the scrypt package:

   - move from "Libraries -> Crypto" to "System tools", since it only
     installs one binary, scrypt, and no library.
   - bump from 1.1.6 to 1.2.0
   - add comment in the .hash file explaining where the hash is coming
     from.
   - add missing dependency on OpenSSL
   - use BSD-2c as the license code, instead of BSD2
   - use 'main.c' as the license file.

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

scrypt: new package

[Thomas:
  - move from "Libraries -> Crypto" to "System tools", since it only
    installs one binary, scrypt, and no library.
  - bump from 1.1.6 to 1.2.0
  - add comment in the .hash file explaining where the hash is coming
    from.
  - add missing dependency on OpenSSL
  - use BSD-2c as the license code, instead of BSD2
  - use 'main.c' as the license file.]

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

package/kodi-visualisation-waveforhue: bump version

Also remove patch applied upstream:
https://github.com/notspiff/visualization.waveforhue/commit/8824f4fa384693da0dfb3df5c30992bfe0fee9a4

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

package/kodi: bump to version 15.1-Isengard

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

package/libdrm: bump to version 2.4.64

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

package/{mesa3d, mesa3d-headers}: bump version to 10.6.4

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

libpam-tacplus: new package

[Thomas:
  - change the inclusion location in package/Config.in to be with the
    other linux-pam plugins.
  - fix indentation of the Config.in file.
  - fix build of the package when a toolchain without SSP support is
    used.
  - fix installation location of the PAM module: it was installed in
    /usr/lib/security, while all other PAM modules are in
    /lib/security.
  - adjust the ordering of the variables in the .mk file to be a bit
    more logical.
  - remove passing of $(TARGET_CONFIGURE_OPTS) in <pkg>_CONF_ENV as it
    is not needed: it is already done by the autotools-package
    infrastructure.]

Signed-off-by: Giovanni Zantedeschi <giovanni.zantedeschi@datacom.ind.br>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libpam-radius-auth: new package

[Thomas:
  - make libpam-radius-auth depend on linux-pam (since it's a
    linux-pam plugin, it needs some header files from linux-pam) by
    adding a "if BR2_PACKAGE_LINUX_PAM...endif" block in
    package/Config.in. This way, we will be grouping all the Linux PAM
    plugins together.
  - Fix the indententation in the Config.in file.
  - Use a better Config.in help text, copy/pasted from the website.
  - Use a tarball instead of github. This allows to remove
    AUTORECONF=YES.
  - Add linux-pam as a dependency.
  - Use $(TARGET_CONFIGURE_OPTS) instead of passing just CC/LD, but
    pass it in the environment rather than as options so that the
    CFLAGS/LDFLAGS added by the package Makefile are still taken into
    account.
  - Use mkdir -p + cp -dpfr instead of $(INSTALL) -D since we are
    copying multiple files.
  - Add a hash file.]

Signed-off-by: Giovanni Zantedeschi <giovanni.zantedeschi@datacom.ind.br>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/libcgroup: allow to build cgroup-tools

Add a new config option that allows to build cgroup-tools - a set of
command-line utils for managing cgroups.

[Thomas: slightly change the prompt of the new Config.in option.]

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

package/qt5/qt5multimedia: fix gstreamer1 support

Fix the missing gstreamer1 build dependencies, which could possibly
prevent the configuration of qt5multimedia from detecting the supported
gstreamer1 features.

Fix the missing gstreamer1 install rules, which resulted in the
following runtime error:

    defaultServiceProvider::requestService(): no service found for - "org.qt-project.qt.mediaplayer"

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

qt5: disable for static-only builds

Even though we have some specific code to support building Qt5 for
static-only configurations, it doesn't work. The first problem is that
our custom qmake.conf always passes -ldl, which makes a number of Qt5
config.tests fail at configure time. Once this problem is fixed by
removing -ldl from QMAKE_LIBS and adding it to QMAKE_LIBS_DYNLOAD
instead, the next problem is that the plugin infrastructure of Qt5
assumes that Linux has dynamic library support: the qlibrary_unix.cpp
file includes <dlfcn.h>, and the only condition for this file to not
be included is:

Until recently, building Qt5 statically was working because our C
library was not built static-only: it provided <dlfcn.h> and
libdl.so. But now that we have a really static only toolchain, Qt5 no
longer builds.

The easiest solution is to simply make Qt5 depend on dynamic library
support.

Fixes:

   http://autobuild.buildroot.net/results/538/538e0325adba9fabbe4ec8e550fbb6a7219f5e7a/

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

docs/website: fix manual symlink

It should point to the filesystem path on the server, not the HTTP virtual path.

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

perl-http-daemon: fix dependencies

regenerated by support/scripts/scancpan

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

perl-http-cookies: fix dependencies

regenerated by support/scripts/scancpan

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

perl-www-robotrules: fix dependencies

regenerated by support/scripts/scancpan

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

website: make a symlink for the manual at the root of the website

This will give us a shorter URL, that we can more easily refer to in the
documetation itself, in help texts, on IRC...

[Peter: Use buildroot.org everywhere]
Suggested-by: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

packages/linux-pam: disable when using musl

Fixes: http://autobuild.buildroot.net/results/e33/e337d69420ad00b2cc4017d639a31803926f2353/
linux-pam needs some surgery to build under musl, so for the time being
disable it until the issues are solved. Also disables dependent package
python-pam, and dependent suboptions under openvmtools, rsh-redone and
util-linux.

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

tn5250: fix static build with openssl

OpenSSL needs libz. Use pkg-config to find dependencies.

Fixes:
http://autobuild.buildroot.net/results/95f/95f6df0d01e4944ecd84e830e2bbd8389e1d3e3d/
http://autobuild.buildroot.net/results/432/432b1fdec1e6282d51ba1d859c0f4ab9ed2a5efc/
http://autobuild.buildroot.net/results/d6a/d6a7125c21b4f92b69bb9e30b05c4727eedb63ba/

and more.

[Thomas: use += instead of = when assigning <pkg>_CONF_ENV to avoid
future mistakes.]

Reviewed-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Tested-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

tinyalsa: fix build with gcc 5.x

This commit adds a patch to tinyalsa to fix the build with gcc 5.x. It
fixes autobuilder issues like:

  http://autobuild.buildroot.net/results/d19/d19d1779da473d7adee2a18fea990fa19c7a7697/

The fix has been submitted upstream.

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

zyre: bump version to fix static linking problem

The last tagged release of zyre, v1.0.0, was made in May 2014. Since
then, they have switched to pkg-config to detect zmq and czmq, which
fixes static linking problems. However, they made a number of changes
to configure.ac, which make it difficult to backport just the relevant
changes.

Since we already had another backported fix, let's simply bump the
version of zyre to the latest available commit, which builds fine with
no change for shared and static scenarios, thanks to the use of
pkg-config.

An issue was opened upstream to ask them to tag a new release:
https://github.com/zeromq/zyre/issues/324.

Fixes:

  http://autobuild.buildroot.net/results/0ab/0ab4c6a4bb4942d51e7712073d4731d81ecb5251/

Thanks to Vincente Olivert Riera for the initial investigation.

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

openssh: security bump to version 7.1p1

Fixes:
CVE-2015-6563 - Fixed a privilege separation weakness related to PAM
support.
CVE-2015-6564 - Fixed a use-after-free bug related to PAM support that
was reachable by attackers who could compromise the pre-authentication
process for remote code exectuion.
CVE-2015-6565 - incorrectly set TTYs to be world-writable.

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

package/ltrace: do not enable libunwind support for MIPS

Currently libunwind support in ltrace is broken for MIPS. I'm working
with upstream to fix this issue, but it's not yet ready, so let's
disable it by now.

Fixes:

  http://autobuild.buildroot.net/results/79b/79b51941ed57b0564c68112489b03cac39a04e9a/

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

package/canfestival: don't build in parallel

The Makefiles for canfestival are not correctly written, which leads to
multiple warnings such as:

    make[4]: warning: jobserver unavailable: using -j1. Add `+' to parent make rule.

Since canfestival is relatively small, it builds in less than 6s here
when not in parallell, while a parallel build takes 5s.

Just disable parallel build to avoid future surprises.

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

midori: depend on x11 backend for gtk3

Mirror commit ba25711e for midori as well, otherwise gtk3 with non-X11
backends can trigger build failures. Fixes:
http://autobuild.buildroot.net/results/80ffdf54fd675fbff5a645784cd041cdf862154c/
http://autobuild.buildroot.net/results/1c049845ade60de223296c77c8a70869c4db64f0/

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

erlang-p1-tls: add missing dependency to openssl

Signed-off-by: Johan Oudinet <johan.oudinet@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/libmbim: Now needs libgudev under systemd

libmbim uses code (originally from udev) that has since been split
from the main systemd codebase into libgudev.

Fixes: http://autobuild.buildroot.org/results/638/638dbf05b785a276a33983b0237b7cad54777b85/
Tweak the package files for libmbim to require libgudev when building
with systemd.

Signed-off-by: Nathaniel Roach <nroach44@gmail.com>
Tested-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/network-manager: now needs libgudev when systemd is used

NetworkManager uses code (originally from udev) that has since been
split from the main systemd codebase into libgudev.

Tweak the package files for NetworkManager to require libgudev when
building with systemd.

Signed-off-by: Nathaniel Roach <nroach44@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

libgudev: New package

As libgudev recently was split from the main systemd/udev source, this
library is now required to build certain packages.

This library is only relevant to systemd, as the code it contains is
still contained in eudev.

[Thomas:
 - don't show the dependency comment when systemd is not available,
   since libgudev is anyway useless when you're not using systemd.
 - fix the license, it's LGPLv2.1+ and not GPLv2+
 - remove useless empty lines in the .mk file.]

Signed-off-by: Nathaniel Roach <nroach44@gmail.com>
Tested-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

core/pkg-kconfig: fix configurators

Currently, the configurators are using $($(2)_MAKE_ENV), often derived
from $(TARGE_MAKE_ENV), as the environment to be set when calling the
various configurators.

This means that our host tools are used first, most notably pkg-config
(from host-pkgconf).

However, this is inherently flawed. Our pkg-config, when set for the
host, only searches .pc files in $(HOST_DIR) and never ever uses the
ones from the host. For example, since we do not build a host-qt, our
pkg-config would not find the host's QtCore.pc et al.

Consequently, on some systems (but not on others?) most of the
configurators fail to build, especially the latest kernel versions, as
they have been starting to use pkg-config two years ago.

Fix that by filtering-out sensible values out of the environment, but
only when calling the configurators.

[Thomas: rewrap comment to appropriate length.]

Reported-by: Mauro Condarelli <mc5686@mclink.it>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Mauro Condarelli <mc5686@mclink.it>
Tested-by: Mauro Condarelli <mc5686@mclink.it>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/x11r7/xserver_xorg-server: Add libdrm dependency when dri or dri2 is enabled

Fixes compile error using this defconfig
BR2_TOOLCHAIN_BUILDROOT_WCHAR=y
BR2_TOOLCHAIN_BUILDROOT_CXX=y
BR2_PACKAGE_XORG7=y
BR2_PACKAGE_XSERVER_XORG_SERVER=y
BR2_PACKAGE_XPROTO_DRI2PROTO=y

drmVersionPtr is referenced not only in hw/xfree86/dri2/dri2.c
but also in hw/xfree86/dri/dri.c.

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>

package/x11r7/xserver_xorg-server: Fix libdrm dependency handling

Fixes
http://autobuild.buildroot.net/results/2e1/2e12f1a91812f027d89fcceeb41d6c88ef7b9d32/

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>

package/x11r7/xserver_xorg-server: Fix compilation of modesetting driver

Kernel modesettings support also depends on dri2, see
http://cgit.freedesktop.org/xorg/xserver/tree/hw/xfree86/drivers/modesetting/Makefile.am#n46

Fixes
http://autobuild.buildroot.net/results/f26/f26018d5ad62bc7b61623dd173dc1e92e2c958a4/

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>

package/libwebsock: fix incorrect inline function

Fixes:
    http://autobuild.buildroot.org/results/c02/c0243afc26202cbea98a1a75ddb9e37d8e2e08de/
    http://autobuild.buildroot.org/results/8a0/8a0f34da29fdfb2807aaa7280759e324a3874851/
    http://autobuild.buildroot.org/results/3da/3dac7b3504349e06c05fa0170a033e7828a9ab55/

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

package/vlc: update opencv support

vlc supports both opencv-2.4 and opencv-3, so adjust the vlc package to
reflect this.

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

Revert "package/gstreamer1/gst1-plugins-bad: disable opencv plugin with opencv-3"

This (partially) reverts commit 5e238a87eacf2a0c62736c4123961651afb1ba87.

The dependency is changed from a 'select' to a 'depends on' to avoid a
circular dependency caused by the introduction of OpenCV-3. This means
we can also drop the threads and C++ dependencies, since they are now
inherited via the depends on OpenCV.

Cc: Jonathan Ben Avraham <yba@tkos.co.il>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
[yann.morin.1998@free.fr: fix dependencies]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package/opencv: re-introduce opencv for opencv-2.4

As Jonathan noticed in [1], users' applications may depend on opencv-2.4
APIs removed in opencv-3.0.

So, re-introduce opencv package as it was right before the bump to
opencv-3.0 (i.e.: commit bf00b5a9ea1ab4189fe5c7dea05f40d0bbbf2082).

We do not support both OpenCV-2.4 and OpenCV-3 at the same time, so make
OpenCV-3 depend on !OpenCV-2.4.

[1] http://lists.busybox.net/pipermail/buildroot/2015-August/135270.html

Cc: Jonathan Ben Avraham <yba@tkos.co.il>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
[yann.morin.1998@free.fr:
  - remove legacy symbols, now
  - make opencv3 depends on !opencv, not the other way around
  - slitghly reword the commit log (opencv/opencv3 dependency)
]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

package: rename opencv -> opencv3

Since there is a couple of API breaks between OpenCV 2.4 and 3.0, two
distinct packages mutually exclusive will be integrated in the package
tree.

So, this change prepares the re-introduction of the OpenCV-2.4 package
by renaming the current opencv package (which provides OpenCV-3.0) to
opencv3.

Reverse dependencies (vlc) is fixed to use the new symbols.

Cc: Jonathan Ben Avraham <yba@tkos.co.il>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
[yann.morin.1998@free.fr:
  - fix missed usage in vlc.mk
  - don't remove legacy OpenCV symbols
  - fix 'endif' comment
  - slightly reword commit log (reverse deps)
]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

dhcpdump: fix build with musl

muls exposes udphdr Linux member names only when _GNU_SOURCE is defined.

Fixes:
http://autobuild.buildroot.net/results/dc2/dc2902b758faa7bcfed44cca5fa73aeffdf284ae/
http://autobuild.buildroot.net/results/432/43214b5f7d9e5881fbeba7203ee7fcfc6386a492/
http://autobuild.buildroot.net/results/291/29100641731904b50d73af7cf2db08847fe1567a/

and more.

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

Add link to 2015.08-rc2 announcement e-mail

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

Update for 2015.08-rc2

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

apitrace: bump to version 7.0

Our current apitrace version can't detect the host-python version
correctly, so if both host-python and host-python3 where installed, it
will take the last one and it will fail with an "invalid syntax" error.
The latest apitrace version has this problem solved and it detects the
host-python version correctly.

Fixes:

  http://autobuild.buildroot.net/results/22a/22a73b4ba0adcc874ecc153917ae6edcfd4d37af/

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

tftpd: fix init script

The daemon binary is tftpd, not in.tftpd. While we are at it, drop the
unneeded /usr/local from the PATH.

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