quota: fix static linking

The aim of this patch is to fix bug #7574, i.e fix the static linking
of the quota package. It does so by introducing a patch to the quota
build system that generalizes the use of $(LIBS), and then changes
quota.mk to use LIBS instead of LDFLAGS to link against intl and tirpc
when needed.

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

quota: remove dependency on util-linux

The dependency on util-linux is only present in Config.in, and not in
quota.mk, and quota indeed builds properly without util-linux. It
could be a runtime dependency, but there is no indication that it is
the case, and I don't see why quota would run-time depend on
util-linux utilities.

Looking back at when the quota package was introduced, in one of the
preliminary patch, he following explanation was given by the original
author:

  [Update: I added check for util-linux mount because
   it support usrquota and grpquota mount options.]

But I still don't see why usrquota and grpquota mount options would be
the source of a dependency of the quota utilities on util-linux. Here
is what the util-linux mount man page says about those two mount
options:

   grpquota|noquota|quota|usrquota

         These options are accepted but ignored.  (However, quota
         utilities may react to such strings in /etc/fstab.)

So indeed, the quota tools will look at /proc/mounts and see if those
options are used for certain mount points, but that doesn't create a
dependency of quota on util-linux.

Therefore, this commit gets rid of the dependency of quota on
util-linux. It allows to re-enable quota on Microblaze, since this
dependency was inherited from util-linux.

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

quota: rename patch to the new convention

As a preparation to the introduction of an additional patch to quota,
let's rename the existing patch to the new naming convention.

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

libv4l: fix utilities build with older glibc versions

Fixes:
http://autobuild.buildroot.net/results/9af/9af37fbf344b63b60e59ccac21e010cdf0ea219e/
http://autobuild.buildroot.net/results/bbd/bbd34d5378354e9db7822eaae8d3c4a75a67ac23/
http://autobuild.buildroot.net/results/ebb/ebbaa70e1a277162ba53c7bde8335ee998486703/
http://autobuild.buildroot.net/results/2c1/2c11f4dbb7cfadcfc6fbc267679b2b4eab867221/

And many more.

The utilities use clock_gettime(), which was provided in librt in glibc <
2.17, so ensure we link against -lrt.

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

Config.in.legacy: fix typo

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

Config.in.legacy: add legacy handling for the recently removed BR2_x86_generic option

Buildroot automatically falls back to a sensible CPU variant, but inform the
user of the change anyway so they are aware of it.

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

mongoose: bump to version 5.5

- examples/server was renamed examples/web_server
 - patch was submitted and included upstream so we can drop it

Signed-off-by: Davide Viti <zinosat@tiscali.it>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/dovecot: Adjust LIBDOVECOT in dovecot-config to STAGING_DIR

Fixes
http://autobuild.buildroot.net/results/3e6/3e6c258d7636fedbb87ba62069094291666e6a85/
http://autobuild.buildroot.net/results/410/410b68afece06ddb03a00245cfdc3de4d9a4e5f1/
http://autobuild.buildroot.net/results/bd3/bd37ee92eeb00adb1558dbb61315465a0cdfe635/
http://autobuild.buildroot.net/results/fd1/fd18c3678eded431476b4e61a10c48e160ffd51a//
and many others

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

package/dovecot: Fix build error in SQLite module

Patch occured with "make dovecot" using this defconfig:
http://autobuild.buildroot.net/results/bd3/bd37ee92eeb00adb1558dbb61315465a0cdfe635/

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

collectd: drop memcachec from disable list

As it is explicitly handled below. No functional change, but cleaner this way.

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

package/libupnpp: bump to version 0.8.6

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

package/upmpdcli: bump to version 0.8.6

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

makedevs: Rework README

Make all the example as a space separated list.
The definition of the different type was modified to look like the same
section on the manual.

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

package/*/*.mk: Fix indent

Fix indent for LIBFOO_USERS and LIBFOO_PERMISSIONS  as per the manual example.

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

lttng-tools: bump version

The previous releases was removed from their servers has they did a
releases from a wrong tag, the resulting binary was wrong.
Thanks to "Yann E. Morin" for spotting that.

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

lttng-babeltrace: Bump version

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

opencv: superres - Fix return value VideoFrameSource_GPU

superres module fails to compile with the following error messages:

[100%] Building CXX object
modules/superres/CMakeFiles/opencv_superres.dir/src/super_resolution.cpp.o
/opencv-2.4.10/modules/superres/src/frame_source.cpp: In function
'cv::Ptr<cv::superres::FrameSource>
cv::superres::createFrameSource_Video_GPU(const string&)':
/opencv-2.4.10/modules/superres/src/frame_source.cpp:263:16: error:
expected type-specifier before 'VideoFrameSource'
/opencv-2.4.10/modules/superres/src/frame_source.cpp:263:16: error:
could not convert '(int*)operator new(4ul)' from 'int*' to
'cv::Ptr<cv::superres::FrameSource>'
/opencv-2.4.10/modules/superres/src/frame_source.cpp:263:16: error:
expected ';' before 'VideoFrameSource'
/opencv-2.4.10/modules/superres/src/frame_source.cpp:263:41: error:
'VideoFrameSource' was not declared in this scope
/opencv-2.4.10/modules/superres/src/frame_source.cpp:264:1: error:
control reaches end of non-void function [-Werror=return-type]
cc1plus: some warnings being treated as errors
make[3]: ***
[modules/superres/CMakeFiles/opencv_superres.dir/src/frame_source.cpp.o]
Error 1
make[3]: *** Waiting for unfinished jobs....

This is caused because the return value of the
createFrameSource_Video_GPU function should be a VideoFrameSource_GPU
object.

Backporting an upstream patch to fix this problem in Buildroot:
  https://github.com/Itseez/opencv/commit/2e393ab83362743ba1825ad4b31d4a2925c606b4

Fixes:
  http://autobuild.buildroot.net/results/b09/b0996267197a9016d29d6070804ebc0cb7853548/

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

opencv: rename patches to follow the new name structure

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

package/clamav: add hash

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

python-netifaces: bump to 0.10.4

Change download location and remove obsolete patch.

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

libiscsi: only build the test tool and ld-iscsi if we have shared libs

Backporting an upstream patch to fix a failure when doing a static
build:

/br/output/host/usr/bin/mipsel-ctng-linux-uclibc-gcc -shared -o
ld_iscsi.so ld_iscsi.o -ldl
/br/output/host/opt/ext-toolchain/bin/../lib/gcc/mipsel-ctng-linux-uclibc/4.8.2/../../../../mipsel-ctng-linux-uclibc/bin/ld:
ld_iscsi.o: relocation R_MIPS_HI16 against `__gnu_local_gp' can not be
used when making a shared object; recompile with -fPIC
ld_iscsi.o: could not read symbols: Bad value
collect2: error: ld returned 1 exit status

Upstream commit:
  https://github.com/sahlberg/libiscsi/commit/3d6c2be3428d88682cecfa2a4ad172aaacffb318

Fixes:
  http://autobuild.buildroot.net/results/7a9/7a9caf1f4080c2c4b04ee3b13c1240f475a22ea7/

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

libiscsi: rename patches to follow the new name structure

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

tcl: add hash file

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

tcl: update package disabling

Update package disabling (removing pkgs subdirectories) after
latest tcl version update:

$ ls tcl8.6.1/pkgs/
 itcl4.0.0
 sqlite3.8.0
 tdbc1.0.0
 tdbcmysql1.0.0
 tdbcodbc1.0.0
 tdbcpostgres1.0.0
 tdbcsqlite3-1.0.0
 thread2.7.0

$ ls tcl8.6.2/pkgs/
 itcl4.0.1
 sqlite3.8.6
 tdbc1.0.1
 tdbcmysql1.0.1
 tdbcodbc1.0.1
 tdbcpostgres1.0.1
 tdbcsqlite3-1.0.1
 thread2.7.1

Fixes:
 http://autobuild.buildroot.net/results/07ae8e42acf62fef99de70c8099ec5a0ca89a817/

Could not reproduce the build failure, but disabling pkgs/sqlite3.8.6 should
prevent all build failures in this directory...

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

package/mesa3d: bump to 10.3.3

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

package/cramfs: Needs zlib

cramfs.mk contains

CRAMFS_DEPENDENCIES = zlib

but this dependency was missing in Config.in.

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

vim: install vi symlink

Install a 'vi' symlink to win over busybox vi (more features) and in
case busybox isn't around, for people expecting plain simple 'vi' to
call the editor.

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

arptables: re-enable for static

Add a small patch to re-enable arptables for static builds.
The dlfcn.h is a stray include for a past attempt at loadable plugins
but the code is disabled so there's no need for it.

Also add hash file.

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

package/qemu: fix specifying custom target sub-set

The logic was wrong.

Even though it was working for previous versions of QEMU, it changed in
later versions, and thus now breaks on the version we currently package.

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

arptables: needs dynamic linking support for dlopen()

Fixes:
http://autobuild.buildroot.net/results/b96/b9642befef82e479cc50839e07114cb87314d156/
http://autobuild.buildroot.net/results/57f/57ffebbe925391badf81f6867fdbf16975b4af6f/
http://autobuild.buildroot.net/results/5f1/5f1966f1a7c8800846a8405b7e74f4893f3fcd14/

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

arch/Config.in.x86: drop BR2_x86_generic

The fuzzy generic x86 variant doesn't make much sense in the context of
Buildroot, and the recent change to use -march instead of -mtune broke it.

From the GCC manual:

https://gcc.gnu.org/onlinedocs/gcc-4.9.2/gcc/i386-and-x86-64-Options.html#i386-and-x86-64-Options:

-mtune=cpu-type
    Tune to cpu-type everything applicable about the generated code,
    except for the ABI and the set of available instructions. While
    picking a specific cpu-type schedules things appropriately for that
    particular chip, the compiler does not generate any code that cannot
    run on the default machine type unless you use a -march=cpu-type
    option. For example, if GCC is configured for i686-pc-linux-gnu then
    -mtune=pentium4 generates code that is tuned for Pentium 4 but still
    runs on i686 machines.

    The choices for cpu-type are the same as for -march. In addition,
    -mtune supports 2 extra choices for cpu-type:

    ‘generic’
        Produce code optimized for the most common IA32/AMD64/EM64T
        processors. If you know the CPU on which your code will run,
        then you should use the corresponding -mtune or -march option
        instead of -mtune=generic. But, if you do not know exactly what
        CPU users of your application will have, then you should use
        this option.

        As new processors are deployed in the marketplace, the behavior
        of this option will change. Therefore, if you upgrade to a newer
        version of GCC, code generation controlled by this option will
        change to reflect the processors that are most common at the
        time that version of GCC is released.

        There is no -march=generic option because -march indicates the
        instruction set the compiler can use, and there is no generic
        instruction set applicable to all processors. In contrast,
        -mtune indicates the processor (or, in this case, collection of
        processors) for which the code is optimized.

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

bash: switch to release tarball

Yay!
The release tarball allows us to drop sizable in-tree patches, so switch
to it.

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

systemd: Properly define the __NR_memfd_create macro for MIPS

This is an upstream patch which has been applied to the master branch
and I have adapted it to make it apply to the v216 version.

v217 is not fixed. This patch can be removed when we bump to v218.

Upstream commit:
  http://cgit.freedesktop.org/systemd/systemd/commit/?id=e6c019026b8cfd27a997e6e6ed1349f8f289b7e2

Fixes:
  http://autobuild.buildroot.net/results/596/59617a822fb2c8961881dd78a0dc8e41bdb14ede/

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

systemd: rename patches to follow the new name structure

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

grep: fix build failure for musl

musl doesn't implement re_match and friends and grep fails to
detect this properly, so force internal regex routines for musl.

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

coreutils: fix build failure for musl

musl doesn't implement re_match and friends and coreutils fails to
detect this properly, so force internal regex routines for musl.

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

ffmpeg: typo fix

Add missing bracket.

Signed-off-by: Karoly Kasza <kaszak@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

gdk-pixbuf: host variant needs pkgconfig and libglib2 just like target variant

Fixes:

http://autobuild.buildroot.net/results/576/576d110cb02a8bc655e227bd949bb58ae39c2e38/
http://autobuild.buildroot.net/results/c48/c487a1fff7c74218a31024074aa2fd6255b6d09f/
http://autobuild.buildroot.net/results/b28/b282b7e7a9e3e5a8334d593be9092746509d4702/
http://autobuild.buildroot.net/results/dee/dee288268d81cc19fe685b32d68036b596af260e/

And many more.

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

elfutils: disable for static builds

Even when doing static builds, a shared library is built. This causes a
build failure under some circumstances, for instance when building for
MIPS + uClibc + static.

After asking upstream if it would be possible to add a configure option
to not build the shared library, the answer was that doing a static
build is not a good idea. Here is a small snippet of the conversation:

"Note that fully static builds are problematic.  elfutils uses dlopen to
open the EBL backends (the CPU-specific support snippets), so even if
you link statically, the final binaries are still considerably dynamic."

Related:
  https://lists.fedorahosted.org/pipermail/elfutils-devel/2014-November/004223.html

Fixes:
  http://autobuild.buildroot.net/results/691/6913f5af6519463fbed39ef37b6a40ecf6a67b54/

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

btrfs-progs: bump to version 3.17.1

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: Peter Korsgaard <peter@korsgaard.com>

iproute2: bump to version 3.17.0

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: Peter Korsgaard <peter@korsgaard.com>

pwgen: bump to version 2.07

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: Peter Korsgaard <peter@korsgaard.com>

arch: remove the BR2_GCC_TARGET_TUNE option

The BR2_GCC_TARGET_TUNE option is now unused, so we can get rid of it.

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

toolchain-external: do not use BR2_GCC_TARGET_TUNE anymore

Since the BR2_GCC_TARGET_TUNE value is always empty now, there is no
longer a point in using it in the external toolchain logic.

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

gcc: do not use BR2_GCC_TARGET_TUNE anymore

Since the BR2_GCC_TARGET_TUNE value is always empty now, there is no
longer a point in using it in the gcc package.

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

arch/m68k: get rid of BR2_GCC_TARGET_TUNE

On m68k, we are passing the exact same values of BR2_GCC_TARGET_ARCH
and BR2_GCC_TARGET_TUNE, which is redundant. Therefore, this commit
removes the usage of BR2_GCC_TARGET_TUNE on m68k.

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

arch/powerpc: get rid of BR2_GCC_TARGET_TUNE

According to the gcc documentation for PowerPC options:

'-mtune=CPU_TYPE'
     Set the instruction scheduling parameters for machine type
     CPU_TYPE, but do not set the architecture type, register usage, or
     choice of mnemonics, as '-mcpu=CPU_TYPE' would.  The same values
     for CPU_TYPE are used for '-mtune' as for '-mcpu'.  If both are
     specified, the code generated will use the architecture, registers,
     and mnemonics set by '-mcpu', but the scheduling parameters set by
     '-mtune'.

In the case of Buildroot where we only target a specific system, using
-mtune therefore doesn't make much sense, and using -mcpu would be
more appropriate. As a consequence, this patch makes PowerPC use
BR2_GCC_TARGET_CPU instead of BR2_GCC_TARGET_TUNE.

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

arch/sparc: get rid of BR2_GCC_TARGET_TUNE

On SPARC, BR2_GCC_TARGET_TUNE was only used for one specific case, the
BR2_sparc_v8. There is actually no reason to not use
BR2_GCC_TARGET_CPU instead for this, as all values supported for
-mtune are also supported for -mcpu. Therefore, the only
BR2_GCC_TARGET_TUNE case is moved as a BR2_GCC_TARGET_CPU case.

[Peter: fixup merge conflict after 'sparc: Add leon3 cpu type and remove sparc{s,h}fleon{,v8}]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

arch/x86: get rid of BR2_GCC_TARGET_TUNE

According to the gcc documentation, -march implies -mtune on x86, so
defining both BR2_GCC_TARGET_ARCH and BR2_GCC_TARGET_TUNE is
redundant. Therefore, this commit removes the definition of
BR2_GCC_TARGET_TUNE on x86.

However, while doing so, it adds one new case for BR2_GCC_TARGET_ARCH:
using -march=generic when BR2_x86_generic is selected.

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

ffmpeg: use arch and cpu instead of tune

ffmpeg currently uses BR2_GCC_TARGET_TUNE as the --cpu option
value. However, there are multiple architectures for which
BR2_GCC_TARGET_TUNE is not used. After inspecting the ffmpeg configure
script, we concluded that using the BR2_GCC_TARGET_CPU value if
available, or the BR2_GCC_TARGET_ARCH value as a fallback was the
appropriate behavior.

This allows to remove the reference to BR2_GCC_TARGET_TUNE, which is
one step towards the removal of this option.

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

arch/arm: remove BR2_GCC_TARGET_ARCH definitions on ARM

On ARM, we were defining both the CPU type and the architecture
variant. However, depending on the version of gcc, a given combination
of (CPU, architecture) may not be the same. Since the architecture
variant is implied by the CPU type, given the former is not necessary,
and we can simply specify the latter.

>From the gcc documentation:

  This specifies the name of the target ARM processor. GCC uses this
  name to derive the name of the target ARM architecture (as if
  specified by -march) and the ARM processor type for which to tune
  for performance (as if specified by -mtune). Where this option is
  used in conjunction with -march or -mtune, those options take
  precedence over the appropriate part of this option.

Note that we verified that for all BR2_GCC_TARGET_ARCH value that
existed, a proper BR2_GCC_TARGET_CPU value is defined.

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

arch/arm: do not distinguish revisions of ARM1136JF-S

In commit 88cf3bb91792c9c04586e14f293d89a6e0c13e1d
("arch/Config.in.arm: Use armv6k for arm1136jf-s rev1"), Benoît
Thébaudeau added separate options for the revision 0 and revision 1 of
the ARM1136JF-S processor, so that different -march values could be
used (armv6j for revision 0, armv6k for revision 1).

However, this is preventing the removal of the BR2_GCC_TARGET_ARCH
option, which we need to do to give only the CPU type to gcc, and let
it decide the architecture variant that matches. This is because this
story of revision 0 vs. revision 1 is the only case where -mcpu
doesn't fully define the CPU.

Moreover, a quick test with gcc shows that -march=armv6j
-mcpu=arm1136jf-s is accepted, while -march=armv6k -mcpu=arm1136jf-s
makes gcc complain: " warning: switch -mcpu=arm1136jf-s conflicts with
-march=armv6k switch".

In addition, gcc 5 will apparently no longer allow to pass all of
--with-arch, --with-cpu and --with-tune, so we will anyway have to
rely only on one of them.

As a consequence, this commit basically reverts
88cf3bb91792c9c04586e14f293d89a6e0c13e1d and provides only one option
for ARM1136JF-S. If the two revisions are really different, then they
should be supported in upstream gcc with different -mcpu values.

Note that the removal of the two options should not break existing
full .config, since the hidden option BR2_arm1136jf_s becomes again a
visible option to select the CPU.

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

dhcpcd: do not pass --target option

dhcpcd is using $(BR2_GCC_TARGET_ARCH) as the value of its --target
option. However, depending on the architecture, this value might be
empty.

After inspecting the dhcpcd configure script, it turns out that in
fact the --target option has the following semantic:

  --target=TARGET   configure for building compilers for TARGET [HOST]

So it is very much like the --target option of regular,
autoconf-based, configure scripts. Since dhcpcd is not a compiler,
passing --target is pointless. Moreover, as long as --os=linux is
passed, passing --host or --target is completely ignored by the
configure script.

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

toolchain-external: use BR2_ARM_CPU_ARM* options

Refactor the toolchain-external Config.in file to use the
BR2_ARM_CPU_ARM*. All of the changes are purely mechanical, except for
the Arago ARMv5 toolchain: it had a 'depends on BR2_GCC_TARGET_ARCH !=
"armv5t"', but armv5t was not a possible value for
BR2_GCC_TARGET_ARCH. Since the toolchain is ARMv5TE, the only ARM
architectures we need to exclude are ARMv4 and ARMv4T.

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

libnspr: use the BR2_ARM_CPU_ARMV7A option

The libnspr --enable-thumb2 option was only enabled on Cortex-A8 and
Cortex-A9. This commit switches to using BR2_ARM_CPU_ARMV7A instead,
which will allow using Thumb2 for this package even with the other
Cortex-A cores.

[Peter: use BR2_ARM_CPU_HAS_THUMB2 instead]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

x264: use the BR2_ARM_CPU_ARMV7A option

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

valgrind: use the BR2_ARM_CPU_ARM* options

Instead of using the exhaustive list of Cortex-A ARM cores, use
BR2_ARM_CPU_ARMV7A instead. We also fix a wrong comment that said the
Valgrind package could only be selected for Cortex-A8 and Cortex-A9,
which was true a long time ago, but was no longer true since it was
also enabled for all the other Cortex-A platforms.

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

webkit: use BR2_ARM_CPU_ARM* options

The webkit package is not available on ARM < v5, so this commit
switches to using the newly introduced BR2_ARM_CPU_ARM* options
instead of per ARM core options.

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

vo-aacenc: use the BR2_ARM_CPU_ARM* options

The vo-aacenc package has an option to enable ARMv5 optimizations. It
was enabled based on per ARM core options, but now that we have per
ARM architecture options, this commit switches to using them.

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

speex: use BR2_ARM_CPU_ARM* options

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

qt5: use the BR2_ARM_CPU_ARM* options

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

nodejs: use BR2_ARM_CPU_ARM* options

Adjust the nodejs Config.in dependencies to use the per ARM
architecture BR2_ARM_CPU_ARM* options instead of the per ARM core
options.

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

nettle: use the BR2_ARM_CPU_ARM* options

Adjust the nettle package to use the BR2_ARM_CPU_ARM* option to
disable the usage of assembly code on ARMv4 and ARMv5.

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

libplayer: use BR2_ARM_CPU_ARM* option

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

mplayer: also hide toolchain comment on ARMv4

Like we do for the other arch dependencies.

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

mplayer: allow usage of ARMv6 optimizations on ARMv7 platforms

Similar to what we do for ffmpeg/gst-ffmpeg/gst1-libav.

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

mplayer: use BR2_ARM_CPU_ARM* options

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

gst1-libav: allow usage of ARMv6 optimizations on ARMv7 platforms

Contrary to the ffmpeg package, the gst1-libav package was only
allowing ARMv6 optimizations on ARMv6 platforms. This commit also
allows ARMv6 optimizations on ARMv7 platforms, like the ffmpeg package
does.

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

gst1-libav: use the new BR2_ARM_CPU_ARM* options

Instead of using the Config.in options for each individual ARM core,
use the newly introduced BR2_ARM_CPU_ARM* options, which exist per ARM
architecture.

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

webkit: Add --enable-dependency-tracking to the configure options

--disable-dependency-tracking was added globally to
package/pkg-autotools.mk by this commit:

http://git.buildroot.net/buildroot/commit/?id=3e37b0fc6cbbe2245e39fedb247a381fb4e0d992

This causes a failure in webkit because some source files are generated
after the Makefile is created. To fix this, we add
--enable-dependency-tracking to the configure options.

Fixes:
  http://autobuild.buildroot.net/results/5beea967b02df202bca1d42afd6fa8c2c13383d5/

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

linux-headers: bump 3.{12, 2}.x series

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

barebox: bump to version 2014.11.0

Signed-off-by: Fabio Porcedda <fabio.porcedda@gmail.com>
Reviewed-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Tested-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

stunnel: bump to version 5.07

Also enable optional systemd support.

[Peter: fixup announcement url in .hash as pointed out by Baruch]
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: Peter Korsgaard <peter@korsgaard.com>

libcurl: security bump to version 7.39.0

Fixes:
CVE-2014-3707 - libcurl's function curl_easy_duphandle() has a bug that
can lead to libcurl eventually sending off sensitive data that was not
intended for sending.

Removed patch that was upstream and now in the release.

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: Peter Korsgaard <peter@korsgaard.com>

subversion: ensure --disable-debug doesn't mangle CFLAGS

Fixes:
http://autobuild.buildroot.net/results/fc6/fc69a19c66462585449f7c4dad174d45a84e4947/
http://autobuild.buildroot.net/results/e04/e0471f2a9087d547840a7b18863289963e357b57/
http://autobuild.buildroot.net/results/bd8/bd8cdf976937c7b9029658871929f4be464b7a47/
http://autobuild.buildroot.net/results/652/652c3afe844e912061fbc5991e6fecad98ff6e6f/

And many more.

When --disable-debug is passed to configure, as is automatically done by the
autotools infrastructure since 822a757456e (infra: Move
--enable/--disable-debug to package/Makefile.in), the configure script will
try to strip debugging (-g) options from the C/CXXFLAGS. The logic to do so
is unfortunately buggy, so it ends up mangling options like
-mfloat-gprs=double that we use on certain PowerPC variants, breaking the
build.

Fix it by adjusting the sed regexp to be more selective in what it strips.
The package unfortunately doesn't cleanly autoreconf, so configure is
patched instead of configure.ac.

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

adwaita-icon-theme: depends on host-libgtk3

The adwaita icon theme can only be selected with libgtk3 and depends on
host-libgtk3 to build.

Fixes http://autobuild.buildroot.org/results/f8d/f8d0e2198fa54f9141512e2db6dc46e3cee1cd44/

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

mpd: bump to version 0.19.2

Patch now upstream.

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

lttng-tools: update hash

Fixes:
http://autobuild.buildroot.net/results/4c0/4c0f8583986d72fbc879d21bef65246bc9b4a732/
http://autobuild.buildroot.net/results/db4/db4011b3444c71942b1a269dd16b18652fd22d97/
http://autobuild.buildroot.net/results/220/220ef41676096bafd2e38f4fbb892eb0a7d654d8/
http://autobuild.buildroot.net/results/545/545371f5a39c85aba0d15a41cdc46e1cf6c1ba9f/

And many more.

Upstream silently updated the tarball on 30/10, breaking the build. The only
difference between the old and new tarballs is that the version number is
fixed:

diff -urpN old/lttng-tools-2.5.1/include/version.h new/lttng-tools-2.5.1/include/version.h
--- old/lttng-tools-2.5.1/include/version.h     2014-10-20 21:47:51.000000000 +0200
+++ new/lttng-tools-2.5.1/include/version.h     2014-10-31 16:33:17.000000000 +0100
@@ -20,7 +20,7 @@
 #ifndef VERSION_H
 #define VERSION_H

-#define GIT_VERSION "heads/stable-2.5-0-g56eb774"
+#define GIT_VERSION "2.5.1"

The copy of the old tarball on sources.buildroot.net has also been removed.

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

qextserialport: allow building with Qt5

qextserialport builds fine with both Qt4 and Qt5, so this commit
enables it for Qt5.

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

qjson: allow building with Qt5

Since commit 18714011b231fc326c96782134fbea778ab6a94d ("qjson: bump
version"), qjson can be built either against Qt4 or Qt5. This commit
therefore allows this package to be selected when Qt5 is enabled.

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

qt5: expose a QT5_QMAKE variable for other Qt5 packages

Much like package/qt/qt.mk exposes a QT_QMAKE variable that other
qmake-based Qt4 libraries/applications can use in their .mk file, this
commit adds a QT5_QMAKE variable to allow the same for Qt5 packages.

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

package: prepare to support modules that work with Qt or Qt5

Currently, all the Qt external modules (qtuio, qjson, etc.) can only
be compiled against Qt4. However, some of them (qjson, qextserialport
at least) can compile with either Qt4 or Qt5.

Therefore, this commit adjusts the Config.in logic to make it possible
for certain modules to be selected with either Qt4 or Qt5. We use the
same approach as the one of for Python 2 vs. Python 3 modules:

 - in package/Config.in, the package Config.in files are included
   either if BR2_PACKAGE_QT or BR2_PACKAGE_QT5 are enabled.

 - each individual package is then responsible for having a dependency
   on either BR2_PACKAGE_QT, or BR2_PACKAGE_QT5 or both depending on
   what they support. In this commit, we add a dependency on
   BR2_PACKAGE_QT to all packages, since for now, they only support
   Qt4.

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

qt5: select qt5base

qt5base is (as the name hints) the basics of anything related to qt5, so
enabling qt5 without enabling qt5base doesn't make any sense as nothing will
get built.

Selecting it here also simplifies the upcoming series to add qt5 support to
external qt modules.

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

qemu: needs IPv6

Fixes http://autobuild.buildroot.net/results/f53/f53d4d47a6b65c40c1b122dfad61d9164e533bf0/

Actually only the linux-user code needs it, but with the _CUSTOM_TARGETS
option it is difficult to ensure it only for userland emulation, so just add
it to the main qemu option.

At the same time add the missing wchar/mmu dependencies to the comment.

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

qemu: fix build when python3 is enabled

Fixes:
http://autobuild.buildroot.net/results/04e/04ec455e352319a46dc06def1762c65ad28a12a1/
http://autobuild.buildroot.net/results/f19/f1932657912674af444602ac59211b6caf6841ca/
http://autobuild.buildroot.net/results/18a/18a43df80616280506a44ac3abc327ddf0e7b079/
http://autobuild.buildroot.net/results/44c/44c00b50248018c4e81f4fbb1041d35b78f89640/

And many others.

When python3 for the target is enabled, $(HOST_DIR)/usr/bin/python points to
host-python3, not host-python - Breaking the build. Ensure host-python is used.

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

avahi: fix libgtk3 build issue by patching Makefile.in instead of Makefile.am

Fixes:
http://autobuild.buildroot.net/results/f0b/f0b4f8c46eb1de6cf5ed081dacd248c66c591aa3/
http://autobuild.buildroot.net/results/dac/daca5d030b8ecc6d11901e2936fe2151d0930852/
http://autobuild.buildroot.net/results/05e/05e7fb9a888169ba7c4cb28abdb4febb5215354c/
http://autobuild.buildroot.net/results/743/743f30b1816b17a0a1397b7fd8b5a8571c4dde03/

And many others.

So we no longer need to autoreconf the package. Autoreconfing is non-trivial
as it uses m4 macros from glib, but we don't want to unconditionally pull in
glib.

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

orc: disable for uclinux

Fixes http://autobuild.buildroot.net/results/445/4456e49a6c33a80570b4f59ec9799a8d4a314af1/

The orc configure script doesn't recognize uclinux, so builds using
BR2_BINFMT_FLAT fails.  While this could be patched, orc doesn't have any
acceleration for blackfin and orc isn't a hard dependency of any other
package, so just disable it.

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

libwebsockets: needs host-openssl when built with openssl support

Fixes
http://autobuild.buildroot.net/results/d01/d0190a6e5c6b5aa78cfefc29e4e8a4e4b135450b/
http://autobuild.buildroot.net/results/9bb/9bb8defedb8024de83eb3a609ab584efb5d0d2b0/
http://autobuild.buildroot.net/results/e4a/e4a1b298e7e158d3614b89e61d40ef9a00e7ff67/
http://autobuild.buildroot.net/results/5e2/5e24a90058111f7d5618047410f809b45bc165f6/

And many others.

The build system wants to create a ssl key for testing using host-openssl,
so ensure it is available.

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

host-openssl: use correct prefix so libssl / libcrypto can be found

Fixes (root cause of):
http://autobuild.buildroot.net/results/d01/d0190a6e5c6b5aa78cfefc29e4e8a4e4b135450b/
http://autobuild.buildroot.net/results/9bb/9bb8defedb8024de83eb3a609ab584efb5d0d2b0/
http://autobuild.buildroot.net/results/e4a/e4a1b298e7e158d3614b89e61d40ef9a00e7ff67/
http://autobuild.buildroot.net/results/5e2/5e24a90058111f7d5618047410f809b45bc165f6/

And many others.

The host version of openssl should be configured to use the host directory
as its prefix and the INSTALL_PREFIX dance shouldn't be used, as otherwise
host-openssl will look in /usr for its libraries.

While we are at it, ensure our host CFLAGS are used similar to how we do for
the target build.  Also enable zlib support as we already list it as a
dependency.

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

dbus: Remove DBUS_USERS duplicate

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

targetcli-fb: bump version

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

package/ffmpeg: bump version to 2.4.3

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

package/dovecot-pigeonhole: bump version to 0.4.6

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

package: indentation cleanup

Signed-off-by: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

unzip: Use the "-q" option to silence unzipping of source files

Add and use the "UNZIP" variable instead of calling directly unzip
because the variable contains the "-q" option to silence "unzip" so it
doesn't show the list of files extracted just like when tar files are
being unpacked.

Signed-off-by: Fabio Porcedda <fabio.porcedda@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

arc: add support of ARC HS38 core

Synopsys has recently announced its new ARC HS38 core that is capable of
running Linux -
http://www.synopsys.com/dw/ipdir.php?ds=arc-hs38-processor

ARC HS38 is based on ARCv2 ISA and requires special settings of gcc and
libc.

Also in case of HS38 atomic extensions (LLOCK/SCOND instructions) are
built-in by default, so enabling atomic extensions in Buildroot as well.

This commit adds support of the core in buildroot.

[Peter: string type, so must be in quotes as noted by Yann]
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Anton Kolesov <anton.kolesov@synopsys.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

uclibc: add explicit setup of ARC-specific options

As with other architectures for ARC we need to have an ability to set
specific options in uClibc.

In particular this is required for selection of ARC ISA version.

[Peter: string type, so must be in quotes as noted by Yann]
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Anton Kolesov <akolesov@synopsys.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

arc: add explicit selection of CPU templates ARC 750D and ARC770D

This separation allows to specify unique options and features for each
CPU.

For example ARC 770D has LLOCK/SCOND instructions built-in by default.
Also this new scheme simplifies selection of proper configuration for
users - preconfigured options now match templates for ARC CPUs.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Anton Kolesov <akolesov@synopsys.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

live555: bump to version 2014.11.01

Also add a hash file. Upstream provides md5 only, so add a locally generated
sha256.

Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

board: add support for Minnowboard MAX

Very similar to the older Minnowboard, except that it works with mainline
Linux, uses 64bit firmware and a realtek NIC needing firmware.

The Linux configuration is based on the configuration fragment on elinux:
http://elinux.org/Minnowboard:MinnowMaxLinuxKernel

Many thanks to Circuitco for sponsoring a board.

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