libllcp: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

liblinear: bump to version 2.20

- Slightly update first patch so it can apply on this version
- Add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libldns: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libfreeimage: fix build with glibc

Add patch to rename powf64 into powf_64 as powf64 is already defined in
glibc

Fixes:
 - http://autobuild.buildroot.net/results/9b1e6d63db9eae46a92487713b0d72bd039deaba

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

core/pkg-infra: check for an SCM in the correct variable

The site method is stored in FOO_SITE_METHOD, not in FOO_SITE.

Fixes:
http://autobuild.buildroot.net/results/131/13196dd779bc9e3b172c74851546dd4c4752aa02/

[Peter: add autobuilder reference]
Reported-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libgsasl: add mandatory dependency on wchar

libgsasl needs wctomb, if not available it will define it to
rpl_wctomb. However, as there is no rpl_wctomb implementation, the .so
ends up with an undefined reference to wctomb.

This will cause linker errors for packages trying to use it such as
msmtp:
/home/peko/autobuild/instance-0/output/host/bin/../microblazeel-buildroot-linux-uclibc/sysroot/usr//lib/libgsasl.so:
undefined reference to `wctomb'

Fixes:
 - http://autobuild.buildroot.net/results/1bc7cb9a2dd1af746c78f3150528206b7256e40f

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

DEVELOPERS: Add myself to some packages

Add myself to lftp, libidn, libjpeg and mutt

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libjson: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libjpeg: bump to version 9c

Add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

python-subprocess32: needs kernel headers v3.10 on MIPS N64

MIPS N64 ABI has been missing getdents64 wire up until kernel version
3.10. Make python-subprocess32 depend on newer headers for MIPS N64.

Fixes:
http://autobuild.buildroot.net/results/af1/af1c0dbfcc27fa93992de57f986657d65163800e/
http://autobuild.buildroot.net/results/e5f/e5f9e96d02512359348886a9db3aa4cea9db2081/
http://autobuild.buildroot.net/results/8ba/8ba4c75d724b45c4fe10b4b6c90602b68a811fac/

Cc: Asaf Kahlon <asafka7@gmail.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux-headers: bump 4.{4, 9, 14, 16}.x series

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux: bump default to version 4.16.7

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python: add upstream GCC8 build fix

Fedora 28 switched to GCC8.

Signed-off-by: Stefan Becker <chemobejk@gmail.com>
[Thomas: fixup location of SoB in the patch.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

Config.in: add BR2_HOST_GCC_AT_LEAST_8

Fedora 28 switched to GCC 8.x.

Signed-off-by: Stefan Becker <chemobejk@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/clang: delete unnecessary files from target

Delete:
Binaries in: /usr/bin
     /usr/libexec

Directories: /usr/lib/clang
     /usr/share/clang
     /usr/share/opt-viewer
     /usr/share/scan-build
     /usr/share/scan-view

Manual:      /usr/share/man/man1/scan-build.1

Signed-off-by: Valentin Korenblit <valentin.korenblit@smile.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

cargo: bump version to 0.26.0

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

cargo-bin: bump version to 0.26.0

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

rust: bump version to 1.25.0

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

rust-bin: bump version to 1.25.0

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/llvm: delete LLVMHello.so from /usr/lib

LLVMHello.so contains the Hello World example of an
LLVM pass. It is not needed on the target.

Signed-off-by: Valentin Korenblit <valentin.korenblit@smile.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

DEVELOPERS: Merge my two entries

Merge my two entries in the DEVELOPERS file, because the mail address
<sven.haardiek@greenbone.net> is outdated.

Signed-off-by: Sven Haardiek <sven.haardiek@iotec-gmbh.de>
[Thomas: fix alphabetic ordering.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

DEVELOPERS: add myself for some packages

Add myself for libgee, libgtk2, libgtk3 and pcmanfm

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libilbc: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libcue: bump to version 2.2.1

- Remove patch (already in version)
- Include buildroot patch http://patchwork.ozlabs.org/patch/907206
  currently in new state

Fixes:
- http://autobuild.buildroot.net/results/02712e85db8e538045497e9242b476c040198f1d

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

s6-linux-init: update license hash

Copyright year update.

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

execline: update license hash

Copyright year update.

Fixes:
http://autobuild.buildroot.net/results/6f0dcacbba4abd48927ccae044c32e07a96a50a4/

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

libiscsi: add hash for license files

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

core/pkg-infra: don't enforce site-method for extra downloads

The site method only ever applies to the main download, while extra
downloads are always to be fetched with wget.

However, the site method is prepended to the URL from within the
DOWNLOAD macro (well, a variable evaluated in the DOWNLOAD macro),
which is called for each download of a package, thus effectively
prepending the site method to all downloads, even the extra ones (and
the patches).

We fix that by prepending the site method from within the
generic-package infra, so that it only applies to the main download.

For that, we move the main _SOURCE out of the foreach loop, so that
we can prepend the site-method to it, without impacting the other
downloads.

Reported-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Luca Ceresoli <luca@lucaceresoli.net>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Reviewed-by: Luca Ceresoli <luca@lucaceresoli.net>
Tested-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

core/pkg-infra: set no-check-hash only for main download

Currently, when the main download if from an SCM, we exclude all
downloads from being hash-checked, on the assumption that we don't
have hashes for downloads from an SCM.

However, the exclusion is computed on the DOWNLOAD macro, which is
called for each download of a package, thus effectively disabling
hash checks for extra downloads, even though those are only ever
download with wget.

What we really wanted to do, in fact, was to exclude just the main
download.

We fix that by appending the main source file to the global list of
excluded files, from within the generic-package infra itself.

Reported-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Luca Ceresoli <luca@lucaceresoli.net>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Reviewed-by: Luca Ceresoli <luca@lucaceresoli.net>
Tested-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/mesa3d: add missing elfutils dependency with R600 and RadeonSI driver

Commit "add dependency on elfutils for R600 with LLVM" [1] select
elfutils package at Kconfig level without adding it the make dependency.

Enabling r600 support lead to a build issue when building from
scratch with "make mesa3d".

Fixes:
checking for RADEON... yes
configure: error: r600 requires libelf when using llvm

elfutils dependency is needed by r600 with llvm support and radeonSI.

Fixes:
checking for RADEON... yes
checking for AMDGPU... yes
configure: error: radeonsi requires libelf when using llvm

Add a new BR2_PACKAGE_MESA3D_NEEDS_ELFUTILS hidden Config.in boolean
selected by R600 and RadeonSI at Kconfig level.
When selected, BR2_PACKAGE_MESA3D_NEEDS_ELFUTILS add elfutils dependency
at make level.

[1] 92fda8761aa8024be53a8edb8c102d1ba6401bce

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Michał Kalbarczyk <michal@gamecodehq.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/mesa3d: RadeonSI needs llvm support

Fixes:
checking for AMDGPU... yes
configure: error: --enable-llvm is required when building radeonsi

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Michał Kalbarczyk <michal@gamecodehq.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/mesa3d: RadeonSI needs libdrm-radeon and libdrm-amdgpu

Fixes:
checking for EXPAT... yes
checking for RADEON... no
configure: error: Package requirements (libdrm >= 2.4.71 libdrm_radeon >= 2.4.71)

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Michał Kalbarczyk <michal@gamecodehq.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/mesa3d: radeonSI: specify drm platform

When BR2_PACKAGE_MESA3D_OPENGL_EGL is selected with RadeonSI drvier,
mesa platform must be specified with drm.

Fixes:
hecking for RADEON... yes
checking for AMDGPU... yes
configure: error: radeonsi requires one of these:
                  1) --with-platforms=drm (X, Wayland, offscreen rendering based on DRM)
                  2) --with-platforms=surfaceless (offscreen only)
                  3) --with-platforms=android (Android only)
                  Recommended options: drm,x11

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libimxvpuapi: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libiio: bump to version 0.14

Add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

meson: bump version to 0.46.0

Bump version and drop obsolete patch.

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

vlc: switch to proper Git formatted patches

VLC uses Git as its version control system, so it makes sense to use
Git formatted patches.

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

package/vlc: fix build with updated x264

Fixes
http://autobuild.buildroot.net/results/909/909d4c1b4bf8b3135607025e12363db1b2196011/

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

msmtp: add optional libidn dependency

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

prosody: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

mutt: bump to version 1.9.5

Add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libgeotiff: bump to version 1.4.2

- Add patch to fix autoheader issue: "missing template:
  GEO_NORMALIZE_DISABLE_TOWGS84"
- Add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

skeleton: add /dev/fd, /dev/std{in, out, err} symlinks for static /dev on readonly rootfs

Some applications, e.g. bashs process subsitution feature, rely on the
convention of `/dev/fd` being a symbolic link to `/proc/self/fd`.

When a static /dev is used on a readonly rootfs then the runtime ln
invocations in the inittab will fail, so we need to add the symlinks at
build time.  Makedevs doesn't support creating symlinks, so instead add the
symlinks to the default skeleton.

For non-static /dev setups, the kernel will mount devtmpfs which shadows the
/dev of the rootfs, but then the runtime ln invocations in inittab will
create the symlinks.

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

sysvinit: add /dev/std{in, out, err} symlinks to inittab

Some applications, e.g. bashs process subsitution feature, rely on the
convention of `/dev/fd` being a symbolic link to `/proc/self/fd`.

This symbolic link and his companions `/dev/std*` are created by (e)udev [1],
but not by mdev, resulting in the following error when using the following
expression:

```
bash: /dev/fd/62: No such file or directory
```

For the sake of simplicity, lets fix this by creating the symlinks in inittab.
It is only really needed if eudev isn't used, but it doesn't really hurt to
create them even if eudev will recreate them afterwards.

Note, that we do not create the symlink `/dev/core` as `/proc/kcore` is
not available on all platforms, e.g. ARM, and the feature is not much
appreciated [2].

[1] https://github.com/gentoo/eudev/blob/8943501993322c59a6eb5be456b0d716aafff21e/src/shared/dev-setup.c#L35-L40
[2] https://lwn.net/Articles/45315/

[Peter: redirect errors to /dev/null for ro rootfs]
Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

busybox: add /dev/std{in, out, err} symlinks to inittab

Some applications, e.g. bashs process subsitution feature, rely on the
convention of `/dev/fd` being a symbolic link to `/proc/self/fd`.

This symbolic link and his companions `/dev/std*` are created by (e)udev [1],
but not by mdev, resulting in the following error when using the following
expression:

```
bash: /dev/fd/62: No such file or directory
```

For the sake of simplicity, lets fix this by creating the symlinks in inittab.
It is only really needed if eudev isn't used, but it doesn't really hurt to
create them even if eudev will recreate them afterwards.

Note, that we do not create the symlink `/dev/core` as `/proc/kcore` is
not available on all platforms, e.g. ARM, and the feature is not much
appreciated [2].

[1] https://github.com/gentoo/eudev/blob/8943501993322c59a6eb5be456b0d716aafff21e/src/shared/dev-setup.c#L35-L40
[2] https://lwn.net/Articles/45315/

[Peter: redirect output (errors) to /dev/null for ro rootfs]
Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libidn: bump to version 1.34

- Remove both patches (already in version)
- Remove AUTORECONF = YES (patch removed)
- Add hash for license files

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

bluez-alsa: fix select of libbsd

bluez-alsa/Config.in must select BR2_PACKAGE_LIBBSD, not the
non-existing BR2_PACKAGE_BSD.

Fixes:

  http://autobuild.buildroot.net/results/ab6751615ac058d892258be9e7a8e110e68c62c5/

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

libg7221: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

s6-linux-init: bump version to 0.4.0.0

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

s6-rc: bump version to 0.4.0.1

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

s6: bump version to 2.7.1.1

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

execline: bump version to 2.5.0.0

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

skalibs: bump version to 2.6.4.0

Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

msmtp: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

msmtp: add optional support for libgsasl

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

iproute2: bump to version 4.16.0

Drop uClibc-ng compatibility patch. Current uClibc-ng provides the
needed definitions.

Drop the libc-config.h musl compatibility workaround patch. iproute2
update its headers copy to current kernel version for which this
workaround is no longer needed.

Don't modify tc/Makefile to make XT_LIB_DIR get the value of
IPT_LIB_DIR. Pass the correct value directly in CFLAGS.

Add a patch removing arpd dependency on threading support.

Rename IPROUTE2_WITH_IPTABLES to IPROUTE2_DISABLE_IPTABLES which better
describes what it does.

Don't patch the configure script for the cross compiler. The CC
environment variable is enough.

Don't force the CCOPTS make variable. The CFLAGS environment variable
does the trick for target code.

Don't patch CCOPTS out of netem/Makefile. The CBUILD_CFLAGS environment
variable can pass the host CFLAGS to host code.

The ifcfg script is now POSIX shell compatible. Remove its installation
dependency on bash.

Pass installation DESTDIR in the environment. Don't force SBINDIR, DOCDIR, and
MANDIR. The default values are fine.

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

bluez-alsa: needs NPTL

bluez-alsa calls pthread_setname_np() which is only provided by the NPTL
pthread implementation.

Fixes:
http://autobuild.buildroot.net/results/707/707fd46921b1cfda90e9c30be33ca918f2d0cb00/
http://autobuild.buildroot.net/results/7a6/7a632c2b6b9021aa6edcd5126400de26e2b8ca42/

Cc: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

download/git: always do full-clone

We currently attempt a shallow clone, as tentative to save bandwidth and
download time.

However, now that we keep the git tree as a cache, it may happen that we
need to checkout an earlier commit, and that would not be present with a
shallow clone.

Furthermore, the shallow fetch is already really broken, and just
happens to work by chance. Consider the following actions, which are
basically what happens today:

    mkdir git
    git init git
    cd git
    git remote add origin https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
    git fetch origin --depth 1 v4.17-rc1
    if ! git fetch origin v4.17-rc1:v4.17-rc1 ; then
        echo "warning"
    fi
    git checkout v4.17-rc1

The checkout succeeds just because of the git-fetch in the if-condition,
which is initially there to fetch the special refs from github PRs, or
gerrit reviews. That fails, but we just print a warning. If we were to
ever remove support for special refs, then the checkout would fail.

The whole purpose of the git cache is to actually save bandwidth and
download time, but in the long run. For one-offs, people would
preferably use a wget download (e.g. with the github macro) instead of
a git clone.

We switch to always doing a full clone. It is more correct, and pays off
in the long run...

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

download/git: ensure we can checkout repos with submodule conversions

When a git tree has had sub-dir <-> sub-module conversions, or has had
submodules added or removed over the course of time, checking out a
changeset across those conversions/additions/removals may leave
untracked files, or may fail because of a conflict of type.

So, before we checkout the new changeset, we forcibly remove the
submodules. The new set of submodules, if any, will be restored later.

Ideally, we would use a native git command: git submodule deinit --all.
However, that was only introduced in git 1.8.3 which, while not being
recent by modern standards, is still too old for some enterprise-grade
distributions (RHEL6 only has git-1.7.1).

So, instead, we just use git submodule foreach, to rm -rf the submodules
directory.

Again, we would ideally use 'cd $toplevel && rm -rf $path', but
$toplevel was only introduced in git 1.7.2. $path has always been there.

So, instead, we just cd back one level, and remove the basename of the
directory.

Eventually, we need to get rid of now-empty and untracked directories,
that were parents of a removed submodule. For example. ./foo/bar/ was a
submodule, so ./foo/bar/ was removed, which left ./foo/ around.

Yet again, recent-ish git versions would have removed it during the
forced checkout, but old-ish versions (e.g. 1.7.1) do not remove it with
the forced checkout.

Instead we rely on the already used forced-forced clean of directories,
untracked, and ignored content, to really get rid of extra stuff we are
not interested in.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Reviewed-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

download/git: ensure we checkout to a clean state

Force the checkout to ignore and throw away any local changes. This
allows recovering from a previous partial checkout (e.g. killed by
the user, or by a CI job...)

git checkout -f has been supported since the inception of git, so we
can use it without any second thought.

Also do a forced-forced clean, to really get rid of all untracked stuff.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Reviewed-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

download/git: try to recover from utterly-broken repositories

In some cases, the repository may be in a state we can't automatically
recover from, especially since we must still support oldish git versions
that do not provide the necessary commands or options thereof.

As a last-ditch recovery, delete the repository and recreate the cache
from scratch.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

download/git: quickly exit when the cset does not exist

Check that the given cset is indeed something we can checkout. If not,
then exit early.

This will be useful when a later commit will trap any failing git
command to try to recover the repository by doing a clone from scratch:
when the cset is not a commit, it does not mean the repository is broken
or what, and re-cloning from scratch would not help, so no need to trash
a good cache.

Reported-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Reviewed-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

download/git: run all git commands in the current directory

That way, we can pushd earlier, which will help with last-ditch recovery
in a followup commit.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Reviewed-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

download/git: add warning not to use our git cache

We really want the user not to use our git cache manually, or their
changes (committed or not) may eventually get lost.

So, add a warning file, not unlike the one we put in the target/
directory, to warn the user not to use the git tree.

Ideally, we would have carried this file in support/misc/, but the git
backend does not have access to it: the working directory is somewhere
unknown, and TOPDIR is not exported in the environment.

So, we have to carry it in-line in the backend instead.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Reviewed-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

cups: disable pam support

Make the build consistent.

Cc: Olivier Schonken <olivier.schonken@gmail.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Acked-by: Olivier Schonken <olivier.schonken@gmail.com>
Signed-off-by: Baruch Siach &lt;<a href="mailto:baruch@tkos.co.il" target="_blank" rel="noreferrer">baruch@tkos.co.il</a>&gt;<br>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

core/pkg-generic: only save latest package list

When rebuilding a package, simply appending the package's file list to
the global list means that the package list grows for every rebuild, as
does the time taken to check for files installed by multiple packages.
Furthermore, we get false positives where a file is reported as being
installed by multiple copies of the same package.

With this approach we may end up with orphaned files in the target
filesystem if a package that has been updated and rebuilt no longer
installs the same set of files, but we know that only a clean build will
produce reliable results.  In fact it may be helpful to identify these
orphaned files as evidence that the build is not clean.

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

lftp: bump to version 4.8.3

Add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libcurl: fix building against libressl

LibreSSL 2.7.x breaks libcurl 7.59.0 with the error:
error: static declaration of ‘OpenSSL_version_num’ follows non-static
declaration

This failure has since been fixed upstream with commit:
https://github.com/curl/curl/commit/7c90c93c0b061da81f69fabdd57125b2783c15fb

Signed-off-by: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

gtksourceview: bump to version 3.24.7

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

cups: fix build without PAM

Add upstream patch removing MD5 code that builds when PAM is not
enabled.

Fixes:
http://autobuild.buildroot.net/results/48d/48d53bdeceafdbb51756d5c0b9936ce7c98b4ddc/
http://autobuild.buildroot.net/results/13c/13cb03a8d0ae215c418fe2520a27b2940efd4a39/
http://autobuild.buildroot.net/results/fd3/fd3c56b60bb0411b38dc7279880d1fab7228c20a/

Cc: Olivier Schonken <olivier.schonken@gmail.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Acked-by: Olivier Schonken <olivier.schonken@gmail.com>
Signed-off-by: Baruch Siach &lt;<a href="mailto:baruch@tkos.co.il" target="_blank" rel="noreferrer">baruch@tkos.co.il</a>&gt;<br>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libgudev: bump to version 232

Add --disable-umockdev (enabled by default)
Add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libgtk3: bump to version 3.22.30

Add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux-headers: bump 4.{9, 14, 16}.x series

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

linux: bump default to version 4.16.6

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libid3tag: add hash for license files

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libiconv: add hash for license files

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libical: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libhttpparser: bump to version 1.8.1

Add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libhid: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

libglade: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

sdl2: update license file hash

Copyright year update.

Fixes:
http://autobuild.buildroot.net/results/2c8/2c865463a4b7524114518c04dce9c94252433460/

Cc: Guillaume Gardet <guillaume.gardet@oliseo.fr>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

imx-alsa-plugins: new package

Freescale specific alsa-lib plugins for the i.MX processor family.

For now, the only provided plugin is:
  * Rate Converter Plugin Using Freescale ASRC Resampler

The ASRC hardware resampler can be found on i.MX53 and most i.MX 6
processors.

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
[Thomas:
 - Remove the mention "This package is provided by Freescale as-is and
   doesn't have an upstream." from the Config.in help text, since
   according to Gary Bisson, there is a Git repository for such
   projects, and therefore some form of upstream.
 - Add missing entry to DEVELOPERS file.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

python-influxdb: new package

InfluxDB client.

Signed-off-by: Sven Haardiek <sven.haardiek@iotec-gmbh.de>
Reviewed-by: Yegor Yefremov <yegorslists@googlemail.com>
[Thomas: as suggested by Yegor, add entry in the DEVELOPERS file, and
fix commit title.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/pkg-generic: add check that target variant is defined before host variant

Update the documentation accordingly.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
[yann.morin.1998@free.fr: slight rephrasing in error message, update manual]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/qemu: declare target variant before host variant

Our package infrastructure uses inheritance of a number of values from
the target package to the host package, which assumes the target
package is defined before the host package. In addition, future
changes are going to make this requirement even more important.

Therefore, let's fix the qemu package so that it declares its target
variant before its host variant, like all other packages in
Buildroot. We handle qemu separately from other packages, because
unlike other packages, it didn't had the "eval" for the host and
target packages at the end of the file, but rather all variables
related to the host variant first, then the call to the package
infrastructure for the host variant, then the variables related to the
target variant, and finally the call to the package infrastructure for
the target variant. We are inverting the order of those two big parts
in this commit.

Signed-off-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@bootlin.com>

package: declare target variant before host variant

Our package infrastructure uses inheritance of a number of values from
the target package to the host package, which assumes the target
package is defined before the host package. In addition, future
changes are going to make this requirement even more important.

Therefore, let's fix the android-tools, gauche, lcms2,
linux-syscall-support and pngquant packages, so that they declare
their target variant before their host variant, like all other
packages in Buildroot.

Signed-off-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@bootlin.com>

libgtk2: bump to version 2.24.32

Add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libgsm: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libgsasl: add hash for license files

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libgphoto2: bump to version 2.5.17

Add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libebml: fix build on some toolchains

Replace climits include by limits (retrieve upstream patch from:
https://github.com/Matroska-Org/libebml/commit/e46906b80e7662fb78d305f576412f9fa4a22218)

Fixes:
 - http://autobuild.buildroot.net/results/3c2402a7052acb67fc7e03ad2ff27e84534aaa36

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

sdl2_image: security bump to version 2.0.3

Fixes the following security issues:

CVE-2017-12122: An exploitable code execution vulnerability exists in the
ILBM image rendering functionality of SDL2_image-2.0.2.  A specially crafted
ILBM image can cause a heap overflow resulting in code execution.  An
attacker can display a specially crafted image to trigger this
vulnerability.

CVE-2017-14440: An exploitable code execution vulnerability exists in the
ILBM image rendering functionality of SDL2_image-2.0.2.  A specially crafted
ILBM image can cause a stack overflow resulting in code execution.  An
attacker can display a specially crafted image to trigger this
vulnerability.

CVE-2017-14441: An exploitable code execution vulnerability exists in the
ICO image rendering functionality of SDL2_image-2.0.2.  A specially crafted
ICO image can cause an integer overflow, cascading to a heap overflow
resulting in code execution.  An attacker can display a specially crafted
image to trigger this vulnerability.

CVE-2017-14442: An exploitable code execution vulnerability exists in the
BMP image rendering functionality of SDL2_image-2.0.2.  A specially crafted
BMP image can cause a stack overflow resulting in code execution.  An
attacker can display a specially crafted image to trigger this
vulnerability.

CVE-2017-14448: An exploitable code execution vulnerability exists in the
XCF image rendering functionality of SDL2_image-2.0.2.  A specially crafted
XCF image can cause a heap overflow resulting in code execution.  An
attacker can display a specially crafted image to trigger this
vulnerability.

CVE-2017-14449: A double-Free vulnerability exists in the XCF image
rendering functionality of SDL2_image-2.0.2.  A specially crafted XCF image
can cause a Double-Free situation to occur.  An attacker can display a
specially crafted image to trigger this vulnerability.

CVE-2017-14450: A buffer overflow vulnerability exists in the GIF image
parsing functionality of SDL2_image-2.0.2.  A specially crafted GIF image
can lead to a buffer overflow on a global section.  An attacker can display
an image to trigger this vulnerability.

For details, see the announcement:

https://discourse.libsdl.org/t/sdl-image-2-0-3-released/23958

Also add a hash for the license file while we're at it.

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

sdl2: bump version to 2.0.8

Drop now upstreamed patch.

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

libglu: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libglob: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libglfw: add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libcue: fix build without C++

By default, CMake assumes that the project is using both C and C++. By
explicitly passing 'C' as argument of the project() macro, we tell
CMake that only C is used, which prevents CMake from checking if a C++
compiler exists.
Patch sent upstream: https://github.com/lipnitsk/libcue/pull/17

Fixes:
 - http://autobuild.buildroot.net/results/a5344d18dbdc86705853fbf8a66d5304de07a9b2

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

pcmanfm: bump to version 1.3.0

Add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

menu-cache: bump to version 1.1.0

Add hash for license file

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

libfm-extra: bump to version 1.3.0.2

Add hash for license files

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>