package/gcc: remove gcc 4.9

In order to reduce the number of choice in gcc selection, remove the
gcc 4.9 version.

This version was kept due to libstdc++ ABI-incompatible changes and
other build issues with kernel and bootloader as reported by Arnout
[1].

Since then, gcc 4.9 is not supported any more since glibc 2.29 [2]
and recent kernel and bootloaders has been fixed to use more recent
compiler version.

[1] http://lists.busybox.net/pipermail/buildroot/2017-June/194374.html
[2] https://www.sourceware.org/ml/libc-alpha/2019-01/msg00723.html

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

configs/odroidc2: remove the defconfig

The odroidc2 defconfig is using gcc 4.9 version in order to build
uboot (2015.01) and kernel (3.14.79) for the board.

We are going to remove gcc 4.9 version in Buildroot and this
defconfig is the only remaining one using this version.

Since we don't have the board, we can't update the defconfig with
newer bootloader and kernel version.

A new defconfig for the odroidc2 board are welcome as soon as
it use a newer uboot and kernel version.

Remove the defconfig from the gitlab yaml.

Signed-off-by: Romain Naour <romain.naour@smile.fr>
Cc: Dagg Stompler <daggs@gmx.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/linux-headers: apply all Linux patches when BR2_KERNEL_HEADERS_AS_KERNEL=y

When BR2_KERNEL_HEADERS_AS_KERNEL=y, we expect that the Linux kernel
headers code will be exactly the same as the Linux kernel code
itself. The code currently takes into account the patches defined by
BR2_LINUX_KERNEL_PATCH, but not the kernel patches that are stored in
linux's BR2_GLOBAL_PATCH_DIR.

So for example, the current qemu_riscv32_virt_defconfig has:

BR2_GLOBAL_PATCH_DIR="board/qemu/riscv32-virt/patches/"

With:

board/qemu/riscv32-virt/patches/
└── linux
    └── 0001-Revert-riscv-Use-latest-system-call-ABI.patch

This patch gets properly applied when the Linux kernel is built, but
not when the linux-headers package is built.

This commit fixes that by making sure patches stored in the "linux"
BR2_GLOBAL_PATCH_DIR subdirectory are taken into account.

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

package/python-redis: new package

The Python interface to the Redis key-value store.

https://github.com/andymccurdy/redis-py

Signed-off-by: Grzegorz Blach <grzegorz@blach.pl>
[Peter: hiredis is an optional runtime dependency, not build time]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python-hiredis: new package

Python extension that wraps protocol parsing code in hiredis.
It primarily speeds up parsing of multi bulk replies.

https://github.com/redis/hiredis-py

Signed-off-by: Grzegorz Blach <grzegorz@blach.pl>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/libcurl: introduce options to extend/reduce feature set

Libcurl is more than 250 KiB (libcurl) / 100 KiB (curl binary) in size.
About 50 KiB / 15 KiB of this can be saved by disabling features/protocols
that are not commonly needed:

- proxy support: 15 KiB
- cookies support: 10 KiB
- various less common protocols: 25 KiB (libcurl) + 15 KiB (curl binary)

Note that the exact amount of space saved depends on the architecture,
toolchain, and other factors.

Other packages that are selecting libcurl might require protocols from the
'extra' set. But, there is no clear way to find out which packages are in
this situation, in particular because issues may only be visible at runtime.

Note: remove the text 'enable' on the option for 'verbose strings' as that
is more common in Buildroot.

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
[Peter: unconditionally remove the libcurl-option to generate C code]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python-urllib3: bump version to 1.25.3

Updated license hash after upstream commit:
https://github.com/urllib3/urllib3/commit/2a0957ea27e966166f81f88693af8e2f87d19fb6

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

package/domoticz: bump to version 0f411f781ae4fb4a82f592d38a3f40578c149533

- Replace first patch to fix build when using a development version
- Fix build with openzwave 1.6 through:
  https://github.com/domoticz/domoticz/commit/56d3fa099d9c98222d6cf5043891c6ba2f1df6bf
- Add patch to revert cmake 3.14 dependency as buildroot currently
  requires only 3.8

Fixes:
 - http://autobuild.buildroot.org/results/3dbb6612205020cc156d245a75c76a7886f7d529

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

package/python-mwclient: bump version to 0.10.0

Switched license file to LICENSE.md.

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

package/python-requests-oauthlib: bump version to 1.2.0

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

package/python-oauthlib: bump version to 3.0.2

Updated license hash after upstream bump to 2019
https://github.com/oauthlib/oauthlib/commit/f4273e75250dd36c88d63dc075ae45650a5172e9

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

package/python-certifi: bump version to 2019.6.16

Updated project URL to https.

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

package/ddrescue: remove patch

Not needed anymore after upstream change:
https://lists.gnu.org/archive/html/bug-ddrescue/2017-02/msg00002.html

Build-tested using br-x86-64-core2-full.config

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

package/gsl: bump version to 2.5

Added license hash.

Release notes:
https://lists.gnu.org/archive/html/info-gnu/2018-06/msg00007.html

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

package/xorriso: bump version to 1.5.0

Removed patch for sys/xattr.h handling and replaced it with the new
upstream-provided configure option --disable-xattr-h-pref-attr
https://dev.lovelyhq.com/libburnia/libisofs/commit/848e039e6d5222b07da7ec810ce4cca1c60bd935

Removed autoreconf, not needed anymore.

Added license hashes.

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

package/time: add upstream sha1 hash

Also added a reference for upstream-provided sha1 & sha256 hashes.

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

package/zstd: bump version to 1.4.2

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

package/ddrescue: bump version to 1.24

Added license hash and upstream sha256 hash for tarball.

Release notes:
1.23: https://lists.gnu.org/archive/html/info-gnu/2018-02/msg00004.html
1.24: https://lists.gnu.org/archive/html/info-gnu/2019-02/msg00012.html

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

package/gdbm: bump version to 1.18.1

Added license hash.

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

package/mtools: bump version to 4.0.23

Switched to .lz tarball.

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

package/neon: add license hash

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

package/ncurses: add license hash

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

package/mjpegtools: add license hash

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

package/lzo: add license hash

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

package/libsilk: add license hash

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

package/kompexsqlite: add license hash

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

package/iostat: add license hash

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

package/haserl: add license hash

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

package/giflib: add license hash

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

package/ftop: add license hash

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

package/flot: add license hash

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

package/flickcurl: add license hash

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

package/fbdump: add license hash

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

package/enscript: add license hash

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

package/ebtables: add license hash

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

package/dfu-util: add license hash

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

package/cegui06: add license hash

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

package/cairomm: add license hash

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

package/nfs-utils: bump version to 1.3.4

Added license hash.

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

package/diffutils: bump version to 3.7

Added license hash.

Release notes:
https://lists.gnu.org/archive/html/info-gnu/2019-01/msg00000.html

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

package/python-twisted: bump to version 19.2.1

Also change hash for the license file due to a year change.

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

package/python-docutils: bump to version 0.15.2

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

package/python-cffi: bump to version 1.12.3

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

package/python-autobahn: bump to version 19.8.1

In addition:
 - select python-cryptography as it's now a runtime dependency
 - Fix a typo in the help.

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

[PATCH 01/05 package/python-cryptography: bump to version 2.7

Also change the hash for LICENSE.APACHE due to changing http to https
in the license URL.

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

support/config-fragments/autobuild: set gcc version for RISC-V toolchains

Prior to b3ba26150d49dbc6b06ad402c5f6fbd398a71451
("toolchain/toolchain-external/toolchain-external-custom: be more
flexible on gcc version"), the default gcc version selected by
Buildroot for custom external toolchain was affected by the
BR2_ARCH_NEEDS_GCC_AT_LEAST_xyz definitions.

Since BR2_riscv selects BR2_ARCH_NEEDS_GCC_AT_LEAST_7, gcc 7.x was the
default gcc version assumed to be used in a custom RISC-V external
toolchain, so our config snippets for RISC-V toolchains were correct.

With b3ba26150d49dbc6b06ad402c5f6fbd398a71451 applied, the default gcc
version assumed for custom external toolchains is the latest one
(currently gcc 9.x), while our RISC-V toolchains use gcc 7.x. So we
now need to explicitly give the gcc version used by our RISC-V
toolchains, otherwise the build fails with:

  Incorrect selection of gcc version: expected 9.x, got 7.4.0

Fixes:

  http://autobuild.buildroot.net/results/b872befe1adec2633b9cbcc49bc0eb7619f606c2/

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

libdrm: change to meson build system

- remove legacy patch
  0003-configure-Makefile.am-use-pkg-config-to-discover-lib.patch

- add patch to fix meson atomic ops detection
  0003-meson.build-fix-intel-atomics-detection.patch

- add patch to enable static build
  0004-meson.build-enable-static-build.patch

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

package/elfutils: fix build with glibc < 2.16

Fixes:
 - autobuild.buildroot.net/results/1053e2b4b51bc225c4a1a29c93946101a7a53be9

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

package/qemu: drop host kernel version check

There is no clean way to check if a program will actually run using
host-qemu, making this check too restrictive.

Add a warning in the help text.

Signed-off-by: Adam Duskett <Aduskett@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/python-zeroconf: depends on ifaddr instead of netifaces

Starting from 0.21.0 zeroconf uses pure-python ifaddr module
instead of netifaces.

Currently we have zeroconf 0.23.0, so this module raises
ModuleNotFoundError exception during import.

Signed-off-by: Grzegorz Blach <grzegorz@blach.pl>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/python-ifaddr: new package

Enumerates all IP addresses on all network adapters of the system.

https://github.com/pydron/ifaddr

Signed-off-by: Grzegorz Blach <grzegorz@blach.pl>
[Thomas: add license file.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

docs/manual: document providers from br2-external

Add documentation about how a br2-external tree can provide an external
toolchain or a libjpeg or openssl alternative implementation.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vadim Kochan <vadim4j@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

core: allow br2-external trees to provide opensl

Similar to toolchains and jpeg, we now offer a way for br2-external
trees to provide their openssl implementation, which gets included in
the openssl choice.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vadim Kochan <vadim4j@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

core: allow br2-external trees to provide libjpeg

Similar to toolchains, we now offer a way for br2-external trees to
provide their libjpeg implementation, which gets included in the jpeg
choice.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vadim Kochan <vadim4j@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

core: allow br2-external trees to provide pre-configured toolchains

Since we have a choice for the pre-configured pre-built toolchains,
there is no possbility for a br2-external to provide its own. The
only solution so far for defconfigs in br2-external trees is to use
BR2_TOOLCHAIN_EXTERNAL_CUSTOM and define all the bits by itself...

This is not so convemient, so offer a way for br2-external trees to
provide such pre-configured toolchains.

To allow for this, we now scan each br2-external tree and look for a
specific file, provides.toolchains.in. We generate a kconfig file that
sources each such file, and that generated file is sourced from within
the toolchain choice, thus making the toolchains from a br2-external
tree possible and available in the same location as the ones known to
Buildroot:

    Toolchain  --->
        Toolchain type (External toolchain)  --->
        Toolchain  --->
            (X) Arm ARM 2019.03
            ( ) Linaro ARM 2018.05
            ( ) Custom toolchain
                *** Toolchains from my-br2-ext-tree: ***
            ( ) My custom ARM toolchain
                *** Toolchains from another-br2-ext-tree: ***
            ( ) Another custom ARM toolchain
            ( ) A third custom ARM toolchain

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vadim Kochan <vadim4j@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

core: split generated kconfig file

Currently, the kconfig part contains two things: the kconfig option
with the paths to br2-external trees, and the kconfig menus for the
br2-external trees.

When we want to include more kconfig files from the br2-external tree
(e.g. to get definitions for pre-built toolchains), we will need to
have the paths defined earlier, so they can be used from the br2-external
tree to include files earlier than the existing menus.

Split the generated kconfig file in two: one to define the paths, which
gets included early in our main Config.in, and one to actually define
the existing menus, which still gets included at the same place they
currently are.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vadim Kochan <vadim4j@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

toolchain: allow PIC/PIE without RELRO

In commit 7484c1c3b806 (toolchain/toolchain-wrapper: add BR2_RELRO_),
we added the PIC/PIE flags, but based on the RELRO_FULL condition.

It is however totally possible to do a PIC/PIE executable without
RELRO_FULL, as it is also valid to do a PIC/PIE build with RELRO_PARTIAL.

Add a new option that now governs the PIC/PIE flags.

Note: it is unknown if RELRO_FULL really needs PIC/PIE or not, so we
keep the current situation, where RELRO-FULL forces PIC/PIE compilation.
Decoupling can come later from an interested party.

Signed-off-by: "Yann E. MORIN" <yann.morin@orange.com>
Cc: Matt Weber <matthew.weber@rockwellcollins.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Reviewed-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

toolchain: -fstack-protector-strong can be back-ported

Currently, use of -fstack-protector-strong is only available for gcc
starting with 4.9, on the assumption that it appeared with that version.

Although this is true, it happens that quite a few vendors will have
back-ported -fstack-protector-strong to older gcc versions (at least 4.8
seen in the wild).

Remove the guard against gcc>=4.9, and expand the help text.

Note: we could have changed the guard to something like:
    depends on BR2_TOOLCHAIN_GCC_AT_LEAST_4_9 || BR2_TOOLCHAIN_EXTERNAL_CUSTOM

However, the latest gcc we support in the internal toolchain now *is*
gcc-4.9, and similarly all external toolchains except Sourcery ARM are
4.9 or higher. So except for the Sourcery toolchain, the condition would
have always been true. For that one toolchain, we can allow it to hit
the SSP check, and just drop the condition entirely.

Signed-off-by: "Yann E. MORIN" <yann.morin@orange.com>
Cc: Matt Weber <matthew.weber@rockwellcollins.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

toolchain: check the SSP option is known

Some toolchain vendors may have backported those options to older gcc
versions, and we have no way to know, so we have to check that the
user's selection is acceptable.

Extend the macro that currently checks for SSP in the toolchain, with
a new test that the actual SSP option is recognised and accepted.

Note that the SSP option is either totaly empty, or an already-quoted
string, so we can safely and easily assign it to a shell variable to
test and use it.

Note that we do not introduce BR2_TOOLCHAIN_HAS_SSP_STRONG, because:

  - our internal toolchain infra only supports gcc >= 4.9, so it has
    SSP strong;

  - of the external pre-built toolchains, only the codesourcery-arm
    one has a gcc-4.8 which lacks SSP strong, all the others have a
    gcc >= 4.9;

  - we'd still have to do the actual check for custom external
    toolchains anyway.

So, we're not adding BR2_TOOLCHAIN_HAS_SSP_STRONG just for a single
case.

Signed-off-by: "Yann E. MORIN" <yann.morin@orange.com>
Cc: Matt Weber <matthew.weber@rockwellcollins.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/gdb: update host gdb minimum host gcc version

Following gdb 7.12.1 removal [1] the host gcc version needs to be updated
since gdb >= 8.x now requires a C++11 compiler (gcc >= 4.8).

While at it, move BR2_HOST_GCC_AT_LEAST_4_8 under BR2_PACKAGE_HOST_GDB since
it's not an architecture dependency.  Add a comment when the host gcc is too
old to build host gdb.

[1] d36f2c73331b3ab4e238eeb47d7d79d81a895a59

Fixes:
http://autobuild.buildroot.org/results/822/822a747a6717db57705d1ce198a61988aa1173b1

Signed-off-by: Romain Naour <romain.naour@smile.fr>
Cc: Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/ethtool: bump to version 5.2

Signed-off-by: Pierre-Jean Texier <pjtexier@koncepto.io>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/bzip2: security bump version to 1.0.8

Switched to new maintainer source:
https://sourceware.org/ml/bzip2-devel/2019-q2/msg00022.html

Version 1.0.7 fixes CVE-2016-3189 & CVE-2019-12900

Version 1.0.8 fixes the fix for CVE-2019-12900 from 1.0.7:
https://sourceware.org/ml/bzip2-devel/2019-q3/msg00031.html

Rebased 0002-improve-build-system.patch.

Removed 0003-Make-sure-nSelectors-is-not-out-of-range.patch, applied
upstream:
https://sourceware.org/git/?p=bzip2.git;a=commitdiff;h=7ed62bfb46e87a9e878712603469440e6882b184
and reverted later on
https://sourceware.org/git/?p=bzip2.git;a=commitdiff;h=b07b105d1b66e32760095e3602261738443b9e13

Added upstream sha512 hash and updated license hash after upstream
commits:
https://sourceware.org/git/?p=bzip2.git;a=history;f=LICENSE;h=81a37eab7a5be1a34456f38adb74928cc9073e9b;hb=HEAD

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

package/libhdhomerun: bump version to 20190621

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

core: prepare for generating multiple kconfig fragments

We currently redirect the output of each helper function. This was nice
as long as we were generating single .mk and .in fragments.

But we are soon to need more .in fragments.

So, do the redirection inside the .in helpers.

We do not (currently) need to generate more than one .mk fragment, but
for consistency, do the redirection in the .mk helper too.

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Vadim Kochan <vadim4j@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

core: drop now-useless prepare-kconfig rule

This rule was added back in 9429e7b698 (core: introduce an intermediate
rule before the configurators) when the kconfig-side br2-external file
was generated separately from the Makefile-side one.

Now that they are generated together very early in the Makefile, we no
longer need this intermediate rule. Drop it.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vadim Kochan <vadim4j@gmail.com>
[Peter: also drop outdated reference in the manual]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

core: generate all br2-external files in one go

When we introduced support for multiple br2-external trees, we
introduced two files, one on the Makefile side, needed very early,
and one on the kconfig side, needed later in the configuration
process. We naturally introduced a two-step generation, as it looked
like the simplest and most obvious way.

But now, we are on the verge of generating more files on the kconfig
side, and it does not make sense to add even more steps to generate
them.

And even better yet, we can generate both the Makefile-side and
kconfig-side files at the same time, in fact.

Make it so.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vadim Kochan <vadim4j@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

core: simplify removal of generated br2-external files

Now that all the br2-external generated files are named after the same
pattern, it gets easier to remove them all using a glob.

Furthermore, we're on the verge of introducing more such generated
files, so removing them at one fell swoop will be simpler too.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vadim Kochan <vadim4j@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

core: rename generated .br-external.mk file

Now that the two (all of them!) br2-external related files are generated
in the same location, it makes sense they are named after the same
pattern.

When initial support for (then single) br2-external trees was added back
in a4239f7fd1 (core: introduce the BR2_EXTERNAL variable), it was not
clear-cut why that file was not named with a br2 prefix.

So rename it now.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vadim Kochan <vadim4j@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

core: move generated .br2-external kconfig file to $(BASE_DIR)

Currently, that file is generated rather late in the configuration
process, so BUILD_DIR is known (and exists) by then.

We're soon to generate that file much earlier, at a point where
BUILD_DIR is not yet known, so we have two options:
 1- declare BUILD_DIR earlier;
 2- generate the file in an already-known location.

We go with the second solution, as we're already generating a
br2-external related file in BASE_DIR, so we can as well generate all
br2-external files in the same place.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vadim Kochan <vadim4j@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

.gitlab-ci.yml: regenerate after adding TestCheckPackage

When adding the check-package test, the committer (Arnout) merged the
TestCheckPackageBasicUsage class into the TestCheckPackage class, but
failed to regenerate .gitlab-ci.yml. Do this now.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/icu: Fix nios2 build

Fixes:
http://autobuild.buildroot.net/results/91e/91eaec34708d91f8a05af189243be0b7cabce31b/

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/dcron: add patch to fix error message

The error message issued when the creation of the log file fails lacks
an ending newline. Add a patch already submitted upstream[1] to fix it.

1. https://github.com/dubiousjim/dcron/pull/22

Signed-off-by: Carlos Santos <unixmania@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/qt5base: fix libtool la file dependency_libs entries

Fixes [1]:

  libtool: error: cannot find the library '' or unhandled argument '/.../host/riscv64-buildroot-linux-gnu/sysroot/usr/lib/libQt5Widgets.so'

Add upstream suggested patch ([2]) to change la file dependency_libs entries
to -L<path> -l<library> version.

[1] http://autobuild.buildroot.net/results/79c1e1b7a1bc53c1e9b2ae0c9acb443e6d2e2994
[2] https://codereview.qt-project.org/c/qt/qtbase/+/269146

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Tested-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/python3: adjust _REMOVE_USELESS_FILES fix for new layout

python3 nowadays appends the triplet to the config-<version>m directory:

echo target/usr/lib/python3.7/config-*
target/usr/lib/python3.7/config-3.7m-powerpc-linux-gnu

Likewise, there is no longer a pyconfig.h:

ls target/usr/lib/python3.7/config-3.7m-powerpc-linux-gnu
config.c  config.c.in  install-sh  libpython3.7m.a  Makefile
makesetup  python-config.py  python.o  Setup  Setup.local

So adjust the removal logic to match.  Use a wildcard rather than
$GNU_TARGET_NAME as buildroot and python3's idea of the triplet doesn't
always match (E.G.  for musl/uclibc).

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

package/python3: fix configure issue for musl/uclibc GCC 8+ toolchains on powerpc

Fixes:
http://autobuild.buildroot.net/results/cb4/cb49c539501342e45cbe5ade82e588fcdf51f05b

GCC commit 6834b83784dcf0364eb820e8 (multiarch support for non-glibc linux
systems), which is part of GCC 8+, changed the multiarch logic to use
$arch-linux-musl / $arch-linux-uclibc rather than $arch-linux-gnu.

This then causes the python3 configure script to error out:

checking for the platform triplet based on compiler characteristics... powerpc-linux-gnu
configure: error: internal configure error for the platform triplet, please file a bug report

http://autobuild.buildroot.net/results/cb4/cb49c539501342e45cbe5ade82e588fcdf51f05b

As it requires that the --print-multiarch output (if not empty) matches the
deduced triplet (which always uses -linux-gnu).

It isn't quite clear why --print-multiarch returns something for a
non-multiarch toolchain on some architectures (E.G.  PowerPC), but as a
workaround, add a patch to rewrite the --print-multiarch output to match
older GCC versions to keep the configure script happy.

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

package/sshguard: need threads

Fixes:
 - http://autobuild.buildroot.net/results/01fb32fdb21c60ffece09ea5e7ba7b58d989ed76/

Signed-off-by: Pierre-Jean Texier <pjtexier@koncepto.io>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

support/scripts/br2-external: drop help for internal helper script

We do not usually provide help for our internal scripts. Besides, such
help has a tendency to bitrot pretty quickly anyway.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vadim Kochan <vadim4j@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

support/scripts/br2-external: declare missing local variables

Commit b14b02698 (core/br2-external: restore compatibility with old
distros) switched to using 'eval' to emulate associative arrays, for
those distros too old to have bash-4+.

In so doing, it forgot to declare the new local variables in the
respective helper functions.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Vadim Kochan <vadim4j@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/spdlog: new package

Signed-off-by: Asaf Kahlon <asafka7@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

fs/common.mk: enable multithreaded xz compression

xz help indicates only 1 thread is used unless we set threads:
-T, --threads=NUM   use at most NUM threads; the default is 1; set to 0
                    to use as many threads as there are processor cores

Since this splits the file into blocks, the result will be not
bit-for-bit identical to single-threaded compression. Therefore, don't
enable this in BR2_REPRODUCIBLE builds.

Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
Reviewed-by: Matthew Weber <matthew.weber@rockwellcollins.com>
[Arnout: append the option instead of repeating the entire command]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

boot/uboot: add option to pass custom variables to U-Boot build

U-Boot supports a number of environment variables to pass specific
information. The following patches were submitted in the past to one
some specific Config.in option to pass some of these variables:

 - http://patchwork.ozlabs.org/patch/881197/ proposed an option to
   pass a custom EXT_DTB= variable

 - http://patchwork.ozlabs.org/patch/1018245/ proposed an option to
   pass a custom DEVICE_TREE= variable

Instead of adding one Config.in option for each of those variables,
let's provide a generic mechanism to pass arbitrary variables during
U-Boot build step.

Cc: Konstantin Porotchkin <kostap@marvell.com>
Cc: Clemens Gruber <clemens.gruber@pqgruber.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/cppzmq: bump to version 4.4.1

Remove patch (already in version)

Signed-off-by: Pierre-Jean Texier <pjtexier@koncepto.io>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

support/cmake: Explicitly set CMAKE_SYSTEM

Some packages test for CMAKE_SYSTEM explicitly[1]

CMAKE_SYSTEM is comprised of CMAKE_SYSTEM_NAME and CMAKE_SYSTEM_VERSION.
It defaults to CMAKE_SYSTEM_NAME if CMAKE_SYSTEM_VERSION is not set[2]

At the point CMAKE_SYSTEM_NAME is set to "Linux" CMAKE_SYSTEM is already
constructed. Setting it explicitly ensures that it is the correct value.

This is because we do set CMAKE_SYSTEM_NAME twice, in fact:

  - first in toolchainfile.cmake, so that we tell cmake to use the
    "Buildroot" platform,

  - second, in the Buildroot.cmake platform definition itself, so that
    we eventually behave like the Linux platform.

We also set CMAKE_SYSTEM_VERSION to 1, and so the real CMAKE_SYSTEM
value should be set to Linux-1 if we were to follow the documentation to
the letter.

However, for Linux, the version does not matter, and in some situations
may even be harmful (that was reported in one of the commits that
introduce Buildroot.cmake and toolchainfile.cmake).

[1] Fluidsynth https://github.com/FluidSynth/fluidsynth/blob/0cd44d00e1ec9a905d51163efde7b70ff8ae0ba3/CMakeLists.txt#L80
[2] https://cmake.org/cmake/help/git-master/variable/CMAKE_SYSTEM.html#variable:CMAKE_SYSTEM

Signed-off-by: Frank Vanbever <frank.vanbever@mind.be>
Acked-by: Yann E. MORIN <yann.morin.1998@free.fr>
[Peter: update commit message with description from Yann]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/bitcoin: new package

Bitcoin Core is an open source project which maintains and releases
Bitcoin client software called “Bitcoin Core”.

Signed-off-by: Fabio Urquiza <fabiorush@gmail.com>
[Thomas:
 - Don't create a new blockchain applications sub-menu for now, put
   this package in "Miscellaneous applications"
 - Do not select BR2_INSTALL_LIBSTDCPP, use depends on instead, and
   add the corresponding comment.
 - Do not select BR2_TOOLCHAIN_BUILDROOT_USE_SSP. Instead pass
   --disable-hardening, and let Buildroot pass the appropriate CFLAGS
   when hardening features are enabled system-wide.
 - Add missing BR2_TOOLCHAIN_HAS_ATOMIC dependency
 - Add quirky !(BR2_arm || BR2_armeb) || BR2_USE_MMU because the
   Cortex-M toolchains don't provide 8-byte __atomic intrinsics, but
   we don't have a good way to express that today
 - Add missing BR2_TOOLCHAIN_HAS_GCC_BUG_64735 due to the use of
   std::future
 - Use only one BITCOIN_CONF_OPTS assignment to pass all options]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/procps-ng: add init script for sysctl

Add a simple init script that invokes sysctl early in the initialization
process to configure kernel parameters. This is already performed by
systemd (systemd-sysctl) but there is no sysvinit/busybox counterpart.

Files are read from directories in the following list in the given order
from top to bottom:

    /run/sysctl.d/*.conf
    /etc/sysctl.d/*.conf
    /usr/local/lib/sysctl.d/*.conf
    /usr/lib/sysctl.d/*.conf
    /lib/sysctl.d/*.conf
    /etc/sysctl.conf

Signed-off-by: Carlos Santos <unixmania@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/busybox: add init script for sysctl

Add a simple init script that invokes sysctl early in the initialization
process to configure kernel parameters. This is already performed by
systemd (systemd-sysctl) but there is no sysvinit/busybox counterpart.

Files are read from directories in the following list in the given order
from top to bottom:

    /run/sysctl.d/*.conf
    /etc/sysctl.d/*.conf
    /usr/local/lib/sysctl.d/*.conf
    /usr/lib/sysctl.d/*.conf
    /lib/sysctl.d/*.conf
    /etc/sysctl.conf

A file may be used more than once, since there can be multiple symlinks
to it. No attempt is made to prevent this.

Signed-off-by: Carlos Santos <unixmania@gmail.com>
Reviewed-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

boot/optee-os: support alternate image files

Some platform may generate specific boot image files instead of
the generic files tee.bin and tee-*_v2.bin when building OP-TEE OS
package.

This change introduces optee-os configuration directive
BR2_TARGET_OPTEE_OS_CORE_IMAGES that allows board configuration
to specify its expected boot image file names.

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
[Thomas: use the current hardcoded values as the default for the new
config option, to avoid breaking existing setups, and therefore use
$(wildcard ...) to support wildcards]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/busybox/udhcpc.script: fix domain search comment

The domain search option is from RFC3397, not RFC3359 (which is about TLV
codepoints), so fix that.

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

busybox: Enable domain search list support in udhcpc

This is useful in networks with internal resources as it allows
to use much shorter names.

E.g. instead of "server.internal.company.com" it's possible
to use just "server" if DHCP server is configured with:
---------------------------->8-----------------------
option domain-search "internal.company.com";
---------------------------->8-----------------------

This improvement consists of 2 parts:

1. Enable handling of RFC3397 so DHCP client is ready for processing
   corresponding data from DHCP server.

2. Some DHCP servers always send out search list if it is set in server's
   configuration and some servers only provide search list if client
   asks for that (sending list of options it expects to get).

   And exactly for those stubborn DHCP servers we need to add "-O search"
   to udhcp's command line via CONFIG_IFUPDOWN_UDHCPC_CMD_OPTIONS.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Ignacy Gawedzki <ignacy.gawedzki@green-communications.fr>
Cc: Peter Korsgaard <peter@korsgaard.com>
Acked-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

ARC: Add support for ARC HS48 v3.1 processor

This patch introduces a recently released significant update to ARC HS
family: ARC HS48.

One of the major ARC HS48 features is dual-issue pipeline which requires a
little bit modified instruction scheduling compared to single-issue cores
(HS38), thus new "-mcpu/--with-cpu=hs4x".

Also to address some peculiarities of early designs based on HS48 we
introduced yet another "-mcpu/--with-cpu=hs4x_rel31" which we're going to use
as well on some of our development boards.

Signed-off-by: Evgeniy Didin <didin@synopsys.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: arc-buildroot@synopsys.com
[Peter: fixup check-package warnings]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

support/testing: test check-package ignores external.mk

Check external.mk is ignored only when in the root path of a
br2-external.

Add a file called external.mk as a fixture to be used by the test case.

Signed-off-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
[Arnout: wrap at 80 columns]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

check-package: ignore external.mk

The external.mk file in a br2-external usually contains raw makefile
targets. This file is common code and not a package recipe so it should
not be tested against the code-style of a package .mk file.

When using this script to check files in a br2-external tree, usually
the user is responsible for not passing files that check-package do not
understand. But external.mk is special because it is part of the
br2-external structure, so it is likely someone expects it to be
checkable by an in-tree script.

Instead of adding another blob to the manual, just ignore this file.
Only do that when a br2-external is being tested (so with option -b
passed to the script) and also check that it is on the root path of the
br2-external to allow someone to have a package called external.

Reported on bug #11271.

Reported-by: Vitaliy Lotorev <lotorev@gmail.com>
Signed-off-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
[Arnout: wrap at 80 columns]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

support/testing: add test for check-package

Check the basic usage for check-package.

It can be called using either absolute path, relative path or from PATH.
Files to be checked can be passed with either absolute path or relative
path (also including files in the current directory).

Also check it ignores some special files when checking intree files,
i.e. package/pkg-generic.mk, while still generating warnings for out-of-tree
files when called with -b.
In order to allow the later, add an empty line to the Config.in in the
br2-external being tested so the script does generate a warning.

Catches bug #11271.

More tests can be added later, for example compatibility to Python 3.

Suggested-by: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
[Arnout: wrap at 80 columns where appropriate; merge into a single
class.]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

support/testing: allow run-tests to be called from anywhere

Currently run-tests must be called from the Buildroot top directory.

Derive the top directory from the script path, so run-tests can be called from
any path.
As a consequence the test infra will always test the repo it belongs to.

Suggested-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

check-package: fix check of file in current dir with -b

One of the possible usages of check-package is to first cd to the
directory that contains the files to test (e.g. a package directory) and
then call the script passing the files in the current dir.
It already works when used for intree files, but for files in a
br2-external it throws an exception because some check functions (from
utils/checkpackagelib/lib_*.py) do need the name of the file being
processed and assume there will be a slash before the name.

Fix all check functions that assume that the full filename being checked
contains a slash. Do not use regexps to extract the filename, use
os.path functions instead.

Notice RemoveDefaultPackageSourceVariable and TypoInPackageVariable lead
to an exception in this case, but ApplyOrder instead generates a false
warning.

Fixes bug #11271.

Reported-by: Vitaliy Lotorev <lotorev@gmail.com>
Signed-off-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Vitaliy Lotorev <lotorev@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/jack1: new package

This patch adds a new jack1 package alongside the existing jack2.

While jack1 and jack2 are two equivalent implementations of the same
protocol, they differ in a few details and both of them are being
actively used (jack2 is not considered to be a replacement of jack1).

It is not possible to enable both at the same time, so hide away jack1
when jack2 is enabled (to keep existing defconfig files working).

For more information, see:
https://github.com/jackaudio/jackaudio.github.com/wiki/Q_difference_jack1_jack2

Signed-off-by: Adam Heinrich <adam@adamh.cz>
[yann.morin.1998@free.fr:
  - fix coding style
  - use the release tarball, not a git clone
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
[Peter: add host-pkgconf dependency, reorder dependencies]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python-pyasn1: bump to version 0.4.6

Signed-off-by: Pierre-Jean Texier <pjtexier@koncepto.io>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python-pyparsing: bump to version 2.4.2

Also add a hash for LICENSE file

Signed-off-by: Pierre-Jean Texier <pjtexier@koncepto.io>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python-pytz: bump to version 2019.02

Signed-off-by: Pierre-Jean Texier <pjtexier@koncepto.io>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

toolchain/toolchain-external/toolchain-external-custom: be more flexible on gcc version

The custom external toolchain logic asks the user to specify which gcc
version is provided by the toolchain. The list of gcc versions given
by Buildroot is restricted depending on the selected CPU architecture
using the BR2_ARCH_NEEDS_GCC_AT_LEAST_xyz config options.

However, these config options generally indicate in which upstream gcc
version the support for the selected architecture was introduced. But
in practice, it is possible that an external toolchain uses some
non-upstream gcc code, providing support for a CPU architecture before
it was merged in upstream gcc.

A specific example is that there are pre-built external toolchains for
the C-SKY CPU architecture that are based on gcc 6.x, even if the
support for it was only added in upstream gcc 9.x.

Due to the BR2_ARCH_NEEDS_GCC_AT_LEAST_xyz options, only gcc >= 9.x
can be selected for C-SKY, preventing the use of such a custom
toolchain.

In addition, those dependencies are in fact not really needed:
Buildroot will check that the gcc version provided matches what the
user declared in the configuration. And if the gcc provided by the
toolchain does support that CPU architecture, then well, so be it,
there's no need to restrict the gcc version selected.

So we simply get rid of these dependencies on
BR2_ARCH_NEEDS_GCC_AT_LEAST_xyz, and also don't use them anymore to
chose a default value for the gcc version.

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