fs/ubi: expose BINARIES_DIR to ubinize.cfg

It is often necessary to refer to other images that are present in
BINARIES_DIR from a custom ubinize configuration e.g. to include the
kernel in a UBI volume.

As we do for BR2_ROOTFS_UBIFS_PATH, replace BINARIES_DIR when copying
the file.

Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
[yann.morin.1998@free.fr:
  - add the blurb in the help text
  - rewrap commit log
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/libdrm: fix build failure

Add patch to fix build failure with Sourcery Codebench toolchain.

The patch is provided upstream:
https://gitlab.freedesktop.org/mesa/drm/-/merge_requests/75

Fixes:
http://autobuild.buildroot.net/results/b7910460015b077fcb2431da608c34f95c2f2fbb/

Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/redis: add optional support for systemd

- Use the unit type=notify (Redis contacts systemd when ready)
- Start redis with the supervised option, so it knows it should
  talk to systemd
- USE_SYSTEMD is not documented, but it is used in the src/Makefile

Signed-off-by: Titouan Christophe <titouan.christophe@railnova.eu>
[yann.morin.1998@free.fr: add and explain USE_SYSTEMD]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/redis: add optional support for TLS connections

TLS support is only enabled when BUILD_TLS is set to 'yes'. Anoy other
value, as well as empty or not set, will disable TLS. However, for
consistency with the rest of the code (e.g. for autotools packages where
we use both the --enable and --disable flags), we explcitly pass
BUILD_TLS=no. This also avoids an environment variable to leak into the
build.

Signed-off-by: Titouan Christophe <titouan.christophe@railnova.eu>
[yann.morin.1998@free.fr: add and explain the =no case]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

support/testing: add runtime test for Redis

Signed-off-by: Titouan Christophe <titouan.christophe@railnova.eu>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/redis: bump to version 6.0.4

- Update dependencies in Config.in because of new requirements
- Update the help text in Config.in to a more recent wording
- Update the hash file to the new 2 spaces convention

Signed-off-by: Titouan Christophe <titouan.christophe@railnova.eu>
[yann.morin.1998@free.fr:
  - rewrap help text
  - wrap long depends line in comment
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/libhttpserver: bump to version 0.18.0

- Remove patch (already in version)
- Update indentation in hash file (two spaces)

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Reviewed-by: Stephan Hoffmann <sho@relinux.de>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/libhttpserver: disable examples

examples are enabled by default since version 0.17.5 and
https://github.com/etr/libhttpserver/commit/012d014a7c52ef91a35c521d5d4f83ee11de5068

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Reviewed-by: Stephan Hoffmann <sho@relinux.de>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

linux: introduce BR2_KERNEL_DTB_KEEP_DIRNAME

In linux, newer architectures like ARM64 and RISC-V keep their device
tree sources organized by subdirectories. When these device trees are
installed by the kernel they will keep the directory names. But
buildroot strips the prefixes when installing them into /boot or the
images directory.
Sometimes the bootloader references the device tree by name (e.g.
u-boot has the environment variable 'fdtfile') which also includes
the prefix directory.
Make it possible to keep this prefix during installation, so we can
be compatible with other distributions and make it easier for the
user, because he doesn't have to change that environment variable.

Signed-off-by: Michael Walle <michael@walle.cc>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/gloox: fix build on musl

Fixes:
 - http://autobuild.buildroot.org/results/a61a91bb6e1156eff5fb756f9dccfc2bb6c0f277

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/python-markdown2: drop patches

On master, commit 544007dcc4 itroduced patches to fix CVE-2020-11888.
On next, commit 604fe08806 itroduced the exact same patches for the
exact same reason.

But on next, commit 81b3fd8654 bumped the version and dropped the
patches.

When next was merged into master in commit a6569f2b3d, the patches
introduced by 544007dcc4 (on master) were retained.

Fixes:
 - http://autobuild.buildroot.org/results/bf305c78dddd035b97e88943a1d19a8ceb6b41f7

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[yann.morin.1998@free.fr: rewrite commit log with detailed explanations]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

{linux, linux-headers}: add version 5.7

Signed-off-by: Michael Walle <michael@walle.cc>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/openssh: bump to version 8.3p1

Format hash file with two spaces separators.

Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/msmtp: bump to version 1.8.11

Format hash file with two spaces delimiters.

Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/perl-timedate: bump to version 2.33

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/perl-path-tiny: bump to version 0.114

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/perl-mozilla-ca: bump to version 20200520

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/perl-net-dns: bump to versionn 1.24

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/perl-json-maybexs: bump to version 1.004002

now, LICENSE has its own file

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/perl-http-message: bump to version 6.24

LICENSE diff:
- This software is copyright (c) 1994-2017 by Gisle Aas.
+ This software is copyright (c) 1994 by Gisle Aas.

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/perl-exporter-tiny: bump to version 1.002002

LICENSE diff:
- This software is copyright (c) 2018 by Toby Inkster.
+ This software is copyright (c) 2020 by Toby Inkster.

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/perl-date-manip: bump to version 6.82

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/perl-http-daemon: bump to version 6.12

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/ell: needs headers >= 4.12

ell needs headers >= 4.12 since version 0.31 and
https://git.kernel.org/pub/scm/libs/ell/ell.git/commit?id=fa950995a48bbcd8f0d3dfbbe9a7f768aba5c50f

Indeed NLM_F_CAPPED, NLM_F_ACK_TLVS, NLMSGERR_ATTR_MSG and
NETLINK_EXT_ACK are only available since
https://github.com/torvalds/linux/commit/2d4bc93368f5a0ddb57c8c885cdad9c9b7a10ed5

So add this dependency to ell and its reverse dependencies

Fixes:
 - http://autobuild.buildroot.org/results/bafc0ee2a3bcb2418a3612da38eb9dcd586b57d4

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

boot/grub2: Fix GRUB i386-pc build with Ubuntu gcc

Backport a patch from upstream to fix the build on certain versions of
gsc, notably:

    Ubuntu 19.10 with gcc (Ubuntu 8.3.0-26ubuntu1~19.10) 8.3.0
    Ubuntu 19.10 with gcc (Ubuntu 9.2.1-9ubuntu2) 9.2.1 20191008

The upstream patch is simply a change in the gentpl.py script, which is
used to generate parts of the automake machinery, so if we just backport
the upstream patch, we need to call the script to regenerate those files.

However, the modified script is a python script, so we would need to add
a dependency on host-python (2 or 3), which is not so nice.

Furthermore, calling the script is not enough: it needs a specific set
of optionss for each file it is to generate. That set of options is not
static; it is constructed in the convoluted autogen.sh. Calling
autogen.sh is usally not so good an idea in the Buildroot context, and
indeed this fails becasue it calls to autoreconf, but without our
carefuly crafted options and environment variables.

There was a little light in the tunnel, in that autogen.sh can be told
not to run autoreconf, by setting the environemnt variable
FROM_BOOTSTRAP to an non-=empty string, but this is fraught with various
other side-effects, as in that cause, autogen.sh expects to be valled by
an upper sciopt, bootstrap, which is not provided in the tarball
distribution...

So, between all those issues, autogen, bootstrap, and a host-python (2
or 3) dependency, we choose another route: path the script *and* the one
generated file affected by the change. Since that patched file is a .am
file, we also patch the corresponding .in file

However, we're faced with another issue: the other generated file is
now older than the script, so the automake machinery will now want to
re-run autoconf et al during the build step, which is still not a good
idea for us. So we touch the other generated file so it is mopre recent
than the script.

This is still not sufficient, because the patched file also has a
dependency on the generated file, so we need to touch as well.

Fixes:
 - https://bugs.buildroot.org/show_bug.cgi?id=12946

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[yann.morin.1998@free.fr:
  - keep the hunk about patching gentpl.py
  - make it a git-formatted patch
  - add the touch
  - drastically expand the commit log
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/gloox: disable tests and examples

Fixes:
http://autobuild.buildroot.net/results/91385a34c80c5043519e96395e203eef662ac18d/

Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

Revert "package/prosody: use correct bit32 package"

This reverts commit fa84c176c2148a60103e850204180f86aa5baa73 that
replace luabitop by lua_bit32 package when lua 5.1 is used.

Since this change the prosody test in gitlab is fail due to
missing lua-bitops [1]:

Starting prosody:
**************************
Prosody was unable to find lua-bitops
This package can be obtained in the following ways:

    Source:           http://bitop.luajit.org/
    Debian/Ubuntu:    sudo apt-get install lua-bitop
    luarocks:         luarocks install luabitop

WebSocket support will not be available
More help can be found on our website, at https://prosody.im/doc/depends
**************************

The upstream documentation [2] is misleading (or not uptodate)
about lua-bit32 dependency.

Since bitop is builtin since lua5.2, we probably need to select
luabitop package only when lua 5.1 is used as lua interpreter.

Tested with run-tests:
./support/testing/run-tests tests.package.test_prosody.TestProsodyLua51

[1] https://gitlab.com/buildroot.org/buildroot/-/jobs/576271975
[2] https://prosody.im/doc/depends#bitop

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: James Hilliard <james.hilliard1@gmail.com>
Cc: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/axel: bump version to 2.17.9

Signed-off-by: Ismael Luceno <ismael@iodev.co.uk>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/valgrind: bump version to 3.16.0

- removed 0003-mips-fix-mips32r6-and-mips64r6-compilation-issue.patch
  (backport from upstream commit [1])

[1] https://sourceware.org/git/?p=valgrind.git;a=commit;h=50dd9600ab1ad8933704ee85289191d9c65ba444

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

package/libglib2: bump version to 2.64.3

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

package/fluidsynth: bump to version 2.1.3

Changes since v2.1.2:
https://github.com/FluidSynth/fluidsynth/releases/tag/v2.1.3
- fix a cross-compilation failure from Win32 to WinARM
- fix issues while fluid_player is seeking
- fix a NULL pointer dereference if synth.dynamic-sample-loading is enabled
- fix a NULL pointer dereference in delete_rvoice_mixer_threads()
- fix a NULL pointer dereference in the soundfont loader
- fix dsound driver playing garbage when terminating fluidsynth
- avoid memory leaks when using libinstpatch

./utils/test-pkg --package fluidsynth
6 builds, 1 skipped, 0 build failed, 0 legal-info failed

Signed-off-by: Julien Olivain <juju@cotds.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/doxygen: bump version to 1.18.18

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

package/mosquitto: bump to v1.6.10

This release fix some bugs in the broker and client libraries,
as well as building with below C99 suport.

Read the whole announcement on:
https://mosquitto.org/blog/2020/05/version-1-6-8-released/

Signed-off-by: Titouan Christophe <titouan.christophe@railnova.eu>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/libqmi: bump to 1.24.14

* libqmi-glib:
   ** Fix build with GLib < 2.44.
   ** Fix UTF-8 string validation when the string contains a trailing NUL byte
      (e.g. the Sierra specific "DMS Swi Get Current Firmware" command).

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/ethtool: bump to 5.6

- Now by default ethtool depends on libmnl so make this optional

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/edid-decode: bump version to 1889504

- edid-decode changed from C to C++, adjust make command
  accordingly and add BR2_INSTALL_LIBSTDCPP dependency

- now shipping a LICENSE file, change from (gone) edid-decode.c to this one

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

package/perl: security bump to version 5.30.3

Fixes the following security issues:

[CVE-2020-10543] Buffer overflow caused by a crafted regular
                 expression

[CVE-2020-10878] Integer overflow via malformed bytecode produced by a
                 crafted regular expression

[CVE-2020-12723] Buffer overflow caused by a crafted regular
                 expression

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

package/iwd: bump version to 1.7

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

package/ell: bump version to 0.31

- removed 0001-ell-ecc.h-fix-build-with-uclibc.patch
  (upstream applied [1])

- update 0002-ell-rtnl-fix-compile-with-older-toolchains.patch
 (add upstream link)

[1] https://git.kernel.org/pub/scm/libs/ell/ell.git/commit/?id=aa30f1884cf72d590fdca1db11d4fb2c2d526141

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

package/lcms2: bump to version 2.10

- Drop patch (already in version)
- Update indentation in hash file (two spaces)
- Update hash of COPYING (update in year:
  https://github.com/mm2/Little-CMS/commit/4ba97f4b686c9988b7632638f915ead64fb6cd69)

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

package/sysrepo: bump version to 1.4.66

Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/libyang: bump to version 1.0.176

Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/libnetconf2: bump to version 1.1.26

Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/strace: bump to version 5.7

Drop upstream patch.

Fixes:
http://autobuild.buildroot.net/results/ca298d6543c07efbf77f2adeb4832bbac00ae73f/
http://autobuild.buildroot.net/results/f26a4f2bb5a9b25739e55be5e5ded2b83a0937ac/
http://autobuild.buildroot.net/results/53d6dac4047742ae2acd682a0dd97d986ba611bb/

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

package/libresslL: bump version to 3.1.2

Signed-off-by: Adam Duskett <Aduskett@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/rt-tests: bump to version 1.8

Drop patch #1. libnuma is now a mandatory dependency. The ostype and
machinetype variables are not used for the Linux target.

Drop patch #3. SH is not a supported target architecture anymore.

Add mandatory dependency on libnuma (numactl). Propagate numactl
architecture dependency.

Update homepage link. The kernel wiki is dead.

Drop Python build time dependency. Python is an optional run-time
dependency.

Add reference to upstream provided tarball hash.

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

package/numactl: add BR2_PACKAGE_NUMACTL_ARCH_SUPPORTS

For use in other packages that select numactl.

Cc: Pierre-Jean Texier <pjtexier@koncepto.io>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/python-django: security bump to version 3.0.7

Fixes the following security issues:

- CVE-2020-13254: Potential data leakage via malformed memcached keys

  In cases where a memcached backend does not perform key validation,
  passing malformed cache keys could result in a key collision, and
  potential data leakage.  In order to avoid this vulnerability, key
  validation is added to the memcached cache backends.

- CVE-2020-13596: Possible XSS via admin ForeignKeyRawIdWidget

  Query parameters for the admin ForeignKeyRawIdWidget were not properly URL
  encoded, posing an XSS attack vector.  ForeignKeyRawIdWidget now ensures
  query parameters are correctly URL encoded.

For details, see the announcement:
https://docs.djangoproject.com/en/dev/releases/3.0.7/

Additionally, 3.0.5..3.0.7 contains a number of non-security related
bugfixes.

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

package/pciutils: bump to version 3.7.0

Use CROSS_COMPILE for toolchain prefix. Makefile derives CC, AR, and
RANLIB from that.

Remove sed manipulation of the lib/configure script. Host uname is not
used when the HOST variable is set as we do.

Remove sed manipulation of Makefile. Set STRIP to an empty string
instead.

Format hash file with two space separators.

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

boot/uboot: change from FTP to HTTPS for official releases

The HTTPS URL seems to be more reliable and quicker for download than
FTP.  FTP may also be a blocked protocol on some sites and in CI/CD
setups.

Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

docs/website: update for 2020.02.3

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

Update for 2020.02.3

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit d42f3adaae24a6aa3abc2de4f39fa8023f971d31)
[Peter: drop Makefile changes]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/bcm2835: bump version to 1.65

Changes since 1.63:
 - 1.64 2020-04-11 Fixed error in definitions of BCM2835_AUX_SPI_STAT_TX_LVL
   and BCM2835_AUX_SPI_STAT_RX_LVL
 - 1.65, 1.66 2020-04-16 Added support for use of capability cap_sys_rawio
   to determine if access to /dev/mem is available for non-root users

That latter part (using capabilities) is not supported, because it is
broken upstream (the code is messed up using two similar #defines to
test and enable it; messy...) Since it previously required root access
to work, and still does now, this is not a regression, so do not add
support for capablities.

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
[yann.morin.1998@free.fr: explain why we don't support capabilities]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/bcm2835: disable doxygen doc generation

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/nodejs: security bump to 12.18.0

This is a security release.

Vulnerabilities fixed:

CVE-2020-8172: TLS session reuse can lead to host certificate verification bypass (High).
CVE-2020-11080: HTTP/2 Large Settings Frame DoS (Low).
CVE-2020-8174: napi_get_value_string_*() allows various kinds of memory corruption (High).

See https://nodejs.org/en/blog/release/v12.18.0/

Signed-off-by: Martin Bark <martin@barkynet.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/nghttp2: security bump version to 1.41.0

Fix CVE-2020-11080 Denial of service: Overly large SETTINGS frames

Signed-off-by: Martin Bark <martin@barkynet.com>
[yann.morin.1998@free.fr: two spaces in hash files]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/qemu: remove uneeded upstream patches

Although those patches were properly dropped when the origianl bump was
applied to the next branch (commit 4675c7d441), both net and master also
had a commit that moved the patches around when the csku fork was
removed (commit 58af9a70cc and 20f45029cc, respectively).

This seemed to have caused some confusion with git-merge, though, and
the y re-appeared after the merge.

Remove them again for good, this time.

Fixes: http://autobuild.buildroot.net/results/0adfb031c243709b0bac71599ed419b64cc514a4
Signed-off-by: Adam Duskett <Aduskett@gmail.com>
[yann.morin.1998@free.fr:
  - rewrite commit log to explain why the patches reappeared
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/erofs-utils: bump version to 1.1

- removed 0001-erofs-utils-fix-configure.ac.patch [1];
- removed 0002-erofs-utils-avoid-_LARGEFILE64_SOURCE-and-_GNU_SOURC.patch [2];
- removed 0003-erofs-utils-avoid-using-old-compatibility-type-uint.patch [3];
- removed 0004-erofs-utils-avoid-PAGE_SIZE-redefinition.patch [4];
- add host-pkgconf, util-linux dependencies for uuid support.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs-utils.git/patch/?id=eefd95b37e1042992cb07bec1ac3f6dbe199d8f0
[2] https://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs-utils.git/patch/?id=d4a161552becafeb1ebb98ec7e28675cb25fc548
[3] https://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs-utils.git/patch/?id=989947348dddf03a8292b5e32bca538f0a325cd9
[4] https://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs-utils.git/patch/?id=bdbabe54112d04c05819ebebf4e6f88ae863d436

Signed-off-by: Gao Xiang <hsiangkao@aol.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/systemd: bump version to 245.6

Signed-off-by: Adam Duskett <Aduskett@gmail.com>
[yann.morin.1998@free.fr: two sapces in hash file]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/libv4l: bump version to 1.20.0

- removed 0002-keytable-use-input_event-properly.patch
  (upstream [1])

- removed 0003-keytable-add-compatibility-for-input_event_sec.patch
  (upstream [2])

[1] https://git.linuxtv.org/v4l-utils.git/patch/?id=38f4ce74275ae4625463f7eec78764715a0b6246
[2] https://git.linuxtv.org/v4l-utils.git/patch/?id=8b7e6ce9367fe09ca9398b5f3cc75bba2598b162

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

package/iptables: bump version to 1.8.4

For changelog details see [1].

[1] https://netfilter.org/projects/iptables/files/changes-iptables-1.8.4.txt

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

package/hwdata: bump version to 0.335

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

package/fetchmail: bump version to 6.4.6

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

package/liburiparser: bump to version 0.9.4

- Drop patch (already in version)
- Update hash of COPYING (BSD-3 license fixed with
  https://github.com/uriparser/uriparser/commit/a3e81383591de85c65918a42e68fbe736832f7c1)

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

package/c-periphery: bump version to 2.2.1

Since commit "cmake: add cmake build support"
(https://github.com/vsergeev/c-periphery/commit/952e1e906a5d65b78932128af24b7dbb8cce2e9dvsergeev/c-periphery@d0a973c),
c-periphery implement cmake build, so use cmake-package instead of
generic-package. Due to this, it now builds a shared library, so we
drop the INSTALL_TARGET = NO.

The hash of the license file is updated due to an update in the
copyright year:

  - Copyright (c) 2014-2019 vsergeev / Ivan (Vanya) A. Sergeev
  + Copyright (c) 2014-2020 vsergeev / Ivan (Vanya) A. Sergeev

Signed-off-by: Joris Offouga <offougajoris@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/python-periphery: bump to version 2.1.0

See changelog https://github.com/vsergeev/python-periphery/blob/master/CHANGELOG.md

Update the license hash for a change in copyright years:
-Copyright (c) 2015-2019 vsergeev / Ivan (Vanya) A. Sergeev
+Copyright (c) 2015-2020 vsergeev / Ivan (Vanya) A. Sergeev

Also switch to the new 2 spaces convention for the hash file

Signed-off-by: Pierre-Jean Texier <pjtexier@koncepto.io>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

configs/orangepi_zero: bump kernel and uboot versions

Bump kernel to version 5.6 and uboot to version 2020.04 for
orangepi-zero configuration

Signed-off-by: Gonçalo Salazar <glbsalazar@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/checksec: bump to 2.2.2

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/protobuf: bump to 3.12.2

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

Merge branch 'next'

A few conflicts had to be resolved:

 - Version number and hash for mesa3d-headers/mesa3d
 - Patches added in qemu, and the qemu version number
 - The gnuconfig README.buildroot

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

Kickoff 2020.08 cycle

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

docs/website/news.html: add 2020.05 announcement link

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

Update for 2020.05

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

{linux, linux-headers}: bump 4.{4, 9, 14, 19}.x / 5.{4, 6}.x series

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

package/wampcc fix build with musl 1.2.0

Fixes:
 - http://autobuild.buildroot.org/results/da996e189220499b85efbdb541a891ac18db38c6

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

DEVELOPERS: cleanup rockwellcollins.com maintainers

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/glib-networking: security bump to version 2.62.4

- Fix CVE-2020-13645: In GNOME glib-networking through 2.64.2, the
  implementation of GTlsClientConnection skips hostname verification of
  the server's TLS certificate if the application fails to specify the
  expected server identity. This is in contrast to its intended
  documented behavior, to fail the certificate verification.
  Applications that fail to provide the server identity, including Balsa
  before 2.5.11 and 2.6.x before 2.6.1, accept a TLS certificate if the
  certificate is valid for any host.
- Update indentation in hash file (two spaces)

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[Peter: bump to 2.62.4 rather than 2.64.3]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/libusb-compat: set LIBUSB_1_0_SONAME

LIBUSB_1_0_SONAME is detected since version 0.1.6 and
https://github.com/libusb/libusb-compat-0.1/commit/b6f5a2fe12ca19d658d7180e106254b31cf1f8f5

The detection mechanism is based on sed, here are the more relevant
parts:

 shrext_regexp=`echo "$shrext_cmds" | sed 's/\./\\\\./'`
[...]
 [AS_VAR_SET([ac_Lib_SONAME], [`ldd conftest$ac_exeext | grep 'lib[$2]'$shrext_regexp | sed 's/^@<:@ \t@:>@*lib[$2]'$shrext_regexp'/lib[$2]'$shrext_regexp'/;s/@<:@ \t@:>@.*$//'`])])

However, this mechanism is broken with sed 4.7 and will return the
following 'silent' error:

checking for SONAME of libusb-1.0... sed: -e expression #1, char 40: Invalid back reference
unknown

Moreover, it also raises the following build failure on one of the
autobuilder because an empty line is added to LIBUSB_1_0_SONAME:

checking for SONAME of libusb-1.0... checking
libusb-1.0.so.0
checking for GNU extensions of errno.h... no
configure: WARNING: cache variable au_cv_lib_soname_LIBUSB_1_0 contains a newline
checking that generated files are newer than configure... done
configure: creating ./config.status
config.status: creating libusb.pc
config.status: creating libusb-config
config.status: creating Makefile
config.status: creating libusb/Makefile
config.status: creating examples/Makefile
config.status: creating config.h
config.status: executing depfiles commands
config.status: executing libtool commands
config.status: executing default commands
configure: WARNING: unrecognized options: --disable-gtk-doc, --disable-gtk-doc-html, --disable-doc, --disable-docs, --disable-documentation, --with-xmlto, --with-fop, --enable-ipv6, --disable-nls
configure: WARNING: cache variable au_cv_lib_soname_LIBUSB_1_0 contains a newline
[7m>>> libusb-compat 0.1.7 Building[27m
PATH="/usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output-1/host/bin:/usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output-1/host/sbin:/accts/mlweber1/bin:/accts/mlweber1/libexec/git-core:/accts/mlweber1/usr/bin:/accts/mlweber1/usr/local/bin:/accts/mlweber1/bin:/accts/mlweber1/libexec/git-core:/accts/mlweber1/usr/bin:/accts/mlweber1/usr/local/bin:/accts/mlweber1/bin:/accts/mlweber1/libexec/git-core:/accts/mlweber1/usr/bin:/accts/mlweber1/usr/local/bin:/accts/mlweber1/bin:/accts/mlweber1/libexec/git-core:/accts/mlweber1/usr/bin:/accts/mlweber1/usr/local/bin:/accts/mlweber1/bin:/accts/mlweber1/libexec/git-core:/accts/mlweber1/usr/bin:/accts/mlweber1/usr/local/bin:/accts/mlweber1/bin:/accts/mlweber1/libexec/git-core:/accts/mlweber1/usr/bin:/accts/mlweber1/usr/local/bin:/accts/mlweber1/bin:/accts/mlweber1/libexec/git-core:/accts/mlweber1/usr/bin:/accts/mlweber1/usr/local/bin:/accts/mlweber1/bin:/accts/mlweber1/libexec/git-core:/accts/mlweber1/usr/bin:/accts/mlweber1
 /usr/local/bin:/usr/lib64/qt-3.3/bin:/usr/local/bin:/usr/bin:/usr/local/sbin:/usr/sbin"  /usr/bin/make -j8  -C /usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output-1/build/libusb-compat-0.1.7/
make[1]: Entering directory `/usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output-1/build/libusb-compat-0.1.7'
Makefile:284: *** missing separator.  Stop.

We could patch patch m4/au_check_lib_soname.m4 to fix the mechanism
however this is difficult without reproducing the autobuilder failure
and upstream seems dead so just set LIBUSB_1_0_SONAME

Fixes:
 - http://autobuild.buildroot.org/results/12d771d85d30594929cfe3e1c783fc70857e7f5f

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[yann.morin.1998@free.fr: extract the actual SONAME from the library]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/qemu: bump to 5.0.0

Remove upstream patches [1][2].

Switch to libcap-ng for virtio support [3].

Remove bluez option [4].

Disable container build [5] since we don't want to use containers for cross-building.

Disable io_uring [6] since there is no such package in Buildroot (yet).

The ARM Cortex-m7 cpu is now supported [7] a defconfig can be added in follup patch.

[1] https://git.qemu.org/?p=qemu.git;a=commit;h=00b5032eaddb7193f03f0a28b10286244d2e2a7b
[2] https://git.qemu.org/?p=qemu.git;a=commit;h=21bf9b06cb6d07c6cc437dfd47b47b28c2bb79db
[3] https://git.qemu.org/?p=qemu.git;a=commit;h=7e46261368d129c5ee8be927f5bcadc7ecd800d7
[4] https://git.qemu.org/?p=qemu.git;a=commit;h=1d4ffe8dc77cbc9aafe8bcf514ca0e43f85aaae3
[5] https://git.qemu.org/?p=qemu.git;a=commit;h=afc3a8f9f1df09c091f9903eaef82b35c152cacf
[6] https://git.qemu.org/?p=qemu.git;a=commit;h=c10dd8565defdb14695580c9369b20f4544e65a2
[7] https://git.qemu.org/?p=qemu.git;a=commit;h=cf7beda5072e106ddce875c1996446540c5fe239

See:
https://wiki.qemu.org/ChangeLog/5.0
https://www.qemu.org/2020/04/29/qemu-5-0-0/

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

package/qemu: remove csky fork

We have a qemu fork for csky cpus [1] but since qemu version
bump to 4.2.0 [2] and libssh2/libssh change the csky build is
broken.

The csky fork is based on Qemu 3.0.0 but unlike autotools packages
any unknown option is handled as error.

Since we don't want to support all options from previous qemu
release and the github repository has been removed [3] and the
only remaining archive is located on http://sources.buildroot.net,
remove the qemu csky fork as suggested by [4].

[1] https://git.buildroot.net/buildroot/commit/?id=f816e5b276f1ef15840bec6667f1e8219717ab7d
[2] https://git.buildroot.net/buildroot/commit/?id=0ea17054ce7dfc54efca5634133cef786445e7b1
[3] https://github.com/c-sky/qemu
[4] http://lists.busybox.net/pipermail/buildroot/2020-May/281885.html

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Guo Ren <ren_guo@c-sky.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
[Peter: move patches out of 4.2.0 subdir]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/libcap-ng: add host variant

Qemu 5.0.0 recently switched to libcap-ng [1].
Add the host variant for host-qemu package.

[1] https://git.qemu.org/?p=qemu.git;a=commit;h=7e46261368d129c5ee8be927f5bcadc7ecd800d7

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

package/linux-tools/perf: bear the kernel options munging

perf by itself is not a standalone package; instead, it is part of a
bigger package, linux-tools.

Even though perf is the only one to need kernel .config fixups, we still
do it in a generic way, as it blends nicely in the existing variables,
which all use a loop over all the tools.

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

package/linux-headers: add support for CIP kernel versions with same-as-kernel

When the linux-headers are configured to use the same source as the
kernel (BR2_KERNEL_HEADERS_AS_KERNEL), and the kernel is configured
to be one of the two CIP versions (BR2_LINUX_KERNEL_LATEST_CIP_VERSION
or BR2_LINUX_KERNEL_LATEST_CIP_RT_VERSION), the build fails if the
kernel sources are not already downloaded:

    $ cat defconfig
    BR2_LINUX_KERNEL=y
    BR2_LINUX_KERNEL_LATEST_CIP_VERSION=y
    $ make defconfig BR2_DEFCONFIG=$pwd)/defconfig
    $ make linux-headers-source

    >>> linux-headers 4.19.118-cip25 Downloading
    --2020-05-13 19:28:44--  https://cdn.kernel.org/pub/linux/kernel/v4.x/linux-4.19.118-cip25.tar.xz
    Resolving cdn.kernel.org (cdn.kernel.org)... 2a04:4e42:1d::432, 151.101.121.176
    Connecting to cdn.kernel.org (cdn.kernel.org)|2a04:4e42:1d::432|:443... connected.
    HTTP request sent, awaiting response... 404 Not Found
    2020-05-13 19:28:45 ERROR 404: Not Found.

    make[1]: *** [package/pkg-generic.mk:171: /home/ymorin/dev/buildroot/O/build/linux-headers-4.19.118-cip25/.stamp_downloaded] Error 1
    make: *** [Makefile:23: _all] Error 2

We fix that by adding yet another duplication of information out of
the linux.mk, to use the CIP-specific git tree where to get the
archives as snapshots.

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

linux: forcibly disable use of gcc plugins

The soon-to-be-released linux 5.7 has changed the way it detects the
ability of gcc to use plugins, when it dropped support for gcc 4.7 or
older [0].

To detect the ability to use gcc plugins, the kernel has to check
whether the host gcc is capable enough to build them.

When we call one of the configurator for the Linux kernel, we explicitly
pass a value of HOSTCC=$(HOSTCC_NOCCACHE), because there might be a
discrepancy between the ncurses headers and libraries as found by the
Linux kconfig build [1] [2].

But then, when we build the kernel, we pass another value to use [3]
HOSTCC="$(HOSTCC) $(HOST_CFLAGS) $(HOST_LDFLAGS)" which boils down to
roughly: gcc -I.../host/include -L.../host/lib -Wl,-rpath,.../host/lib
This is needed so that at build time, the kernel can build host tools
that link with our openssl et al.

So, the two HOSTCC we pass to the kernel may have different behaviours.
For example, on a machine where gmp is missing in the system, it is
available in $(O)/host/ when using an internal toolchain (and under a
few other conditions).

In that case, when configuring the kernel, it decides that the host
compiler can't build plugins, so the dependencies of CONFIG_GCC_PLUGINS
are not met, and that option is not present in the linux' .config file
(neither as "=y" nor as "is not set"). But then, when we build the
kernel, the host compiler suddenly becomes capable of building the
plugins, and the internal syncconfig run by the kernel will notice that
the dependencies of CONFIG_GCC_PLUGINS are now met, and that the user
shall decide on its value. And this blocks a build on an interactive
console (abbreviated):

    * Restart config...
    * GCC plugins
    GCC plugins (GCC_PLUGINS) [Y/n/?] (NEW) _

But most problematic is the behaviour when run in a shell that is not
interactiove (e.g. a CI job or such) (abbreviated):

    * Restart config...
    * GCC plugins
    GCC plugins (GCC_PLUGINS) [Y/n/?] (NEW)
    Error in reading or end of file.
      Generate some entropy during boot and runtime (GCC_PLUGIN_LATENT_ENTROPY) [N/y/?] (NEW)
    Error in reading or end of file.
      Randomize layout of sensitive kernel structures (GCC_PLUGIN_RANDSTRUCT) [N/y/?] (NEW)
    Error in reading or end of file.
    * Memory initialization
    Initialize kernel stack variables at function entry
    > 1. no automatic initialization (weakest) (INIT_STACK_NONE)
      2. zero-init structs marked for userspace (weak) (GCC_PLUGIN_STRUCTLEAK_USER) (NEW)
      3. zero-init structs passed by reference (strong) (GCC_PLUGIN_STRUCTLEAK_BYREF) (NEW)
      4. zero-init anything passed by reference (very strong) (GCC_PLUGIN_STRUCTLEAK_BYREF_ALL) (NEW)
    choice[1-4?]:
    Error in reading or end of file.
    Poison kernel stack before returning from syscalls (GCC_PLUGIN_STACKLEAK) [N/y/?] (NEW)
    Error in reading or end of file.
    Enable heap memory zeroing on allocation by default (INIT_ON_ALLOC_DEFAULT_ON) [N/y/?] n
    Enable heap memory zeroing on free by default (INIT_ON_FREE_DEFAULT_ON) [N/y/?] n

The most obvious and simple solution would be to unconditionally disable
gcc plugins altogether, in the KCONFIG_FIXUP hook. But that can't work
either, because after applying the fixups, we call olddefconfig (or the
likes) with the incapable HOSTCC, so the disabled option would be removed
anyway, and we'd be back to square one.

So, in addition to the above, we also forcibly hack the same call just
before actually building the kernel.

Note that the two are needed: the one in the fixups is needed for those
that have a system that already allows building gcc plugins, and the
second is needed in the other case, where the system does not allow it
but would work with our additional headers and libs in $(O)/host/. The
two ensure there is a very similar experience in the two situations.

Forcibly disabling the use of gcc plugins is not a regression on our
side: it has never been possible to do so so far. We're now making sure
that can't work by accident.

Reported-by: Ganesh <ganesh45in@gmail.com>,
Reported-by: Heiko Thiery <heiko.thiery@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Michael Walle <michael.walle@kontron.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Tested-by: Heiko Thiery <heiko.thiery@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/qt5/qt5webengine: don't link with libstdc++.a on the host

While cross-compiling, qt5webengine is building a host tool, 'gn', and
by default wants to link it statically with libstdc++, when the tool is
otherwise dynamically linked with other libraries:

    $ ldd 3rdparty/gn/out/Release/gn
        linux-vdso.so.1 (0x00007ffc1c999000)
        libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f48a3c06000)
        libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f48a3be4000)
        libc.so.6 => /lib64/libc.so.6 (0x00007f48a3a1b000)
        /lib64/ld-linux-x86-64.so.2 (0x00007f48a3c53000)

Not all ditributions have the static libraries installed by default; for
example, on Fedora, libstdc++-static is not installed on a fresh system,
leading to build issues:

  [185/185] LINK gn
  FAILED: gn
  /usr/bin/g++ -O3 -fdata-sections -ffunction-sections -Wl,--gc-sections -Wl,-strip-all -Wl,--as-needed -static-libstdc++ -pthread -o gn -Wl,--start-group tools/gn/gn_main.o base.a gn_lib.a -Wl,--end-group -ldl
  /usr/bin/ld : unable to find -lstdc++
  [...]
  Project ERROR: GN build error!

The root cause is the addition in [0] of a command line option to the
build of gn, that requests static linking with libstdc++ by default.

Explicitly pass that option now, to avoid static linking with libstdc++
and get a fully dynamicallty linked executable:

    $ ldd 3rdparty/gn/out/Release/gn
        linux-vdso.so.1 (0x00007ffd3f160000)
        libstdc++.so.6 => /lib64/libstdc++.so.6 (0x00007f68138e7000)
        libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f68138c5000)
        libc.so.6 => /lib64/libc.so.6 (0x00007f68136fc000)
        libm.so.6 => /lib64/libm.so.6 (0x00007f68135b6000)
        /lib64/ld-linux-x86-64.so.2 (0x00007f6813b13000)
        libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f681359c000)

[0] https://github.com/qt/qtwebengine-chromium/commit/cfab9198a9917f42cf08b1caf84ab9b71aac1911#diff-905c8f054808213577c0a92d1b704615

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Gaël Portay <gael.portay@collabora.com>
[yann.morin.1998@free.fr:
  - rewrite the commit log with extra details and explanations
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/gcc/9.3.0: fix host-gcc-final when ccache is used

As reported by several Buildroot users [1][2][3], the gcc build
may fail while running selftests makefile target.

The problem only occurs when ccache is used with gcc 9 and 10,
probably due to a race condition.

While debuging with "make -p" we can notice that s-selftest-c target
contain only "cc1" as dependency instead of cc1 and SELFTEST_DEPS [4].

  s-selftest-c: cc1

While the build is failing, the s-selftest-c dependencies recipe is
still running and reported as a bug by make.

  "Dependencies recipe running (THIS IS A BUG)."

A change [5] in gcc 9 seems to introduce the problem since we can't
reproduce this problem with gcc 8.

As suggested by Yann E. MORIN [6], move SELFTEST_DEPS before
including language makefile fragments.

With the fix applied, the s-seltest-c dependency contains
SELFTEST_DEPS value.

  s-selftest-c: cc1 xgcc specs stmp-int-hdrs ../../gcc/testsuite/selftests

[1] http://lists.busybox.net/pipermail/buildroot/2020-May/282171.html
[2] http://lists.busybox.net/pipermail/buildroot/2020-May/282766.html
[3] https://github.com/cirosantilli/linux-kernel-module-cheat/issues/108
[4] https://gcc.gnu.org/git/?p=gcc.git;a=blob;f=gcc/c/Make-lang.in;h=bfae6fd2549c4f728816cd355fa9739dcc08fcde;hb=033eb5671769a4c681a44aad08a454e667e08502#l120
[5] https://gcc.gnu.org/git/?p=gcc.git;a=commitdiff;h=033eb5671769a4c681a44aad08a454e667e08502
[6] http://lists.busybox.net/pipermail/buildroot/2020-May/283213.html

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Ben Dakin-Norris <ben.dakin-norris@navtechradar.com>
Cc: Maxim Kochetkov <fido_max@inbox.ru>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/sysrepo: fix SysV init script

The current script (S51sysrepo-plugind) is not able to stop the daemon.

Possible options to fix the problem:

A) By adding the "-m -p $PIDFILE" option to start the pid file will be
   created but it will not contain the correct PID used by the daemon.
   This is obviously because the daemon forks.
B) By not starting the daemon in background (sysrepo-plugind -d) and
   let do it by start-stop-daemon with "-b" option. But then the log
   messages of the daemon will not longer ends in the syslog but to stderr.
C) Start the daemon without a pidfile and stop the daemon with the
   "-x" option.

The only valid option is C to fix that.

Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com>
[yann.morin.1998@free.fr: introduce EXECUTABLE]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

DEVELOPERS: remove Carlos Santos

Goodbye!

Signed-off-by: Carlos Santos <unixmania@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/xen: security bump to version 4.13.1

- Fix CVE-2020-11739: An issue was discovered in Xen through 4.13.x,
allowing guest OS users to cause a denial of service or possibly gain
privileges because of missing memory barriers in read-write unlock
paths. The read-write unlock paths don't contain a memory barrier. On
Arm, this means a processor is allowed to re-order the memory access
with the preceding ones. In other words, the unlock may be seen by
another processor before all the memory accesses within the "critical"
section. As a consequence, it may be possible to have a writer executing
a critical section at the same time as readers or another writer. In
other words, many of the assumptions (e.g., a variable cannot be
modified after a check) in the critical sections are not safe anymore.
The read-write locks are used in hypercalls (such as grant-table ones),
so a malicious guest could exploit the race. For instance, there is a
small window where Xen can leak memory if XENMAPSPACE_grant_table is
used concurrently. A malicious guest may be able to leak memory, or
cause a hypervisor crash resulting in a Denial of Service (DoS).
Information leak and privilege escalation cannot be excluded.

- Fix CVE-2020-11740: An issue was discovered in xenoprof in Xen through
4.13.x, allowing guest OS users (without active profiling) to obtain
sensitive information about other guests. Unprivileged guests can
request to map xenoprof buffers, even if profiling has not been enabled
for those guests. These buffers were not scrubbed.

- Fix CVE-2020-11741: An issue was discovered in xenoprof in Xen through
4.13.x, allowing guest OS users (with active profiling) to obtain
sensitive information about other guests, cause a denial of service, or
possibly gain privileges. For guests for which "active" profiling was
enabled by the administrator, the xenoprof code uses the standard Xen
shared ring structure. Unfortunately, this code did not treat the guest
as a potential adversary: it trusts the guest not to modify buffer size
information or modify head / tail pointers in unexpected ways. This can
crash the host (DoS). Privilege escalation cannot be ruled out.

- Fix CVE-2020-11742: An issue was discovered in Xen through 4.13.x,
allowing guest OS users to cause a denial of service because of bad
continuation handling in GNTTABOP_copy. Grant table operations are
expected to return 0 for success, and a negative number for errors. The
fix for CVE-2017-12135 introduced a path through grant copy handling
where success may be returned to the caller without any action taken. In
particular, the status fields of individual operations are left
uninitialised, and may result in errant behaviour in the caller of
GNTTABOP_copy. A buggy or malicious guest can construct its grant table
in such a way that, when a backend domain tries to copy a grant, it hits
the incorrect exit path. This returns success to the caller without
doing anything, which may cause crashes or other incorrect behaviour.

- Fix CVE-2020-11743: An issue was discovered in Xen through 4.13.x,
allowing guest OS users to cause a denial of service because of a bad
error path in GNTTABOP_map_grant. Grant table operations are expected to
return 0 for success, and a negative number for errors. Some misplaced
brackets cause one error path to return 1 instead of a negative value.
The grant table code in Linux treats this condition as success, and
proceeds with incorrectly initialised state. A buggy or malicious guest
can construct its grant table in such a way that, when a backend domain
tries to map a grant, it hits the incorrect error path. This will crash
a Linux based dom0 or backend domain.

https://xenproject.org/downloads/xen-project-archives/xen-project-4-13-series/xen-project-4-13-1

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/mp4v2: fix build with gcc <= 5

Fixes:
 - http://autobuild.buildroot.org/results/14937c96a82fb3d10e5d83bd7b2905b846fb09f9

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[yann.morin.1998@free.fr: expand the patch' commit log]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

boot/arm-trusted-firmware: ignore licencing check for user defined official version

The commit [1] "licensing info is only valid for v1.4" fixed the legal-info
issues when a custom ATF tarball or a version from git is used.
But we need to ignore licencing for a used defined official ATF version.

Althougt the ATF version are licensed under BSD-3-Clause, the license
file can be updated between version (for example between v1.4 and v2.0).

Ignore the licencing check if the user provide a custom official version.

[1] d1a61703f728340ec894c367398d2a3a394a3360

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Yann E. MORIN <yann.morin.1998@free.fr>
[yann.morin.1998@free.fr: use positive logic with the _LATEST option]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

configs/freescale_imx8mnevk: new defconfig

This patch adds support for the i.MX8M Nano EVK Board [1].

The final boot image is created from uboot and firmware binaries in post
image script board/freescale/common/imx/imx8-bootloader-prepare.sh

Note that this config is based on 4.19.35_1.1.0 ; hence, the kernel
configuration needs host-openssl to build.

[1] https://www.nxp.com/design/development-boards/i-mx-evaluation-and-development-boards/evaluation-kit-for-the-i-mx-8m-nano-applications-processor:8MNANOD4-EVK

Signed-off-by: Stephane Viau <stephane.viau@oss.nxp.com>
Reviewed-by: Maeva Manuel <maeva.manuel@oss.nxp.com>
Reviewed-by: Julien Olivain <julien.olivain@oss.nxp.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

board/freescale/common/imx: add support for i.MX8M Nano

Signed-off-by: Stephane Viau <stephane.viau@oss.nxp.com>
Reviewed-by: Maeva Manuel <maeva.manuel@oss.nxp.com>
Reviewed-by: Julien Olivain <julien.olivain@oss.nxp.com>
Reviewed-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

board/freescale/common/imx: use generic ddr_fw.bin name

For some SoCs (i.e.: i.MX8M Nano for now), the DDR training firmware
in the boot image depends on the type of memory installed on a board
(LPDDR4, DDR4, etc...). The previous patch makes this firmware name
agnostic of the DDR type and uses ddr_fw.bin as a generic name. This
patch makes imx8-bootloader-prepare.sh use this generic name,
independently of the kind board DDR type.

Signed-off-by: Stephane Viau <stephane.viau@oss.nxp.com>
Reviewed-by: Maeva Manuel <maeva.manuel@oss.nxp.com>
Reviewed-by: Julien Olivain <julien.olivain@oss.nxp.com>
Reviewed-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/freescale-imx/firmware-imx: add choice for DDR training binaries

Several i.MX8 (e.g.: 8M, 8MM, 8MN) support many DDR types (LPDDR4, DDR4,
etc.), for which the DDR training is performed in the bootloader.
Some boards have LPDDR4 (e.g.: nitrogen8mn) and some others have the DDR4
(e.g.: NXP's reference board EVK). This patch allows the selection of either
of the binaries used to train the DDR.

Note that DDR and HDMI FW copies are now separated, which makes latter
being properly copied only for the relevant SoC (i.MX8MQ).
Suggested-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Stephane Viau <stephane.viau@oss.nxp.com>
Reviewed-by: Maeva Manuel <maeva.manuel@oss.nxp.com>
Reviewed-by: Julien Olivain <julien.olivain@oss.nxp.com>
Reviewed-by: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/freescale-imx: Add option for DDR FW need

Only some i.MX8 needs a DDR training firmware (8M, 8MM, 8MN). Some other
i.MX8 (QuadMax, QuadXPlus) rely on system controller for that task.

Suggested-by: Julien Olivain <julien.olivain@oss.nxp.com>
Signed-off-by: Stephane Viau <stephane.viau@oss.nxp.com>
Reviewed-by: Gary Bisson <gary.bisson@boundarydevices.com>
[yann.morin.1998@free.fr:
  - use the new variable to drive the firmware installation
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/dav1d: bump version to 0.7.0

Release notes:
https://code.videolan.org/videolan/dav1d/-/releases/0.7.0

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

package/mongoose: bump to to version 6.18

See https://github.com/cesanta/mongoose/releases/tag/6.18

Signed-off-by: Pierre-Jean Texier <pjtexier@koncepto.io>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/libubootenv: bump to version c91dcca

c91dcca ubi, write: fix invalid envsize ptr to UBI_IOCVOLUP

Signed-off-by: Pierre-Jean Texier <pjtexier@koncepto.io>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/{fmc,fmlib}: change repository location

Now that Freescale has been wholly swallowed into NXP, the public-facing
git repositories that were hosting those two packages are no longer
available.

Fortunately, they had been mirrored on Code Aurora forge (a Linux
Foundation project, so relatively stable and trustworthy), which has the
tags we need, and that generates the exact same archives.

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Matthew Weber <matthew.weber@rockwellcollins.com>
Reviewed-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>