busybox: rewrite logging init script

- Split S01logging into S01syslogd and S02klogd. Install them only if no
  other syslog package is selected and the corresponding daemons are
  selected in the Busybox configuration.
- Support /etc/default/$DAEMON configuration files.
- Detect and report start/stop errors (previous version ignored them and
  always reported OK).
- Use a separate function for restart.
- Implement reload as restart.

The dependency of busybox on rsyslog and syslog-ng was only needed
because those packages also installed S01logging. Since now they no
longer install the same file, these dependencies are no longer needed.
The dependency on sysklogd is still needed since that one installs the
syslogd and klogd executables with the same name as busybox.

The -n option of syslogd/klogd is obligatory because start-stop-daemon
starts it in the background. Therefore, move it out of the
SYSLOGD_ARGS resp. KLOGD_ARGS variable so the user can no longer remove
it.

Signed-off-by: Carlos Santos <casantos@datacom.com.br>
Reviewed-by: Matt Weber <matthew.weber@rockwellcollins.com>
[Arnout: keep dependency on sysklogd]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/waf: purge trailing '/' from WAF_SITE

<PKG>_SITE cannot have a trailing slash.

This was not detected by the check in generic-package because it is a
host-only package without Config.in symbol.

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

package/cryptopp: purge trailing '/' from CRYPTOPP_SITE

<PKG>_SITE cannot have a trailing slash.

This was not detected by the check in generic-package because it is a
host-only package without Config.in symbol.

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

package/pkgconf: purge trailing '/' from PKGCONF_SITE

<PKG>_SITE cannot have a trailing slash.

This was not detected by the check in generic-package because it is a
host-only package without Config.in symbol.

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

package/netsurf: new package

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
[Thomas:
 - rename BR2_PACKAGE_NETSURF_FRAMEBUFFER to BR2_PACKAGE_NETSURF_SDL,
   and use "sdl" everywhere instead of "framebuffer", since really SDL
   can display on something else than the framebuffer
 - use a "depends on BR2_PACKAGE_LIBGTK2" instead of a "depends on
   BR2_PACKAGE_XORG7 + many other depends on + select
   BR2_PACKAGE_LIBGTK2"
 - use PKG_CONFIG_HOST_BINARY for the pkg-config path]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/python-sentry-sdk: new package

Signed-off-by: Joseph Kogut <joseph.kogut@gmail.com>
Reviewed-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/qt5/qt5base: do not change QMAKE_COMPILER in g++-base.conf

The variable QMAKE_COMPILER determines the compiler flavour (e.g.
gcc, clang, llvm, msvc etc.) and not a compiler path. Changing
it confuses the platform test of the qt5webengine package
(mkspecs/features/platform.prf):

  defineTest(isPlatformSupported) {
    QT_FOR_CONFIG += gui-private
    linux {
      if(!gcc:!clang)|intel_icc {
        skipBuild("Qt WebEngine on Linux requires clang or GCC.")
        return(false)
      }
      gcc:!clang:!isGCCVersionSupported(): return(false)
    }
  }

Revert the variable change introduced by commit 'package/qt5/qt5base: use
ccache for building host code' ([2]).

Fixes raspberrypi3_qt5we_defconfig/qt5webengine build ([1]):

  Qt WebEngine on Linux requires clang or GCC.
  QtWebEngine will not be built.

[1] https://gitlab.com/buildroot.org/buildroot/-/jobs/131217091
[2] https://git.buildroot.net/buildroot/commit/?id=c2ea056a1b34fd1354100d08e603b9374e225968

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

package/x11r7/xcb-proto: remove pkgconf fix

Not needed anymore after pkgconf bump.

Fixes
http://autobuild.buildroot.net/results/175/1752915119b795c916d75594e51a7c79126afd5f/

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

package/{mesa3d, mesa3d-headers}: bump version to 18.3.0

Rebased patch 0002 after upstream commit
https://cgit.freedesktop.org/mesa/mesa/commit/configure.ac?h=18.3&id=6b8657aff0adb9952aa3178d3b332de19d354561

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

package/gcc: remove unused HOST_GCC_FAKE_TESTSUITE

HOST_GCC_FAKE_TESTSUITE was removed since commit [1].

[1] 545ca6a0f2f4e0d8b3812949476ac2856264b83c

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

package/nodejs: security bump to version 8.14.0

Fixes the following security vulnerabilities:

- Node.js: Denial of Service with large HTTP headers (CVE-2018-12121)
- Node.js: Slowloris HTTP Denial of Service (CVE-2018-12122 / Node.js)
- Node.js: Hostname spoofing in URL parser for javascript protocol
  (CVE-2018-12123)
- Node.js: HTTP request splitting (CVE-2018-12116)
- OpenSSL: Timing vulnerability in DSA signature generation (CVE-2018-0734)
- OpenSSL: Microarchitecture timing vulnerability in ECC scalar
  multiplication (CVE-2018-5407)

For more details, see the announcement:
https://nodejs.org/en/blog/release/v8.14.0/

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

utils/scanpypi: import third party modules after the standard ones

Move imports from six package after the standard modules. Resolves
pylint warnings.

Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/gdb: fix compilation for arc-2018.09 in some cases

In some cases which seem to depend on how toolchain headers
got installed or configured we may face well-known by now failure
fixed in upstream with [1]:
------------------------------>8----------------------------
In file included from nat/linux-ptrace.c:20:
nat/linux-ptrace.h:175:22: error: expected identifier before numeric constant
 # define TRAP_HWBKPT 4
                      ^
nat/linux-ptrace.h:175:22: error: expected '}' before numeric constant
In file included from .../output/host/arc-buildroot-linux-gnu/sysroot/usr/include/signal.h:58,
                 from build-gnulib/import/signal.h:52,
                 from .../output/host/arc-buildroot-linux-gnu/sysroot/usr/include/sys/wait.h:36,
                 from ./common/gdb_wait.h:23,
                 from nat/linux-ptrace.c:24:
.../output/host/arc-buildroot-linux-gnu/sysroot/usr/include/bits/siginfo-consts.h:156:1: note: to match this '{'
 {
 ^
In file included from nat/linux-ptrace.c:20:
nat/linux-ptrace.h:175:22: error: expected unqualified-id before numeric constant
 # define TRAP_HWBKPT 4
                      ^
In file included from .../output/host/arc-buildroot-linux-gnu/sysroot/usr/include/features.h:428,
                 from .../output/host/arc-buildroot-linux-gnu/sysroot/usr/include/bits/libc-header-start.h:33,
                 from .../output/host/arc-buildroot-linux-gnu/sysroot/usr/include/stdio.h:27,
                 from build-gnulib/import/stdio.h:43,
                 from ./common/common-defs.h:52,
                 from nat/linux-ptrace.c:19:
.../output/host/arc-buildroot-linux-gnu/sysroot/usr/include/sys/wait.h:158:1: error: expected declaration before '}' token
 __END_DECLS
 ^~~~~~~~~~~
------------------------------>8----------------------------

Back-porting the fix for ARC now to get predictably successful results.

[1] https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commit;h=5a6c3296a7a90694ad4042f6256f3da6d4fa4ee8

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/libopenssl: use HTTPS for URL

The host forces HTTPS regardless. This can be seen in the build logs:

 >>> host-libopenssl 1.0.2q Downloading
 URL transformed to HTTPS due to an HSTS policy
 --2018-12-10 09:53:27--  https://www.openssl.org/source/openssl-1.0.2q.tar.gz

Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/qt5/qt5virtualkeyboard: rewrap Config.in help text

check-package complains with:

package/qt5/qt5virtualkeyboard/Config.in:59: help text: <tab><2 spaces><62 chars> (http://nightly.buildroot.org/#writing-rules-config-in)

So let's rewrap the Config.in help text.

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

package/libpjsip: add optional dependency on libgsm

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

package/libpjsip: add optional dependency on speex

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

package/libpjsip: disable build of test binaries

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

package/asterisk: enable for uclibc toolchains

This patch extends the configure checks for re-entrant resolver
functions to fix uclibc builds.

Quoting Yann:
http://lists.busybox.net/pipermail/buildroot/2017-September/203004.html

"As a final stroke of genius, asterisk checks for the re-entrant variant
of res_ninit(), and concludes that all such functions are available,
including res_nsearch(). Uclibc-ng has the former but not the latter, so
the build fails. Since there is no cache variable for that check, we
can't pre-feed that result to configure, and fixing it is a bigger
endeavour. So we make asterisk depend on glibc for now, until someone
is brave enough to fix it."

Musl builds are still broken:

output/build/asterisk-16.0.0/include/asterisk/astmm.h:165:35:
 error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘->’ token
  Do_not_use_calloc__use_ast_calloc->fail(a, b)

output/build/asterisk-16.0.0/include/asterisk/astmm.h:169:77:
 error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘->’ token
  Do_not_use_free__use_ast_free_or_ast_std_free_for_remotely_allocated_memory->fail(a)

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

package/asterisk: bump version to 16.0.0

- removed patches applied upstream, re-numbered remaining patches
- not available for static builds anymore:
  https://github.com/asterisk/asterisk/commit/8e36064109b6fabc6cd8caa59e48f9b523c7485b
- fixed license hashes after upstream whitespace removal
  https://github.com/asterisk/asterisk/commit/fd0ca1c3f9b972a52d48a82b492fd6bac772dc78
- removed configure options not provided by upstream anymore
- fixed configure error, the file is included in asterisk source:
    checking for bridges/bridge_softmix/include/hrirs.h... configure:
    error: cannot check for file existence when cross compiling
- added "-without-pjproject-bundled" as noted in
  https://wiki.asterisk.org/wiki/display/AST/New+in+15
- upstream switched from ncurses to libedit:
  https://github.com/asterisk/asterisk/commit/d6fda173a40e078db219e9aaceb7f3ce9ba1b4e4
- added libatomic when needed
- updated core sound package

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

package/asterisk: libsrtp needs openssl

Fixes configure error

checking for the ability of -lsrtp2 to be linked in a shared object... yes
checking for srtp_crypto_policy_set_aes_cm_256_hmac_sha1_80 in -lsrtp2... yes
checking for srtp_crypto_policy_set_aes_cm_192_hmac_sha1_80 in -lsrtp2... no
checking for srtp_crypto_policy_set_aes_gcm_128_8_auth in -lsrtp2... no
checking for srtp_shutdown in -lsrtp2... yes
checking for srtp2/srtp.h... (cached) yes
configure: WARNING: ***
configure: WARNING: *** OpenSSL required when using libsrtp2, checking for libsrtp instead.
configure: WARNING: ***

using this defconfig:

BR2_PACKAGE_LIBSRTP=y
BR2_PACKAGE_ASTERISK=y

Please note that openssl support in libsrtp is not available for static
builds:
https://git.buildroot.net/buildroot/tree/package/libsrtp/libsrtp.mk#n27

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

configs/liteboard: remove explicit getty port

Using default value (console) works well, so there is no reason to set
tty explicitly.

Signed-off-by: Marcin Niestroj <m.niestroj@grinn-global.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

configs/chiliboard: remove explicit getty port

Using default value (console) works well, so there is no reason to set
tty explicitly. Additionally after selecting newer kernels (tested
with 4.19 and 4.20-rc3) ttyO0 no longer works due to missing device
node.

Signed-off-by: Marcin Niestroj <m.niestroj@grinn-global.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Makefile: offload .gitlab-ci.yml generation

GitLab has severe limitations imposed to triggers.
Using a variable in a regexp is not allowed:
|    only:
|        - /-$CI_JOB_NAME$/
|        - /-\$CI_JOB_NAME$/
|        - /-%CI_JOB_NAME%$/
Using the key 'variables' always lead to an AND with 'refs', so:
|    only:
|        refs:
|            - branches
|            - tags
|        variables:
|            - $CI_JOB_NAME == $CI_COMMIT_REF_NAME
would make the push of a tag not to trigger all jobs anymore.
Inheritance is used only for the second level of keys, so:
|.runtime_test: &runtime_test
|    only:
|        - tags
|tests.package.test_python_txaio.TestPythonPy2Txaio:
|    <<: *runtime_test
|    only:
|        - /-TestPythonPy2Txaio$/
would override the entire key 'only', making the push of a tag not to
trigger all jobs anymore.

So, in order to have a trigger per job and still allow the push of a tag
to trigger all jobs (all this in a follow up patch), the regexp for each
job must be hardcoded in the .gitlab-ci.yml and also the inherited
values for key 'only' must be repeated for every job.
This is not a big issue, .gitlab-ci.yml is already automatically
generated from a template and there will be no need to hand-editing it
when jobs are added or removed.

Since the logic to generate the yaml file from the template will become
more complex, move the commands from the main Makefile to a script.

Using Python or other advanced scripting language for that script would
be the most versatile solution, but that would bring another dependency
on the host machine, pyyaml if Python is used. So every developer that
needs to run 'make .gitlab-ci.yml' and also the docker image used in the
GitLab pipelines would need to have pyyaml pre-installed.
Instead of adding the mentioned dependency, keep using a bash script.

While moving the commands to the script:
 - mimic the behavior of the previous make target and fail on any
   command that fails, by using 'set -e';
 - break the original lines in one command per line, making the diff for
   any patch to be applied to this file to look nicer;
 - keep the script as simple as possible, without functions, just a
   script that executes from the top to bottom;
 - do not perform validations on the input parameters, any command that
   fails already makes the script to fail;
 - do not add an usage message, the script is not intended to be called
   directly.

This patch does not change functionality.

Signed-off-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
[Thomas: make the script output on stdout rather than take the output
file name as second argument.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

configs/aarch64_efi: needs host-{dosfstools, mtools}

Fixes:
    https://gitlab.com/ymorin/buildroot/-/jobs/131924126

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Erico Nunes <nunes.erico@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Reviewed-by: Erico Nunes <nunes.erico@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/lua: remove 5.2.x version

At this time :
- Lua 5.3.5 will be the last one of its serie.
- Lua 5.4 is up coming (lua-5.4.0-work2 is already available).
- Lua 5.2.4 was released on 2015.
For various reasons in the Lua ecosystem, the Lua 5.1 will stay.
On BR, Lua 5.3 is the default version since 2016.02.

So, the serie which could be removed is the 5.2.x.
We could wait some days for other user feedback.

Note: see discussion when 5.3.x was introduced :
http://lists.busybox.net/pipermail/buildroot/2015-January/117638.html

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

package/qt5/qt5virtualkeyboard: add arrow key navigation build option

It allows controlling the keyboard using the arrow and return keys.

Signed-off-by: Sébastien Szymanski <sebastien.szymanski@armadeus.com>
Reviewed-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/proftpd: add mod_sql_sqlite

Enable mod_sql_sqlite as a compile time option

Signed-off-by: Jared Bents <jared.bents@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/proftpd: fix typoes in Config.in help texts

mode_ -> mod_

Reported-by: Jared Bents <jared.bents@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

.gitlab-ci.yml: store .config files as artefacts for defconfig tests

Add Buildroot's own .config file, as well as any package's .config file
(uclibc, linux, and busybox), for later inspection should a build fails,
notably due to changes in the kconfig-package infrastructure.

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

configs/nexbox_a95x: kernel needs host-openssl

Since commit 9f5b07fc64 ("configs: nexbox_a95x_defconfig: bump to
kernel 4.19.8"), the nexbox_a95x_defconfig needs host-openssl to build
the Linux kernel.

Fixes: https://gitlab.com/ymorin/buildroot/-/jobs/131924236
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

docs/manual: document <pkg>_KEEP_DEPENDENCIES in rebar-package infra

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

package/pkg-rebar.mk: change semantic of <PKG>_KEEP_DEPENDENCIES

Set it to NO by default and check if it is equal to NO. This is to
be more consistent with other boolean variables in Buildroot.

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

package/pkg-rebar.mk: fix remove-rebar-config-dependencies

Handle the case where there is only one dependency described in
rebar.config, so when the line starts by '{deps' and ends by '}.'.
Before it was deleting this line but also all next lines until finding
a line that ends by '}.'.

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

support/download/dl-wrapper: fix urlencode option never being sent to backend

Since commit 38de434123 ("download: fix file:// BR2_PRIMARY_SITE
(download cache)"), the urlencode option is no longer passed to the
download backend, because we use ${backend} instead of
${backend_urlencode}.

We must get the urlencode information from backend_urlencode.

Signed-off-by: Damien Thébault <damien.thebault@vitec.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
[Thomas: rework commit log]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/python-aiorwlock: new package

Read write lock for asyncio.

Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
Reviewed-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/luajit: add option to enable Lua 5.2 compatibility

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

package/pkgconf: bump to 1.5.3, change upstream

The pkgconf project now has a website at pkgconf.org, and hosts its
tarballs at https://distfiles.dereferenced.org/pkgconf/, so this
commit updates the upstream location, and uses the xz-compressed
tarball as well.

pkgconf is bumped to 1.5.3. What prompted this update is the fact that
GStreamer uses the --define-prefix option when calling pkg-config, and
this option didn't exist in pkgconf 0.9.2.

The patch 0001-Fix-all-variables-sysroot-prefix-problem.patch is
dropped, because pkgconf now behaves properly, by prefixing all paths
with the sysroot. This has been verified by testing libdir and
includedir in zlib.pc, and adding some dummy pkgdatadir, mapdir and
sdkdir variables:

$ cat staging/usr/lib/pkgconfig/zlib.pc
prefix=/usr
exec_prefix=${prefix}
libdir=${exec_prefix}/lib
sharedlibdir=${libdir}
includedir=${prefix}/include
pkgdatadir=${prefix}/pouet
mapdir=${prefix}/this/is/map/dir
sdkdir=${prefix}/this/is/sdk/dir
[...]

$ ./host/bin/pkg-config --variable=libdir zlib
./host/bin/../arm-buildroot-linux-uclibcgnueabi/sysroot/usr/lib
$ ./host/bin/pkg-config --variable=includedir zlib
./host/bin/../arm-buildroot-linux-uclibcgnueabi/sysroot/usr/include
$ ./host/bin/pkg-config --variable=mapdir zlib
./host/bin/../arm-buildroot-linux-uclibcgnueabi/sysroot/usr/this/is/map/dir
$ ./host/bin/pkg-config --variable=sdkdir zlib
./host/bin/../arm-buildroot-linux-uclibcgnueabi/sysroot/usr/this/is/sdk/dir
$ ./host/bin/pkg-config --variable=pkgdatadir zlib
./host/bin/../arm-buildroot-linux-uclibcgnueabi/sysroot/usr/pouet

So, the 0001-Fix-all-variables-sysroot-prefix-problem patch is no
longer necessary.

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

package/prosody: bump to version 0.11.1

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

package/gcc/7.4.0: remove upstream patch

0001-m68k-coldfire-pr68467.patch is already upstream and backported to
gcc 7.4.0 [1].

[1] https://github.com/gcc-mirror/gcc/commit/2d1d94491cc66e240f9ca2845ec64e8ffe851c74

Fixes:
https://gitlab.com/ymorin/buildroot/-/jobs/131924271
https://gitlab.com/ymorin/buildroot/-/jobs/131924272

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

libmbim: Bump to version 1.16.2

+ change website to https

Signed-off-by: Petr Vorel <petr.vorel@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python-aiojobs: new package

Jobs scheduler for managing background task (asyncio).

Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
Reviewed-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python-aiohttp-remotes: new package

Control remote side information.

Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
Reviewed-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python-aiohttp-jinja2: new package

Jinja2 template renderer for aiohttp.web

Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
Reviewed-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python-aiohttp-security: new package

Security for aiohttp.web.

Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
Reviewed-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python-aiohttp-sse: new package

Server-sent events support for aiohttp.

Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
Reviewed-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python-aiohttp-session: new package

Sessions for aiohttp.web.

Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
Reviewed-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

stress-ng: fix build without swapcontext

Fixes:
 - http://autobuild.buildroot.org/results/f2d3b06afa6e31527a71c03671c8f08eb3f46c36

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

minnowboard-max: defconfigs: bump kernel to 4.19.8

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

libmpd: fix build with strndup

Retrieve a patch from upstream to include config.h otherwise build will
fail when trying to redefine strndup:
libmpd-internal.h:210:10: error: expected identifier or '(' before '__extension__'
 char *   strndup     (const char *s, size_t n);

Indeed, without an include on config.h, HAVE_STRNDUP won't be defined

Fixes:
 - http://autobuild.buildroot.org/results/a174818fa768b029d19b033139f9c5e0aaaed149

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

openzwave: fix restrict error with gcc 8

Fixes:
 - http://autobuild.buildroot.org/results/2a8d2c67bcd1a7c228e9196d3bdb7f608c44c856

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

vte: needs gcc >= 4.8

Fixes:
 - http://autobuild.buildroot.org/results/6dada849a87b68f6f69e3ed760995e86dfa25ad6

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

qt5quickcontrols: fix hash file entry

Fix accidental hash entry change previous commit ('qt5: bump latest version
to 5.11.3' 85ecae213ae5d37c9ddb7390706c18b93035f63b).

Reported-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

qt5: bump latest version to 5.11.3

qt5base: remove 0005-Export-qt_open64-from-QtCore.patch (taken from upstream [1])

[1] http://code.qt.io/cgit/qt/qtbase.git/commit/?id=4fc4f7b0ce0e6ee186a7d7fe9b5dd20e94efe432

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

package/gcc: remove arm soft patch

The patch refers to [1] which says "Unfortuantely, arm-gcc defaults to
generating code for armv5t." Since we always explicitly pass the target
CPU for ARM, the default CPU shouldn't matter.

As suggested by Arnout [2], a test based on qemu_arm_versatile_defconfig
has been done without this patch and there is no regression.

[1] https://sourceware.org/ml/crossgcc/2008-05/msg00009.html
[2] http://lists.busybox.net/pipermail/buildroot/2018-May/222104.html

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/gcc: remove uclibc-conf patch

This patch is present in Buildroot since a long time and has been rebased on
several version of gcc without beqing upstreamed.  Also it only concern
contrib/regression, which is not used at all during the build...

As suggested by Arnout [1], a test based on qemu_x86_defconfig has
been done without this patch and there is no regression.

[1] http://lists.busybox.net/pipermail/buildroot/2018-May/222104.html

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

configs: nexbox_a95x_defconfig: bump to kernel 4.19.8

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

package/gcc: bump to version 6.5.0

Remove upstream patches:

0002-fix-building-on-ppc64.patch
https://github.com/gcc-mirror/gcc/commit/765527ad3725c5f3e82ab2b8e5031120b409983d

0003-libsanitizer-Use-pre-computed-size.patch
https://github.com/gcc-mirror/gcc/commit/61f38c64c01a15560026115a157b7021ec67bd3b

942-Use-ucontext_t-not-struct-ucontext-in-linux-unwind.h.patch
https://github.com/gcc-mirror/gcc/commit/b685411208e0aaa79190d54faf945763514706b8

943-sanitizer-linux.patch
https://github.com/gcc-mirror/gcc/commit/8937b94d1a643fd9760714642296d034a45254a8

870-xtensa-fix-PR-target-82181.patch
https://github.com/gcc-mirror/gcc/commit/dbbb83538168c6648b53f5e164c3a30beda6464f

871-xtensa-fix-PR-target-65416.patch
https://github.com/gcc-mirror/gcc/commit/36f42654cbbc1df0e3d0040ba900c61980a8f8db

872-gcc-xtensa-fix-NAND-code-in-xtensa_expand_atomic.patch using patch
https://github.com/gcc-mirror/gcc/commit/9f149a1a929c0c6c22f572b1cd651c83ac83774c

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

package/gcc: bump to version 7.4.0

Remove upstream patches:

0002-fix-building-on-ppc64.patch
https://github.com/gcc-mirror/gcc/commit/aa65a43516da1d48011ef621ed5988289711d99b

0003-gcc-xtensa-fix-NAND-code-in-xtensa_expand_atomic.patch
https://github.com/gcc-mirror/gcc/commit/8709905244d50b9d81f85d534b21c29eb640078c

0003-xtensa-fix-PR-target-65416.patch
https://github.com/gcc-mirror/gcc/commit/5dcbd70ec0a97505d6153fcbc7ef899b5a2ec7fc

0004-libsanitizer-Use-pre-computed-size.patch
https://github.com/gcc-mirror/gcc/commit/61f38c64c01a15560026115a157b7021ec67bd3b

0005-gcc-xtensa-don-t-force-PIC-for-uclinux-target.patch
https://github.com/gcc-mirror/gcc/commit/6ef0a00dea2564f46a568b8681d8efccb4834cda

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

configs/aarch64_efi: Linux needs host-openssl to build

The Linux kernel builds fails with:

scripts/extract-cert.c:21:25: fatal error: openssl/bio.h: No such file or directory
 #include <openssl/bio.h>

Because it needs host-openssl.

Fixes:

  https://gitlab.com/buildroot.org/buildroot/-/jobs/131216892

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

DEVELOPERS: add entry for aarch64_efi_defconfig

Erico Nunes recently contributed this defconfig, we can safely assume
he is willing to maintain it.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Acked-by: Erico Nunes <nunes.erico@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

configs/bananapi_m2_ultra: add kernel headers version option

The bananapi_m2_plus defconfig uses a 4.18 kernel, but doesn't specify
anything for the kernel headers version, so the Buildroot default gets
used, which currently is 4.19. Since 4.19 is newer than 4.18,
Buildroot rightfully whines:

  Incorrect selection of kernel headers: expected 4.19.x, got 4.18.x

Let's fix this by telling this defconfig to use the headers from the
kernel, and that they are of version 4.18.

Fixes:

  https://gitlab.com/buildroot.org/buildroot/-/jobs/131216938

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>

DEVELOPERS: add entry for the bananapi_m2_ultra_defconfig

Lothar contributed this board not long ago, so let's assume he is
willing to take care of it.

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

support/testing: drop docstring in TestPythonPackageBase::test_run()

Having a docstring in the test_run() method clutters the "run-tests
-l" output:

test_run (tests.package.test_python_crossbar.TestPythonPy3Crossbar)
Test a python package. ... ok
[...]
test_run (tests.package.test_python_pexpect.TestPythonPy2Pexpect)
Test a python package. ... ok
test_run (tests.package.test_python_pexpect.TestPythonPy3Pexpect)
Test a python package. ... ok
test_run (tests.package.test_python_twisted.TestPythonPy2Twisted)
Test a python package. ... ok
test_run (tests.package.test_python_twisted.TestPythonPy3Twisted)
Test a python package. ... ok
test_run (tests.package.test_python_pynacl.TestPythonPy2Pynacl)
Test a python package. ... ok
test_run (tests.package.test_python_pynacl.TestPythonPy3Pynacl)
Test a python package. ... ok

So let's simply drop this docstring that is not particularly useful.

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

boot/afboot-stm32: create destination directory during installation

Following d0f4f95e390bcb1c953efa125f5277a8a235396e ("Makefile: rework
main directory creation logic"), BINARIES_DIR is not implicitly
created by the main Makefile at the beginning of the build, leaving
that up to whatever piece of code needs to install something in
$(BINARIES_DIR).

The afboot-stm32 package didn't pay attention to this, which this
commit fixes.

While at it, we move the afboot-stm32 installation into
<pkg>_INSTALL_IMAGES_CMDS, because using <pkg>_INSTALL_TARGET_CMDS to
install only files to BINARIES_DIR is a bit strange.

Fixes:

  https://gitlab.com/buildroot.org/buildroot/-/jobs/131217111

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>

php: security bump to version 7.2.13

Fixes CVE-2018-19518: University of Washington IMAP Toolkit 2007f on UNIX,
as used in imap_open() in PHP and other products, launches an rsh command
(by means of the imap_rimap function in c-client/imap4r1.c and the tcp_aopen
function in osdep/unix/tcp_unix.c) without preventing argument injection,
which might allow remote attackers to execute arbitrary OS commands if the
IMAP server name is untrusted input (e.g., entered by a user of a web
application) and if rsh has been replaced by a program with different
argument semantics.  For example, if rsh is a link to ssh (as seen on Debian
and Ubuntu systems), then the attack can use an IMAP server name containing
a "-oProxyCommand" argument.

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

pcm-tools: new package

Processor Counter Monitor (PCM) is an application programming interface
(API) and a set of tools based on the API to monitor performance and
energy metrics of Intel(R) Core(TM), Xeon(R), Atom(TM) and Xeon Phi(TM)
processors.

This package contains a patch on the pmu-query.py script to look for the
pcm-core program at the default path. It's not nice to have a Buildroot
specific patch but let's use one while we look for a solution that is
acceptable upstream.

Signed-off-by: Carlos Santos <casantos@datacom.ind.br>
[Peter: Needs C++, force X86_MSR on in linux]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

liboping: fix format-truncation error

liboping.c: In function 'ping_host_add':
liboping.c:207:9: error: '%s' directive output may be truncated writing
up to 255 bytes into a region of size 243 [-Werror=format-truncation=]
    "%s: %s", function, message);
         ^~
liboping.c:1644:40:
    ping_set_error (obj, "getaddrinfo", errmsg);
                                        ~~~~~~
liboping.c:206:2: note: 'snprintf' output between 14 and 269 bytes into
a destination of size 256
  snprintf (obj->errmsg, sizeof (obj->errmsg),
  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    "%s: %s", function, message);

Fixes:
 - http://autobuild.buildroot.org/results/b12d86388b495a96194e0bcbb5c19a4e35cbc53d

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

squashfs: do not force gzip support if lz4/xz/zstd is selected

The logic to ensure at least one compression backend is selected was not
updated when lz4, xz and zstd were introduced -  Fix that.

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Reviewed-by: Peter Seiderer <ps.report@gmx.net>
[Peter: add comment as suggested by Peter Seiderer]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

.gitlab-ci.yml: regenerate after prosody tests addition

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

support/testing: add prosody test

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

package/prosody: fix runtime dependencies

prosody is not available for Lua 5.3, but only 5.1 & 5.2.

prosody needs the module BitOp which is included in LuaJIT

note: the actual config is working only LuaJIT
Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

support/testing: add lua test

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Reviewed-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

.gitlab-ci.yml: update with recently added tests

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

support/testing: add perl-io-socket-ssl test

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

support/testing: add perl-io-socket-multicast test

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

support/testing: add perl-encode-decode test

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

support/testing: add perl-dbd-mysql test

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

utils/scancpan: allow the generation of more tests

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

support/testing/tests/packages/test_perl_*: regeneration

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

utils/scancpan: mark XS modules in docstring of test

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

package/pulseview: drop boost thread

Since bump to 0.4.0, boost thread is not needed anymore:
https://sigrok.org/gitweb/?p=pulseview.git;a=commit;h=b8f1cdeba7cf6c2b749ff60102ce4e260fa1309e

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

support/scripts/graph-depends: introduce MANDATORY_DEPS array

This array will be re-used in another function in a follow-up commit,
so it makes sense to factor it out.

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

support/scripts/graph-depends: use proper rootpkg in remove_extra_deps()

The remove_extra_deps() function removes dependencies that we are not
interested in seeing in the dependency graph. It does this for all
packages, except the 'all' package, which on full dependency graphs is
the root of the tree.

However, this doesn't take into account package-specific dependency
graphs (i.e make <pkg>-graph-depends) where the root is not 'all', but
'<pkg>'. Due to this, dependencies on "mandatory deps" were not
visible at all, i.e the toolchain package (and its dependencies) and
the skeleton package (and its dependencies) were not displayed in
package-specific dependency graphs.

To fix this, we use the existing rootpkg variable instead of
hardcoding 'all'.

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

support/graph-depends: drop legacy target-exceptions

Currently, we avoid drawing the dependencies that we call 'target
exceptions', becasue they initially were returned by 'show-targets',
when they in fact were not really packages and thus should not be on
the graph.

However, those two exceptions have no longer been reported in the output
of show-targets since we merged very old initial top-level parallel
build way back in 2014, with commit a24877586a56 (Makefile: add support
for top-level parallel make), where they had been converted into purely
internal rules.

4 years have passed, we can now drop those exceptions from the
graph-depends script.

This concludes the cleanup initiated three years ago with commit
0b32791f0076 (graph-depends: remove absent targets from
TARGET_EXCEPTIONS).

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>

DEVELOPERS: drop Bryan Brinsko entry

His e-mail @rockwellcollins.com is bouncing:

<bryan.brinsko@rockwellcollins.com>: host
    mail05.rockwellcollins.com[205.175.227.28] said: 550 #5.1.0 Address
    rejected. (in reply to RCPT TO command)

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

package/pulseview: fix a build error with libsigrokdecode disabled

Add a patch that fixes a build error occuring when libsigrokdecode is
not selected.

Fixes:

  http://autobuild.buildroot.net/results/79ed29bdd1eec9bf7cb5003202da5ceaeb8101b6/

Signed-off-by: Bartosz Golaszewski <brgl@bgdev.pl>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/pps-tools: bump version to fix usage without bash

As reported in bug #11426, the ppsfind shell script uses /bin/bash,
but the Buildroot pps-tools package doesn't depend on bash. In fact,
upstream has fixed the problem, and the script can now be used with a
POSIX shell, and the shebang is /bin/sh.

This commit therefore bumps pps-tools to the latest upstream commit,
which is precisely this fix.

Fixes bug #11426.

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

package/keyutils: bump to version 1.6

Refresh patch 0001.

Drop patch 0003; no longer needed after upstream headers rearrange.

Add license hash.

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

package/lynx: add dependency on host-pkgconf

The Lynx configure script uses pkg-config when available:

checking for nios2-buildroot-linux-gnu-pkg-config... /home/thomas/projets/buildroot/output/host/bin/pkg-config
checking pkg-config for openssl... yes
[...]
checking pkg-config for ncurses... yes

Using pkg-config avoids build failures such as:

checking for _nc_freeall... no
configure: error: Configuration does not support color-styles
make: *** [/home/test/autobuild/run/instance-1/output/build/lynx-2.8.9rel.1/.stamp_configured] Error 1

When building with "make lynx", so that pkg-config is not built
before. The issue is that in this case, lynx configure script picks up
the ncurses6-config script for the host ncurses instead of the one in
staging. Using pkg-config solves that nicely.

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

linux: ensure images/ exist before creating files there

When using an initramfs, on the first-pass build, we create a dummy cpio
so that the build succeeeds. The real cpio will come later, and we'll do
a second-pass build to use the actual cpio.

However, when we touch that dummy cpio, the images/ directory may not
yet exist, since commit d0f4f95e39 (Makefile: rework main directory
creation logic) removed its creation at the begining of the build, to
only at the moment we need it, i.e. during the *_INSTALL_IMAGES_CMDS
steps.

However, the linux build is not a _INSTALL_IMAGES_CMDS step, so there is
no guarantee that images/ already exist at that time.

Fix that by explicitly creating images/ before touching the dummy cpio.

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

package/python-aiohttp: add missing zlib dependency

Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/civetweb: fix build without sync

If BR2_TOOLCHAIN_HAS_SYNC_4 is not set, define -DNO_ATOMICS=1 to
fallback on lock

Fixes:
 - http://autobuild.buildroot.org/results/7b3bc71d64aedf617dde35798099e16d7ae46713

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

package/stress-ng: fix build without threads

semaphore.h is available only if HAVE_LIB_PTHREAD and
HAVE_SEM_POSIX are defined

Fixes:
 - http://autobuild.buildroot.org/results/1c95898b2833683a22bbe2ff8471fa08d94210e1

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[Thomas: update upstream status of the patch.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

DEVELOPERS: update my e-mail

Signed-off-by: Gaël PORTAY <gael.portay@collabora.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/libiscsi: add optional dependency to cunit

libiscsi uses cunit for its test suite, and autodetects its presence.

Usually, we just try to disable tests altogether, but there is no way to
do so. So, ensure proper ordering.

Note: there is an ac_cv_have_cunit variable, but it is not a true
ac_cv* cache variable, and the value provided on the command line or
environment is properly ignored.

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

package/libiscsi: do not built the manpages

The pre-rendered, bundled ones are still installed, though, but they
get removed in target-finalize anyway.

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

package/libiscsi: fix build due to warnings

Fixes:
    http://autobuild.buildroot.org/results/55b/55bf50fc7dcd465b71b5757434887dd3d0b25abc/
    http://autobuild.buildroot.org/results/98d/98dcfe5c9fc3babd5c8d3116d5128d437715c44e/

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

lxc: fix build without stack protector

Stack protector has been added in version 3.0.3 by
https://github.com/lxc/lxc/commit/2268c27754152aa538db2c9e3753d72d19bcd17a

However, some compilers could missed the needed library (-lssp or
-lssp_nonshared) at linking step so use ax_check_link_flag instead of
ax_check_compile_flag

Fixes:
 - http://autobuild.buildroot.org/results/0b90e7dca2984652842832a41abad93ac49a9b86

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

package/lxc: fix missing include for va_list

Fixes:
 - http://autobuild.buildroot.org/results/0b90e7dca2984652842832a41abad93ac49a9b86

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