Fix the following build failure on armv7 and sparc raised since bump to
version 3.7.1 in commit cc27267ae4 and
b90c52b169:
In file included from perf_memcpy32_s.c:7:
perf_private.h: In function 'rdtsc':
perf_private.h:72:3: error: 'asm' undeclared (first use in this function)
72 | asm volatile("mrc p15, 0, %0, c9, c14, 0" : "=r"(pmuseren));
| ^~~
Fixes:
- http://autobuild.buildroot.org/results/ceb13c071b1461eb6d73f5940d6b010095127f41
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This is a patch release for version 2 & 2.4, fixing a regression
introduced in 2.4.1.
https://github.com/cisco/libsrtp/releases/tag/v2.4.2
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The default currently is to rely on the package name to decide what to
build and install if not specified by the caller. This works nice for
target packages, where a 'foo' package will by default build and
install a 'foo' executable.
However, for host packages, that will build and install a 'host-foo'
exzcutable, which is not really, even really not, what would be
expected.
We fix that by using the package raw name, i.e. the package name with
the host- prefix yanked away.
It is very improbable that there already are many host-golang packages in
the wild (in br2-external trees), but if there are, they would forcibly
define those variables to a sane value. This change is not incompatible, as
the values provided by packages take precedence; it's just that those
packages now carry superfluous, if innocuous, variable assignments.
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Drop custom install rules which were added since commit
676797d57f. Indeed, they result in a
broken installation. Especially, they are trying to "guess" what must
be installed based on BR2_ARCH but oprofile has its own logic. For
example, goldmont microarchitecture files must be installed in i386
directory even if this architecture is 64 bits:
0ad5a9e6af
This will result in the following runtime failure:
oprofile: could not open unit mask description file /usr/share/oprofile//i386/goldmont/unit_masks
Unable to find info for event cpu_clk_unhalted
Fixes:
- https://bugs.buildroot.org/show_bug.cgi?id=14641
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This reverts commit 6e33e59080.
This unfortunately breaks a number of packages, as glibc errors out if 64bit
time_t is used without 64bit file offsets, and some packages undefine
_FILE_OFFSET_BITS leading to build breakage:
# if ! defined (_FILE_OFFSET_BITS) || _FILE_OFFSET_BITS != 64
# error "_TIME_BITS=64 is allowed only with _FILE_OFFSET_BITS=64"
..
So revert it for 2022.02.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
check-package now gives a warning about this. It's not a problem at the
moment, but it's a risk if an unconditional setting of _CONF_OPTS is
done earlier in the file.
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
As already done for {FOO}_DEPENDENCIES in commit
4910a175b3, check that {FOO}_CONF_OPTS are
never overridden in a conditional
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
When enabling MariaDB (BR2_PACKAGE_MARIADB=y) and systemd
(BR2_INIT_SYSTEMD=y) in buildroot, the mysqld.service fail to start
with a permission error. See output of command:
journalctl --unit=mysqld
Which shows:
systemd[1]: Starting MySQL database server...
install[102]: install: can't create directory '/var/log/mysql': Permission denied
systemd[1]: mysqld.service: Control process exited, code=exited, status=1/FAILURE
Since the service file includes the "User=mysql" directive, the
"ExecStartPre=" is executed as this user, which does not have
permission to create a directory in "/var/log".
This commit fixes this issue by adding the "!" prefix, which will
execute the command with full privileges. See the systemd.service manual
page entry for "ExecStart=", table "Special executable prefixes":
https://www.freedesktop.org/software/systemd/man/systemd.service.html#ExecStart=
or https://github.com/systemd/systemd/blob/v250/man/systemd.service.xml#L339
Moreover, the "mysql_install_db" invocation does not need this special
prefix, as the "/var/lib/mysql" directory on target is already owned
by the "mysql" user. The "chown" command is also useless and is
removed in this commit.
Reported-by: Gilles Talis <gilles.talis@gmail.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Tested-By: Gilles Talis <gilles.talis@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Fix the following build failure without threads raised since the
addition of the package in commit
0d4c8ce8a2:
/home/buildroot/autobuild/instance-0/output-1/host/opt/ext-toolchain/bin/../lib/gcc/arm-buildroot-linux-uclibcgnueabihf/9.3.0/../../../../arm-buildroot-linux-uclibcgnueabihf/bin/ld: cannot find -lpthread
Fixes:
- http://autobuild.buildroot.org/results/170d0114987b39a2b9ea173df7a5e806bd270fd8
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Even with the two upstream patches added in commit
b2e6e376a2, python-pycli still raises the
following build failure since bump of python3 to version 3.10.1 in
commit 25b1fc2898 due to the new "Multiple
Exception types without parentheses" exception
(https://docs.python.org/3/whatsnew/3.10.html):
error: File "/usr/lib/python3.10/site-packages/cli/test.py", line 142
except raises, e:
^^^^^^^^^
SyntaxError: multiple exception types must be parenthesized
So drop the package as the last release was made 10 years ago.
Fixes:
- http://autobuild.buildroot.org/results/6112e1830ce608abcea4a26b659c10e5ff09a66a
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
This is a minor release which fixes a couple of crashes and an incorrect
case of TLS certificate velidation when built with the OpenSSL backend.
Release notes:
b0d45b17f7/NEWS
Signed-off-by: Adrian Perez de Castro <aperez@igalia.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
python-treq needs TLS support in twisted so select
BR2_PACKAGE_PYTHON_TWISTED_TLS moreover it doesn't depend directly on
service_identity and pyOpenSSL since version 17.3.0 and
ef2ac7ea2b
Finally, python-idna was never a direct dependency of python-treq
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Reviewed-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
osm2pgsql doesn't support fmt > 8.0 yet resulting in the following build
failure since bump of fmt to version 8.1.1 in commit
ec7fd50d08:
/home/buildroot/autobuild/instance-0/output-1/host/aarch64-buildroot-linux-gnu/sysroot/usr/include/fmt/core.h:1728:7: error: static assertion failed: Cannot format an argument. To make type T formattable provide a formatter<T> specialization: https://fmt.dev/latest/api.html#udt
1728 | formattable,
| ^~~~~~~~~~~
Fixes:
- http://autobuild.buildroot.org/results/5af1a0ea176559a5dc787ba945a3a83823f5b1bb
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Reviewed-by: Maxim Kochetkov <fido_max@inbox.ru>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Sometimes new boards have not been tested with official docker so let's add
instructions to do it with utils/docker-run.
Cc: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
Reviewed-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
gdb's AC_LIB_HAVE_LINKFLAGS macro hardcodes a search starting with
/usr/lib/ which can lead to libgmp from the wrong architecture with the
result that the test fails. Even if a libgmp is found there it is now
the one that should be used.
This is the same macro used for expat and lzma for which there are
already specific CONF_OPTS flags added here. Add the same flag for
libgmp and move the handling down so that it is logically grouped with
the other similar options.
Note that there is no --with(out)-gmp flag to configure, as the
dependency is mandatory, so only the --with-libgmp-prefix option is
specified.
Signed-off-by: John Keeping <john@metanate.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Commit 616ba9c90b extended rootfs size, but
forgot to add quotes around it. Thus, kconfig removed the option
entirely.
While we're at it, move the option to a more appropriate place together
with the rest of the ext2 options.
Fixes:
https://gitlab.com/buildroot.org/buildroot/-/jobs/2103784289
Signed-off-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
For an unknown reason, installation of headers is broken since commit
7c1bb5c2bf which sets _PYTHON_PROJECT_BASE
As _PYTHON_PROJECT_BASE value seems correct and fix its own build
failures, use --install-headers to set install path for headers and
avoid the following build failure with python-greenlet or python-pygame:
python-greenlet: installs files in /home/peko/autobuild/instance-0/output-1/target//home/peko/autobuild/instance-0/output-1
python-pygame: installs files in /home/peko/autobuild/instance-1/output-1/target//home/peko/autobuild/instance-1/output-1
Fixes:
- http://autobuild.buildroot.org/results/f5b52dee7e72f0a1f72b9f87723bc585743c7b31
- http://autobuild.buildroot.org/results/9eb2685d32756864000e1b38f8a183b57f6ea528
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
This bump will fix the following build failure of kf5-kcoreaddons with
python 3.10 raised since commit
25b1fc2898:
CMake Error at /home/giuliobenetti/autobuild/run/instance-1/output-1/host/riscv32-buildroot-linux-gnu/sysroot/usr/share/ECM/find-modules/FindPythonModuleGeneration.cmake:149 (message):
The max python version in PythonModuleGeneration must be updated.
Call Stack (most recent call first):
src/lib/CMakeLists.txt:208 (find_package)
It will also fix the following build failure of kf5-kcoreaddons raised
since commit 5770a645a3:
In file included from /home/peko/autobuild/instance-0/output-1/host/arm-buildroot-linux-gnueabihf/sysroot/usr/include/qt5/QtCore/qchar.h:43,
from /home/peko/autobuild/instance-0/output-1/host/arm-buildroot-linux-gnueabihf/sysroot/usr/include/qt5/QtCore/qstring.h:49,
from /home/peko/autobuild/instance-0/output-1/host/arm-buildroot-linux-gnueabihf/sysroot/usr/include/qt5/QtCore/QString:1,
from /home/peko/autobuild/instance-0/output-1/build/kf5-kcoreaddons-5.47.0/src/lib/randomness/krandom.h:27,
from /home/peko/autobuild/instance-0/output-1/build/kf5-kcoreaddons-5.47.0/src/lib/randomness/krandom.cpp:22:
/home/peko/autobuild/instance-0/output-1/host/arm-buildroot-linux-gnueabihf/sysroot/usr/include/qt5/QtCore/qglobal.h:1274:80: note: declared here
1274 | Q_CORE_EXPORT QT_DEPRECATED_VERSION_X_5_15("use QRandomGenerator instead") int qrand();
| ^~~~~
/home/peko/autobuild/instance-0/output-1/build/kf5-kcoreaddons-5.47.0/src/lib/randomness/krandom.cpp: In function 'QString KRandom::randomString(int)':
/home/peko/autobuild/instance-0/output-1/build/kf5-kcoreaddons-5.47.0/src/lib/randomness/krandom.cpp:76:27: error: use of deleted function 'QCharRef& QCharRef::operator=(char)'
76 | str[i++] = char(r);
| ^
- Update indentation in hash file (two spaces)
- Update licenses due to
93480398b58ad494d957fa9ef0d848
Fixes:
- http://autobuild.buildroot.org/results/0267ff54ee82d853476ff7d6b14edb4aba308aa2
- http://autobuild.buildroot.org/results/694a2322d9161b55f5db66d1ace57d836360ed84
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
python3 is an optional dependency since the addition of the package in
commit bf48bf8629
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Makefiles hostapd.mk and wpa_supplicant.mk define different macros for
different defconfig options. Options that are not listed in defconfig,
can be set using CONFIG_SET. Options that are listed in defconfig, can
be enabled or disabled using CONFIG_ENABLE or CONFIG_DISABLE. Starting
from hostapd v2.10, option CONFIG_DPP is explicitly listed in defconfig.
So it would be time to switch to enable/disable macros for this option.
On the other hand, this approach looks fragile: we have to track
'promoted' defconfig options for each hostapd/wpa_s update. This patch
removes CONFIG_SET macros, keeping only CONFIG_ENABLED/CONFIG_DISABLED.
CONFIG_SET is replaced by the additional pass over enabled options. All
such options that are not listed in defconfig, will be appended to the
configuration file.
Fixes:
- http://autobuild.buildroot.net/results/3f8058ee6f3913fda795578f206db895731ba1e6/
Signed-off-by: Sergey Matyukevich <geomatsi@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
BR_GOMOD is a remnant from a previous iteration of the golang download
infrastructure; it is currently used nowhere.
Do not set it.
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Anisse Astier <anisse@astier.eu>
Cc: Christian Stewart <christian@paral.in>
Reviewed-by: Christian Stewart <christian@paral.in>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Build with per-package directory is broken, because go is not found in
the PATH, when trying to download and vendor a go package.
This is because FOO_DL_ENV contains $(HOST_GO_COMMON_ENV), which defines
the PATH as PATH=$(BR_PATH). This is correct, except this is expanded at
the time the golang-package macro is evaluated, which means PATH
contains the 'global' BR_PATH, i.e.: $(O)/host/bin:$(O)/host/sbin:...
However, with PPD, this does not yet exist at build time; only the
per-package host directory exists.
We want to have it expanded at the time the recipe is run, so like all
other variables, we need to $$-expand it.
At the same time, also $$-expand two other variables (even though those
are benign, it is better for consistency that they be $$-expanded).
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Anisse Astier <anisse@astier.eu>
Cc: Christian Stewart <christian@paral.in>
Reviewed-by: Christian Stewart <christian@paral.in>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Currently, host-go is a download dependency, because we need go for the
vendoring during the download step.
However, when using a package in override-srcdir, there is no download
step, so host-go is not pulled in as a dependency, which will break the
build of a single package from scratch.
We fix that by adding host-go as a standard dependency too.
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Christian Stewart <christian@paral.in>
Cc: Anisse Astier <anisse@astier.eu>
Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Reviewed-by: Christian Stewart <christian@paral.in>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Currently, the golang infrastructure forcibly sets the package _DL_ENV
variable, instead of appending to it, which prevents packages from
providing their own download environment variables.
We fix that by using an append-assignment.
Note: when introduced, that variable was explicitly not documented, and
is supposed to only be used by packages infrastructures. However, that
variable exists, and it can be (ab)used by br2-external packages (most
probably to ill effects, but heh!). We anyway leave it undocumented,
like _ROOTFS_PRE_CMD_HOOKS.
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Christian Stewart <christian@paral.in>
Cc: Anisse Astier <anisse@astier.eu>
Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Reviewed-by: Christian Stewart <christian@paral.in>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Currently, host-rustc is a download dependency, because we need cargo
for the vendoring during the download step.
However, when using a package in override-srcdir, there is no download
step, so host-rustc is not pulled in as a dependency which breaks
running the build of a single package from scratch:
$ make clean
$ make my-rust-package
[...]
[...] cargo build --offline --release --manifest-path Cargo.toml --locked
/bin/sh: cargo: command not found
We fix that by adding host-rustc as a standard dependency too.
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Some packages have their rust sources as a sub-directory, rather
than at the root of the source tree.
Do like we do for autotools-package, and use the package's _SRCDIR
rather than the top-level directory $(@D).
Additionally, in such a situation, it is more than probable that
the Cargo.toml is also present in that sub-directory, so use that
when vendoring the package.
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Currently, the cargo infrastructure forcibly sets the package _DL_ENV
variable, instead of appending to it, which prevents packages from
providing their own download environment variables.
We fix that by using an append-assignment.
Note: when introduced, that variable was explicitly not documented, and
is supposed to only be used by packages infrastructures. However, that
variable exists, and it can be (ab)used by br2-external packages (most
probably to ill effects, but heh!). We anyway leave it undocumented,
like _ROOTFS_PRE_CMD_HOOKS.
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Disable JIT with mips32r6 to avoid the following build failure raised
since bump to version 5.212.0-alpha4 in commit
df0b0fe691:
/tmp/ccHw2rHQ.s:18: Error: opcode not supported on this processor: mips32r6 (mips32r6) `addi $a0,$sp,-8'
Fixes:
- http://autobuild.buildroot.org/results/5191600b463cda078203eeffae880ed989f8ac0f
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fix the following build failure with gcc 4.8 raised since the addition
of -std=c++11 in commit b9eca02159:
sconex.cpp: In function 'scx::Date& scx::build_time()':
sconex.cpp:52:36: error: unable to find string literal operator 'operator"" __TIME__'
static Date s_build_time(__DATE__" "__TIME__,true);
^
Fixes:
- http://autobuild.buildroot.org/results/83ecf772965b4586eaea2682871a85732df7d62f
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
When using uboot's legacy build system, the 'make uboot-menuconfig'
and 'make uboot-savedefconfig' targets are not available as they
are created by 'kconfig-package'.
Signed-off-by: Simon Doppler <dopsi@dopsi.ch>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fix the following build failure with gcc <= 5 raised since bump to
version 2022-02-01 in commit d2cf91e0c6
and
a022cc0c55
./re2/sparse_array.h:266:37: warning: defaulted and deleted functions only available with -std=c++11 or -std=gnu++11
SparseArray<Value>::SparseArray() = default;
^
./re2/sparse_array.h: In copy constructor 're2::SparseArray<Value>::SparseArray(const re2::SparseArray<Value>&)':
./re2/sparse_array.h:273:3: error: 'copy_n' is not a member of 'std'
std::copy_n(src.sparse_.data(), src.max_size(), sparse_.data());
^
Fixes:
- http://autobuild.buildroot.org/results/8ebbac67f32fe24416a1f72aad0afd42b567b403
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Dependency on sqlite has been removed since v17.04.0-ce.
See: https://github.com/moby/moby/pull/30208
Signed-off-by: TIAN Yuanhao <tianyuanhao3@163.com>
Reviewed-by: Christian Stewart <christian@paral.in>
Tested-by: Marcus Hoffmann <marcus.hoffmann@othermo.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
To use time_t 64-bit for glibc >= 2.34 toolchains we have to set both
_FILE_OFFSET_BITS=64 and _TIME_BITS=64 for glibc toolchains. Buildroot
already define _FILE_OFFSET_BITS=64 since 2008 [1] before the first
release tag 2009.02.
_TIME_BITS is not needed for musl libc since it already year2038
ready [2].
The uclibc-ng libc only support time_t 32-bit (long int) so it will be
affected by the year2038 issue [3].
Fixes (in French, chapter Buildroot 2022 and GlibC):
https://www.blaess.fr/christophe/2038
Runtime tested with qemu_arm_vexpress_defconfig and the Bootlin glibc
bleeding-edge 2021.11-1 toolchain.
Before:
# date
Tue Jan 19 03:14:07 UTC 2038
# date
Thu Jan 1 00:00:00 UTC 1970
After:
# date
Tue Jan 19 03:14:07 UTC 2038
# date
Tue Jan 19 03:14:08 UTC 2038
# date
Tue Jan 19 03:14:09 UTC 2038
[1] 60b5eee76e
[2] https://git.musl-libc.org/cgit/musl/tree/include/alltypes.h.in?h=v1.2.2#n3
[3] https://cgit.uclibc-ng.org/cgi/cgit/uclibc-ng.git/tree/include/time.h?h=v1.0.40#n75https://cgit.uclibc-ng.org/cgi/cgit/uclibc-ng.git/tree/libc/sysdeps/linux/common/bits/types.h?h=v1.0.40#n106
Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Christophe Blaess <christophe.blaess@logilin.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This is a minor release which fixes an installation path and a crash
with certain Wayland compositors. Release notes can be found at:
https://wpewebkit.org/release/cog-0.12.1.html
Signed-off-by: Adrian Perez de Castro <aperez@igalia.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
# ipython -c 'import math; math.floor(12.3)'
-sh: ipython: not found
ipython use a wrong shebang on the target:
# head -n 1 /usr/bin/ipython
#!/home/kubu/buildroot/testsuite-master/TestIPythonPy3/host/bin/python
The issue come from python3-setuptools 60.0.0 that changed
SETUPTOOLS_USE_DISTUTILS default value from stdlib to local [1] to use
its local copy of distutils. This is actually a "breaking" change [2].
We have to override SETUPTOOLS_USE_DISTUTILS to the previous value
"stdlib" to fixes the issue.
While at it, use distutils from stdlib for the host.
Fixes:
https://gitlab.com/buildroot.org/buildroot/-/jobs/2088684091
[1] b6fcbbd00c
[2] https://github.com/pypa/setuptools/blob/v60.0.0/CHANGES.rst#breaking-changes
Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: James Hilliard <james.hilliard1@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
