Passing runtime configuration to tests in BRConfigTest class variables
requires the "fork" start method for multiprocessing. With
"forkserver" (default for POSIX platforms since Python 3.14) class
variables modified in the parent process are lost, because the child
process does not inherit the full Python state. This broke the way
support/testing/run-tests sets configuration for BRConfigTest, so a
start method override was added in commit
3d2141bcee.
Instead this patch adds a settings dataclass (requires Python >= 3.7)
which is serialized into an environment variable from run-tests and
read during BRConfigTest.__init__(). Reading is skipped if the
environment variable is not set so test discovery (not execution)
works without configuration (e.g. utils/get-developers uses this).
With that, run-tests no longer relies on a specific multiprocessing
start method.
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Now that we have added support for GCC 16.x, made GCC 15.x the
default, let's drop support for GCC 13.x.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
[Julien:
- remove remaining gcc-13 patches from commit 10ccbe079c
- move legacy option to 2026.08 version section
]
Signed-off-by: Julien Olivain <ju.o@free.fr>
As Thomas stated in 3bb260cf38:
The br-arm-internal-glibc.config is generally used as a configuration
to test the bleeding edge versions of components. However, it has been
lagging behind somewhat, so let's bring it up-to-date:
- Binutils 2.46.x
- GCC 16.x
Let the fun begin in the autobuilders!
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Now that GCC 16.x support has been added, follow our usual strategy of
making GCC 15.x the default GCC version.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
This patch allows to use an external toolchain based on gcc 16.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
In order to add gcc 16 support in follow-up commits, introduce
BR2_TOOLCHAIN_GCC_AT_LEAST_16 symbol.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
https://github.com/facebook/rocksdb/blob/v11.0.4/HISTORY.md
Removed patch 0001 which is included in this release:
53c8f739fd
Added upstream PR to fix build error.
Updated hash of README.md due to various upstream commits:
https://github.com/facebook/rocksdb/commits/v10.7.5/README.md
Switched build system to cmake.
Added configure parameter -DFAIL_ON_WARNINGS=OFF which converts many
"may be used uninitialized" errors into warnings which fixes build
errors introduced by the upcoming gcc version 16.x.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
[Julien: remove .checkpackageignore entry to fix check-package error]
Signed-off-by: Julien Olivain <ju.o@free.fr>
Added patch to enable rtti, needed for upcoming bump of rocksdb to fix
link error:
output/per-package/rocksdb/host/bin/../lib/gcc/x86_64-buildroot-linux-
gnu/15.2.0/../../../../x86_64-buildroot-linux-gnu/bin/ld:
librocksdb.so.10.7.5: undefined reference to `typeinfo for snappy::Sink'
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
OpenJDK 25 is the latest release.
See: https://endoflife.date/oracle-jdk
- BR2_OPENJDK_VERSION_LATEST is now set to 25.
- Add version-specific patches for 25.0.2+10:
- 0001: Add ARCv2 ISA processors support to Zero
- 0002: Compile OpenJDK in headless mode without requirements
- 0003: Fix ambiguous cmp() overload in aarch64 macro assembler
(older GCC 6.x compatibility)
- 0004: Fix constexpr on non-literal type in Shenandoah GC
(older GCC 6.x compatibility)
- Add -fpermissive to fix template definition error with older GCC.
- Update HOST_OPENJDK_BIN_VERSION for OpenJDK 25.
- Update the expectedVersion variable in JniTest.java from 0x00150000
(JNI 21) to 0x00180000 (JNI 24/25).
Tested with:
$ ./support/testing/run-tests -o ~/br-test-py/ -d ~/br-test-dl/ \
tests.package.test_openjdk.TestOpenJdk
Signed-off-by: Thomas Devoogdt <thomas@devoogdt.com>
[Bernd:
- rebased on bump of versions 17 & 21
- build-tested with gcc 16-snapshot]
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
[Julien:
- add .checkpackageignore entry to fix check-package error
- reformat patches with without numbering to fix check-package error
]
Signed-off-by: Julien Olivain <ju.o@free.fr>
Add comment to clarify that autoreconf is still needed even after the
patches are removed to update ax_cxx_compile_stdcxx.m4 for compatibility
with gcc >= 16.x
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
This patch adds fixes for cups-filters so it can be build with newer
versions of qpdf that made changes to PointerHolder:
https://github.com/qpdf/qpdf/blob/v12.3.2/manual/design.rst#smart-pointers
Build-tested using this defconfig:
BR2_PACKAGE_CUPS=y
BR2_PACKAGE_CUPS_FILTERS=y
arm-aarch64 [ 1/32]: OK
bootlin-aarch64-glibc [ 2/32]: OK
bootlin-aarch64-glibc-old [ 3/32]: SKIPPED
bootlin-arcle-hs38-uclibc [ 4/32]: OK
bootlin-armv5-uclibc [ 5/32]: OK
bootlin-armv7-glibc [ 6/32]: OK
bootlin-armv7m-uclibc [ 7/32]: SKIPPED
bootlin-armv7-musl [ 8/32]: OK
bootlin-m68k-5208-uclibc [ 9/32]: SKIPPED
bootlin-m68k-68040-uclibc [10/32]: OK
bootlin-microblazeel-uclibc [11/32]: OK
bootlin-mips64el-glibc [12/32]: OK
bootlin-mipsel32r6-glibc [13/32]: OK
bootlin-mipsel-uclibc [14/32]: OK
bootlin-openrisc-uclibc [15/32]: OK
bootlin-powerpc64le-power8-glibc [16/32]: OK
bootlin-powerpc-e500mc-uclibc [17/32]: OK
bootlin-riscv32-glibc [18/32]: OK
bootlin-riscv64-glibc [19/32]: OK
bootlin-riscv64-musl [20/32]: OK
bootlin-s390x-z13-glibc [21/32]: OK
bootlin-sh4-uclibc [22/32]: OK
bootlin-sparc64-glibc [23/32]: OK
bootlin-sparc-uclibc [24/32]: OK
bootlin-x86-64-glibc [25/32]: OK
bootlin-x86-64-musl [26/32]: OK
bootlin-x86-64-uclibc [27/32]: OK
bootlin-x86-i686-musl [28/32]: OK
bootlin-xtensa-uclibc [29/32]: OK
br-arm-basic [30/32]: SKIPPED
br-arm-full-nothread [31/32]: SKIPPED
br-arm-full-static [32/32]: SKIPPED
32 builds, 6 skipped, 0 build failed, 0 legal-info failed, 0 show-info failed
This solution avoids[1] a version bump of this package while keeping
compatibility with the newest version of qpdf. The bump of qpdf is
needed to fix build errors with gcc 16.x.
[1] https://lists.buildroot.org/pipermail/buildroot/2025-August/784931.html
Thomas: "which means duplicating all the crazy dependencies of
libcupsfilters yes :/"
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Changelog:
https://qpdf.readthedocs.io/en/stable/release-notes.html
Quoting the release notes for version 12.3.0:
"A C++20 compiler is now required to build or test qpdf."
Updated Config.in accordingly.
Updating source URL (Github) and updating build system to cmake.
Build-tested using this defconfig
BR2_PACKAGE_QPDF=y
BR2_PACKAGE_GNUTLS=y
BR2_PACKAGE_OPENSSL=y
arm-aarch64 [ 1/32]: OK
bootlin-aarch64-glibc [ 2/32]: OK
bootlin-aarch64-glibc-old [ 3/32]: SKIPPED
bootlin-arcle-hs38-uclibc [ 4/32]: OK
bootlin-armv5-uclibc [ 5/32]: OK
bootlin-armv7-glibc [ 6/32]: OK
bootlin-armv7m-uclibc [ 7/32]: SKIPPED
bootlin-armv7-musl [ 8/32]: OK
bootlin-m68k-5208-uclibc [ 9/32]: SKIPPED
bootlin-m68k-68040-uclibc [10/32]: OK
bootlin-microblazeel-uclibc [11/32]: OK
bootlin-mips64el-glibc [12/32]: OK
bootlin-mipsel32r6-glibc [13/32]: OK
bootlin-mipsel-uclibc [14/32]: OK
bootlin-openrisc-uclibc [15/32]: OK
bootlin-powerpc64le-power8-glibc [16/32]: OK
bootlin-powerpc-e500mc-uclibc [17/32]: OK
bootlin-riscv32-glibc [18/32]: OK
bootlin-riscv64-glibc [19/32]: OK
bootlin-riscv64-musl [20/32]: OK
bootlin-s390x-z13-glibc [21/32]: OK
bootlin-sh4-uclibc [22/32]: OK
bootlin-sparc64-glibc [23/32]: OK
bootlin-sparc-uclibc [24/32]: OK
bootlin-x86-64-glibc [25/32]: OK
bootlin-x86-64-musl [26/32]: OK
bootlin-x86-64-uclibc [27/32]: OK
bootlin-x86-i686-musl [28/32]: OK
bootlin-xtensa-uclibc [29/32]: OK
br-arm-basic [30/32]: SKIPPED
br-arm-full-nothread [31/32]: SKIPPED
br-arm-full-static [32/32]: SKIPPED
32 builds, 6 skipped, 0 build failed, 0 legal-info failed, 0 show-info failed
Signed-off-by: Angelo Compagnucci <angelo.compagnucci@gmail.com>
[Bernd:
- bumped to 12.3.2
- raised gcc version to 10
- removed configure option -DSKIP_OS_SECURE_RANDOM=ON as suggested
by Angelo: https://lists.buildroot.org/pipermail/buildroot/2025-August/784962.html
- added build test results]
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Fixes build error with gcc 16.x:
interactive.c: In function 'interactiveShowData':
interactive.c:334:49: error: '%5d' directive output may be truncated
writing between 5 and 11 bytes into a region of size 7
[-Werror=format-truncation=]
334 | snprintf(strFl, sizeof(strFl), "%5d ", convert_altitude(modeC * 100));
| ^~~
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
DTUI is a small TUI for introspecting the state of the system/session
D-Bus. It shows current services and lets you introspect objects and
their interfaces.
This package is built with Rust/Cargo.
Upstream: https://github.com/Troels51/dtui
License: MIT
Signed-off-by: Christopher Obbard <obbardc@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
diff COPYRIGHT:
-Copyright (C) 2005-2025 Mike Pall. All rights reserved.
+Copyright (C) 2005-2026 Mike Pall. All rights reserved.
Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
For change log since version 9.2.9, see:
https://www.swi-prolog.org/ChangeLog?branch=stable&from=9.2.9&to=10.0.2
This new version can no longer build in the source tree. Therefore,
this commit adds SWIPL_SUPPORTS_IN_SOURCE_BUILD = NO. Also, the target
swipl needs the host-swipl binary. It is searched by the cmake script.
This commit changes SWIPL_NATIVE_FRIEND variable to point to the
out-of-source build directory.
swipl has now a stricter ABI check of its prolog library bytecode
compiled files, compared to its builtin expected ABI:
https://github.com/SWI-Prolog/swipl/blob/V10.0.2/src/pl-init.c#L263
When cross-compiling, swipl is built on one side, and library files
are built with the host-swipl. The ABI string of the host-swipl and
target swipl must match. If not matching, swipl fail to start at
runtime with the error:
FATAL: could not find SWI-Prolog home
Tried source: compiled in
Found /usr/lib/swipl: ABI mismatch
This ABI string includes a signature of builtin modules. See:
https://github.com/SWI-Prolog/swipl/blob/V10.0.2/src/pl-init.c#L1653https://github.com/SWI-Prolog/swipl/blob/V10.0.2/src/pl-ext.c#L330
The swipl ncurses support is made with one those builtin module.
ncurses development files are likely present on host and likely not
enabled or needed on target, hence producing a different ABI
signature. For that reason, this commit disables ncurses on both the
host and target.
Signed-off-by: Julien Olivain <ju.o@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Buildroot commit aac3d2b402 added a hidden
boolean option that packages which depend on java on the host can select.
Since buildroot commit 5366b8f734 we can
provide our own host-openjdk-bin package.
Kodi previously used BR2_NEEDS_HOST_JAVA but was switched to host-open-
jdk-bin in this series.
The option BR2_NEEDS_HOST_JAVA is now unused and can be removed.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
This patch removes the dependency on java being installed on the host,
instead we make use of our host-openjdk-bin package.
Build-tested using this defconfig
BR2_x86_64=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_ROOTFS_DEVICE_CREATION_DYNAMIC_EUDEV=y
BR2_PACKAGE_KODI=y
BR2_PACKAGE_MESA3D=y
BR2_PACKAGE_MESA3D_GALLIUM_DRIVER_SOFTPIPE=y
BR2_PACKAGE_MESA3D_OPENGL_EGL=y
BR2_PACKAGE_MESA3D_OPENGL_ES=y
BR2_PACKAGE_PYTHON3=y
BR2_PACKAGE_PYTHON3_PY_ONLY=y
by running 'utils/docker-run make kodi'.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
All patches that were applicable to GDB 16.1 are also applicable to
GDB 17.1. Rebased patch 0005. Added upstream patch 0010 to fix build
error with musl and upstream patch 0011 to fix build error on sparc.
Details about the 17.1 release:
https://sourceware.org/gdb/download/ANNOUNCEMENT
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
Changelog:
* psc_misc: Add function to get PDC firmware version
* ipl: Fix mfg mode on C8000, rp34x0 and rp44x0
This has been tested on hp 712, booting from network and from disk.
Signed-off-by: Vincent Stehlé <vincent.stehle@laposte.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Everyone thinks they like ARM but secretly they want to run
Linux on their Megadrive[0].
Building a working toolchain/userland for nommu m68k is a bit
difficult because of the bitrotten tools all over the place and
one big issue is GCC generates code that uses unaligned
accesses that will cause an exception on 68000 machines if you
aren't careful.
Buildroot can easily create a working toolchain and userland
for nommu m68k so why not?
0 - This isn't a joke, I'm working on this. I have u-boot working
on the Megadrive, Linux is already working on a bunch of 68000
machines and I just need to get everything a little bit smaller
to get into 4MB of RAM.
Signed-off-by: Daniel Palmer <daniel@thingy.jp>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
For release announce on mailing list, see [1].
For release general news, see [2].
This commit removes all package patches as they are all included in
this version, except patch #74 for which an alternate fix is included
in this new version in commit [3].
Tested by compiling and booting the defconfigs using grub2:
qemu_aarch64_sbsa_defconfig
qemu_arm_ebbr_defconfig
qemu_loongarch64_virt_efi_defconfig
qemu_riscv64_virt_efi_defconfig
pc_x86_64_bios_defconfig
pc_x86_64_efi_defconfig
and running the tests:
support/testing/run-tests \
-d dl -o output_folder \
tests.boot.test_grub \
tests.fs.test_iso9660
[1] https://lists.gnu.org/archive/html/grub-devel/2026-01/msg00029.html
[2] https://gitlab.freedesktop.org/gnu-grub/grub/-/blob/grub-2.14/NEWS
[3] 0739d24cd1
Signed-off-by: Julien Olivain <ju.o@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
