Enable building of cutekeyboard with Qt6. With the merge of
QuickControls2 into qtdeclarative, the corresponding Kconfig option has
also disappeared in Buildroot.
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Reviewed-by: Andrea Ricchi <andrea.ricchi@amarulasolutions.com>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
The qmake infrastructure hardcodes qt5base as the Qt provider. Select
the Qt major version from BR2_PACKAGE_QT5, which is mutually exclusive
with BR2_PACKAGE_QT6, and use it to pick the qt<N>base dependency, the
QT<N>_QMAKE command and the QT<N>_QT_CONF_FIXUP hook.
The header sync hook now takes the version from QT<N>_VERSION_MAJOR and
QT<N>_VERSION instead of hardcoding 5.15.
Document in the manual which Qt base package is added to the dependencies.
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
[Arnout: write pkg-qmake with explicitly QT_XXX helper variables]
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
With per-package directories, qt.conf is generated at qt6base install
time and holds the paths of the qt6base per-package host directory.
Add a hook to regenerate it with the paths of the package being built,
as QT5_QT_CONF_FIXUP does for Qt5. The qmake infrastructure will use it.
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
The qmake functionality had been missing due to the new host/target
build requirement, to get it working the main parts of the qt5 qmake
has been kept, we're still making our own device spec which we use to
run qmake with along with passing some additional device options for our
flags/cross-compiler.
qt6 has its own cross-compiling qmake wrapper script. However, that
hardcodes the paths to different places than where we have them.
QT6_INSTALL_QT_CONF will be used by the qmake-package infrastructure
later so it is put in qt6.mk, even though currently it's only used by
qt6base.
A big thanks to Angelo Compagnucci for the encouragement and testing.
[dario: fix HostData and add the target install dirs in qt.conf.in,
quote QT_QMAKE_DEVICE_OPTIONS, use TARGET_CFLAGS/TARGET_CXXFLAGS,
change the commit message]
Signed-off-by: Jesse Van Gavere <jesseevg@gmail.com>
Tested-by: Angelo Compagnucci <angelo@amarulasolutions.com>
Tested-by: Richard GENOUD <richard.genoud@bootlin.com>
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
lldpd bundles some uapi linux files for old distributions, especially
the types.h one.
Since kernel commit d4bf9e08d412 ("arm64: uapi: Use __u128 instead of
__uint128_t in UAPI headers"), the type __u128 is used by sigcontext.h.
This type is defined in types.h, but the old version bundled by lldpd
does not provide it. This lack leads to the following compilation error:
In file included from ../../../../host/aarch64-buildroot-linux-gnu/sysroot/usr/include/bits/sigcontext.h:30,
from ../../../../host/aarch64-buildroot-linux-gnu/sysroot/usr/include/signal.h:301,
from ../../../../host/aarch64-buildroot-linux-gnu/sysroot/usr/include/sys/param.h:28,
from privsep_io.c:5:
../../../../host/aarch64-buildroot-linux-gnu/sysroot/usr/include/asm/sigcontext.h:81:9: error: unknown type name '__u128'; did you mean '__u32'?
81 | __u128 vregs[32];
| ^~~~~~
| __u32
The uapi linux files bundled by lldpd are the ones of kernel 4.19
(https://github.com/lldpd/lldpd/commit/a192a2e91f81), and they are
prepended to the include path, so they shadow the sysroot ones.
Remove them so that lldpd uses the uapi files of the sysroot, but only
when the toolchain provides kernel headers 4.19 or later. With older
headers the bundled files are kept, so those configurations are left
unchanged.
Signed-off-by: Thibaut Collet <thbt.cllt@gmail.com>
Reviewed-by: Vincent Jardin <vjardin@free.fr> (✓ DKIM/buildroot.org)
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
WK_USE_CCACHE=NO is only passed to the configure step
(WPEWEBKIT_CONF_ENV). During the build step, ninja can run cmake again
(e.g. when a cmake input such as toolchainfile.cmake changed), and that
run does not have WK_USE_CCACHE set. WebKit's WebKitCCache.cmake then
sets ccache as compiler launcher again.
That outer ccache runs without the CCACHE_BASEDIR that the toolchain
wrapper sets for its own ccache call, so the cache keys and manifests
contain absolute paths, and the entries only match in the same checkout
directory.
Pass WK_USE_CCACHE=NO to the build step too.
Signed-off-by: Thomas Devoogdt <thomas.devoogdt@barco.com>
Acked-by: Adrian Perez de Castro <aperez@igalia.com>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
WK_USE_CCACHE=NO is only passed to the configure step
(WEBKITGTK_CONF_ENV). During the build step, ninja can run cmake again
(e.g. when a cmake input such as toolchainfile.cmake changed), and that
run does not have WK_USE_CCACHE set. WebKit's WebKitCCache.cmake then
sets ccache as compiler launcher again.
That outer ccache runs without the CCACHE_BASEDIR that the toolchain
wrapper sets for its own ccache call, so the cache keys and manifests
contain absolute paths, and the entries only match in the same checkout
directory.
Pass WK_USE_CCACHE=NO to the build step too.
Signed-off-by: Thomas Devoogdt <thomas.devoogdt@barco.com>
Acked-by: Adrian Perez de Castro <aperez@igalia.com>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
Since OpenSSL 3.5, commit fed9be39ffec ("Make RFC8422 deprecated TLS EC
curves disablable"), OpenSSL has the option to disable support
for legacy TLS EC groups that were deprecated in RFC8422.
In OpenSSL 4.0, the option will be removed again and instead there will
be an option to _enable_ the legacy curves.
Add a Kconfig option to enable the deprecated curves. Use positive logic
already, i.e. the Kconfig option enables them.
Signed-off-by: Konrad Vité <konrad.vite@cherry.de>
Acked-by: Quentin Schulz <quentin.schulz@cherry.de>
[Arnout: clarify that enabling this option is dangerous]
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
The new config deviates from the existing
configs/raspberrypicm4io_64_defconfig only as far as necessary to
demonstrate booting through U-Boot instead of having the firmware load
the kernel directly. The sample boot script loads the kernel from the
rootfs and uses the device tree provided by the firmware, and should
be compatible with any RPi device that runs U-Boot.
Signed-off-by: Fiona Klute (WIWA) <fiona.klute@gmx.de>
[Arnout: update kernel to same as raspberrypicm4io_64_defconfig]
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
When installing DTBs from the rpi-firmware package
(BR2_PACKAGE_RPI_FIRMWARE_INSTALL_DTBS=y, the default), there's no DTB
directly in the images directory because the kernel build doesn't
provide one. The DTBs from rpi-firmware are installed in the
rpi-firmware directory.
Using pre-compiled DTBs is discouraged, but useful for using mainline
kernels until suitable DTS can be merged, or to provide U-Boot with a
DTB (via the firmware bootloader) independent of whatever kernel DTB
is in use.
Signed-off-by: Fiona Klute (WIWA) <fiona.klute@gmx.de>
Reviewed-by: Gaël PORTAY <gael.portay+rtone@gmail.com>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
If using U-Boot as "kernel" for the raspberrypi bootloader, include
boot.scr in the boot partition if it exists.
To use this, use BR2_PACKAGE_RPI_FIRMWARE_CONFIG_FILE to set a config
file that uses U-Boot, and enable building a boot script in the U-Boot
host tool settings.
Signed-off-by: Fiona Klute (WIWA) <fiona.klute@gmx.de>
Reviewed-by: Yann E. MORIN <yann.morin.1998@free.fr>
Tested-by: Gaël PORTAY <gael.portay+rtone@gmail.com>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
Passing linker flags via LDFLAGS is notoriously fragile: packages either
use LDFLAGS with the gcc frontend, or directly with ld. This means that
care must be taken to only pass flags that are recognised by both gcc
and ld, which usually proves a bit challenging, as bd035872be
(package/Makefile.in: Use "-z text" instead of "-Wl, -z, text")
demonstrates (later amended by 3a39e706a9).
It turns out that we already have a better, robust way of passing
LDFLAGS via the gcc frontend: the toolchain wrapper, which we already use
to pass relro for example.
We can revert back to using -Wl,-z,text, as we're sure this is going to
be a call to the gcc frontend (the wrapper is not used for ld), which
makes it consistent with the other similar args for -z relro and -z now.
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Cc: J. Neuschäfer <j.neuschaefer@gmx.net>
Cc: Romain Naour <romain.naour@gmail.com>
Cc: Giulio Benetti <giulio.benetti@benettiengineering.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
ifenslave has not received any recent update, has no Buildroot DEVELOPER
and its upstream NEWS [1] states the following:
This version of the ifenslave package no longer provides /sbin/ifenslave. The
/sbin/ip command from the iproute2 package supports creating bonding masters
and enslaving other interfaces to it.
Its feature is available either in iproute or in Busybox, so we can
simply remove it from Buildroot.
[1] https://salsa.debian.org/debian/ifenslave/-/blob/main/debian/NEWS
Signed-off-by: Titouan Christophe <titouan.christophe@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
If BR2_PACKAGE_CA_CERTIFICATES is present, add the ca-certificates.crt
to the php.ini, to not get SSL errors when using PHP.
Signed-off-by: Julian Vetter <julian@outer-limits.org>
Reviewed-by: Thomas Perale <thomas.perale@mind.be>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
The ICU library requires C++ standard library when linking statically.
Without this, applications using pkg-config --libs --static icu-uc
(e.g. xmlcatalog in libxml2) will fail to link with undefined reference
errors to C++ symbols.
Extend the existing fix for static linking to also add -lstdc++ to the
Libs field, not only Libs.private, when building in static-only mode
(ENABLE_SHARED is not set).
Fixes: https://autobuild.buildroot.org/results/80c/80ce2a1b8ec356dc13e3c51474fe4cac9cbe3cd6
Signed-off-by: Shubham Chakraborty <chakrabortyshubham66@gmail.com>
Acked-by: Andreas Ziegler <br025@umbiko.net>
[Arnout:
- Remove checkpackageignore entry.
- Improve patch comment.]
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
Starting from v2.40, binutils provides libsframe, a library for
manipulating the SFRAME debug format. Buildroot installs libsframe
to the target image only if BR2_PACKAGE_BINUTILS_TARGET option is
enabled. However other dynamic libraries from binutils are linked
against libsframe, including libbfd and libopcodes. Any binary
linked dynamically against libbfd, e.g. perf, will need libsframe
as well. So libsframe should always be installed the target together
with other binutils libraries such as libbfd and libopcodes.
Signed-off-by: Sergey Matyukevich <geomatsi@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
To better track the version of a package before a Buildroot LTS release
is made I'm adding myself to DEVELOPERS file for those one so I receive
alerts for this subset of packages.
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
The BR2_EXTERNAL environment variable is passed into the container,
and each path listed in it mounted. This allows using external trees
when running a build using utils/docker-run. Testing the existence of
the variable instead of a non-empty value allows passing an empty
BR2_EXTERNAL variable to disable currently set external trees.
The --mount option needs an absolute path for the destination. Make the
paths absolute the same way br2-external itself does it. Do not use
realpath since that resolves symlinks and would make it different from
what is in the BR2_EXTERNAL variable.
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
Give both boards a recognizable hostname instead of the generic
"buildroot". The name is used for the login banner and sent as the
DHCP hostname (option 12), so the board is easy to identify on the
network.
Signed-off-by: Vladimir Savelev <anykey196@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Add new KConfig symbols for all actively supported longterm versions of the
Linux kernel. This makes it easier for individual configs to follow updates
of their kernels, without having to frequently update them. This should
generally improve the security of the target systems, which receive frequent
security updates while preserving a stable feature set.
Signed-off-by: Titouan Christophe <titouan.christophe@mind.be>
[Peter: move BR2_LINUX_KERNEL_LATEST_SERIES outside the choice to make
utils/check-symbols happy, add version symlinks]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Since Buildroot commit 40e266ccb9 [1], host-uboot-tools was always selected
for the latest Linux kernel. This change is intended for autobuilders, and
was introduced in Buildroot while the latest kernel was 6.12.
However, further investigation seem to indicate that itb images were
generated for mips for a much longer time. Linux commit eed0eabd12 [2]
seem to enable that image format, and was released in Linux 4.9.
This version being EOL for quite some time, we can therefore assume
any "not so old" Linux for mips will need host-uboot-tools to finalize
this image, and therefore we should select it when building Linux.
[1] linux: select host-uboot-tools with FIT on mips
https://gitlab.com/buildroot.org/buildroot/-/commit/40e266ccb9
[2] MIPS: generic: Introduce generic DT-based board support
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=eed0eabd12ef061821cbfa20d903476e07645320
Signed-off-by: Titouan Christophe <titouan.christophe@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The mail address used to contribute via buildroot mailing list
is required to be subscribed to the mailing list. Otherwise,
messages sent to the mailing list will be automatically rejected,
as server only accepts email from mailing list subscribers.
This behavior is not explicitly stated in the documentation
for contributors (aside from recommendation to subscribe
as the first thing for contributor to do). This may lead to an
unpleasant surprise to developers, who prefer reading mailing
lists using archives.
Add mention of mailing list subscription requirement to 'Preparing
a patch series' section of contribution manual.
Signed-off-by: Nikita Kiryushin <kiryushin@ancud.ru>
The commit [1] introduced the `cdx_spec_version` variable.
The reason for that variable is not immediately clear. Only the license
schema can use patch revision while the spec version should always
exclude the patch revision.
This commit adds a comment clarifying this.
[1] 504bb61bc6 utils/generate-cyclonedx: support custom schema version
Suggested-by: Quentin Schulz <quentin.schulz@cherry.de>
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
The argparse library documentation [1] suggest the usage of
'ArgumentDefaultsHelpFormatter' in the ArgumentParser declaration to
show the default value of every arguments.
Use this to avoid declaring this value ourself which possibly gets
outdated with the actual default value.
Note that this also adds an unnecessary default: FALSE to the --virtual
option, but that's a minor annoyance.
[1] https://docs.python.org/3/library/argparse.html#argparse.ArgumentDefaultsHelpFormatter
Suggested-by: Quentin Schulz <quentin.schulz@cherry.de>
Suggested-by: Titouan Christophe <titouan.christophe@mind.be>
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
Add cramjam, a python bindings to de/compression algorithms.
It is needed by python-snappy package in order to bump the version.
Signed-off-by: Flaviu Nistor <flaviu.nistor@gmail.com>
[Peter: add BR2_PACKAGE_HOST_RUSTC_TARGET_ARCH_SUPPORTS dependency, drop
upstream tarball hash.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
get_warnings_for_symbols_without_definition() skips any symbol
matching BR2_PACKAGE_HOST_* or BR2_PACKAGE_HAS_HOST_*, via
br.re_host_symbol. This blanket exclusion was needed because
we don't require a Config.in.host to be present for host packages,
so BR2_PACKAGE_HOST_FOO may in fact be undefined and that's
perfectly fine.
However, this also excludes any other options that should be defined in
the Config.in.host, e.g. a BR2_PACKAGE_HOST_FOO_BAR suboption. If such
an option is misspelled in the .mk file, we will not notice.
We already treat the BR2_PACKAGE_* symbols specially. They are
considered as "used" even if there is no direct reference to them in the
.mk file. This is done in the handle_eval function. Depending on the
package infra and the type of package (bootloader, toolchain, ...) it
sets the corresponding kconfig symbol as "used".
At that point, we can also handle the special case of the host kconfig
symbols. Instead of a blanket exclusion, we say that the
BR2_PACKAGE_HOST_FOO option is implicitly defined by the eval statement.
We only add BR2_PACKAGE_HOST_FOO as implicitly defined, so any
suboptions will still be checked by
get_warnings_for_symbols_without_definition().
We could alternatively also have skipped the "usage" marking of host
packages. This approach felt more consistent though.
Make handle_eval() in checksymbolslib/makefile.py record the
implicitly defined symbols directly, at the $(eval $(host-...))
line that creates them:
- BR2_PACKAGE_HOST_<PKG> as a definition, for host-*-package and
host-virtual-package;
- BR2_PACKAGE_HAS_HOST_<PKG> as virtual, for host-virtual-package,
mirroring how the plain virtual-package case already registers
BR2_PACKAGE_HAS_<PKG>.
This makes br.re_host_symbol and its use in
get_warnings_for_symbols_without_definition() unnecessary, so drop
them. The check now flags a real symbol name once it lacks both an
explicit and an implicit definition, instead of ignoring the whole
BR2_PACKAGE_HOST_*/BR2_PACKAGE_HAS_HOST_* namespace.
Assisted-by: LLM
Signed-off-by: Raphaël Mélotte <raphael.melotte@mind.be>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
The "host-qemu needs a host gcc >= 8" comment depends on
BR2_PACKAGE_HOST_QEMU_USER_ARCH_SUPPORT, missing the trailing "S"
of the actual symbol BR2_PACKAGE_HOST_QEMU_USER_ARCH_SUPPORTS,
defined a few lines above and used correctly everywhere else in
this file.
Because this typo'd symbol is never defined anywhere, it is always
considered false by Kconfig, so the comment's "depends on" clause
silently loses one of its two intended conditions.
Assisted-by: LLM
Signed-off-by: Raphaël Mélotte <raphael.melotte@mind.be>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
The ability to check with `make check-package` the 'IGNORE_CVES'
variable content reference an existing patches was added in commit [1].
This commit extend it to actually check that the entries that are
referenced in the 'IGNORE_CVES' variable are matching the trailers in
the referenced patch.
With this change the following `foo.mk` content:
```
FOO_IGNORE_CVES += CVE-2000-1234
```
Running `make check-package` will output an error if the content of the
patch is the following:
```
CVE: CVE-2001-1234
```
[1] b00ac4e346 utils/checkpackagelib: add new check MissingCVEPatch
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
[Arnout: adapt to the trailers to using regex]
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
Currently the MissingCVEPatch check will complain if another
identifiers than a CVE-YYYY-XXXX is passed after the 'CVE: '
trailer in the patch header introduced in [1].
If a package 'foo' defines the following:
FOO_IGNORE_CVES += GHSA-abcd-1234-efgh
And correctly references it in a patch header:
CVE: GHSA-abcd-1234-efgh
check-package will report the following:
package/foo/foo.mk: patch file '0001-some-patch.patch' is missing 'CVE:' tag
While we could argue that GHSA identifier should not be present in the
'IGNORE_CVES' variable, it's still valuable for users to have a way to
correctly ignore and link a patch for their vulnerability report.
From our point of view it allows us to track GHSA vulnerabilities that
could later be assigned a CVE so we can update the IGNORE_CVES variable.
This commit relaxes the constraint on the 'CVE' trailer to just check
for the presence of any identifier. We still only allow alphanumerical
characters and dashes, to be sure that any tooling that uses the
identifiers won't break.
[1] 1167d0ff3d docs/manual: mention CVE trailer
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
[Arnout:
- keep the regex but allow any alphanumerical string;
- add a test that FOO:123 doesn't match]
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
The `self.pending_patch` is set but unused.
Fixes: b00ac4e346 utils/checkpackagelib: add new check MissingCVEPatch
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
szip is not maintained upstream anymore: 2.1.1 is the last release,
there is no public repository, and the source is only available from
the HDF Group archive. Its license is not a standard open source one,
and the HDF Group itself recommends libaec as a replacement.
The package also fails to build with GCC >= 15 on non-glibc
toolchains, as only linux-gnu hosts get -ansi from szip's configure
and the code relies on pre-C23 empty parameter lists:
rice.c:1323:15: error: too many arguments to function 'winner_ref_function'; expected 0, have 2
libaec provides libsz, a drop-in replacement with the same API and
soname, so select it through the legacy option.
Fixes:
https://autobuild.buildroot.org/results/5e400d0e13480c58e8b1b0178bf7e5537dd31751
Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>
Assisted-by: Claude:claude-opus-5-5
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
libaec provides fast lossless compression of 1 up to 32 bit wide
signed or unsigned integers. It implements the Golomb-Rice compression
method under the BSD license and includes libsz, a drop-in replacement
for the SZIP library with the same API (szlib.h) and soname
(libsz.so.2).
https://github.com/Deutsches-Klimarechenzentrum/libaec
Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>
Assisted-by: Claude:claude-opus-5-5
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
