BR2_PACKAGE_LIBSSH_OPENSSL unconditionnally selects
BR2_PACKAGE_LIBOPENSSL_ENGINES even though libressl is also supported
as an OpenSSL provider (and BR2_PACKAGE_LIBOPENSSL_ENGINES doesn't
make sense for libressl).
This causes the following Kconfig warning:
WARNING: unmet direct dependencies detected for BR2_PACKAGE_LIBOPENSSL_ENGINES
Depends on [n]: <choice> && BR2_PACKAGE_LIBOPENSSL [=n]
Selected by [y]:
- BR2_PACKAGE_LIBSSH_OPENSSL [=y] && <choice> && BR2_PACKAGE_OPENSSL [=y]
We checked that libssh, with OpenSSL support and libressl selected as
an OpenSSL provider works fine, using the following defconfig:
BR2_aarch64=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_TOOLCHAIN_EXTERNAL_BOOTLIN=y
BR2_TOOLCHAIN_EXTERNAL_BOOTLIN_AARCH64_GLIBC_STABLE=y
BR2_PACKAGE_LIBSSH=y
BR2_PACKAGE_LIBSSH_SERVER=y
BR2_PACKAGE_LIBRESSL=y
Fixes: 62103be918 ("package/libssh: select BR2_PACKAGE_LIBOPENSSL_ENGINES")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
In commit
76241e89e1 ("package/libabseil-cpp: bump
to version 20260817.0"), libabseil-cpp was bumped, which required the
gcc >= 8.x dependency to be upgraded to a gcc >= 10.x dependency. This
was properly done in package/protobuf as part of this commit, as
protobuf is a reverse dependency of libabseil-cpp.
However, usbguard, which is a reverse dependency of protobuf, was
forgotten, and it no longer carries the correct gcc dependency.
This commit fixes this issue.
Fixes: 76241e89e1 ("package/libabseil-cpp: bump to version 20260817.0")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
BR2_PACKAGE_QT5CINEX selects BR2_PACKAGE_QT5BASE_PNG,
BR2_PACKAGE_QT5BASE_WIDGETS and BR2_PACKAGE_QT5BASE_EGLFS, which are
all sub-options of BR2_PACKAGE_QT5BASE_GUI, but we don't explicitly
selects BR2_PACKAGE_QT5BASE_GUI.
It turns out that things work because the package selects
BR2_PACKAGE_QT5GRAPHICALEFFECTS, which selects
BR2_PACKAGE_QT5DECLARATIVE_QUICK, which selects
BR2_PACKAGE_QT5BASE_GUI, but that is rather non-obvious, and it makes
more sense for BR2_PACKAGE_QT5CINEX to directly select
BR2_PACKAGE_QT5BASE_GUI if it also selects sub-options of it.
No functional change.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
- BR2_PACKAGE_IVI_HOMESCREEN_AUDIO_PLAYERS selects gstreamer1, which
has a depends on BR2_USE_MMU, but does not propagate it
- BR2_PACKAGE_IVI_HOMESCREEN_FLUTTER_SECURE_STORAGE_PLUGIN selects
libsecret, which has a depends on BR2_USE_MMU, but does not propagate
it
In practice there is no problem since ivi-homescreen depends on glibc,
and glibc doesn't support any noMMU architecture. But just by walking
the chain of option dependencies, this is not something that is
theoretically guaranteed (making automated verification of
dependencies difficult).
The other "depends on" from gstreamer1 and libsecret, BR2_USE_WCHAR
and BR2_TOOLCHAIN_HAS_THREADS are on the other hand already handled by
the top-level BR2_PACKAGE_IVI_HOMESCREEN option, so there is no
ambiguity.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
BR2_PACKAGE_PULSEVIEW selects BR2_PACKAGE_QT5BASE_PNG and
BR2_PACKAGE_QT5BASE_WIDGETS, which both depend on
BR2_PACKAGE_QT5BASE_GUI. It ends working because we also select
BR2_PACKAGE_QT5SVG, which selects BR2_PACKAGE_QT5BASE_GUI, so there is
no bug, but it's bit inconsistent to select sub-options that have a
"depends on" without selecting the option they depend on.
This not a bug fix, it has no functional implication.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
pocketpy enables thread support by default (PK_ENABLE_THREADS=ON) and
then requires Threads from cmake, which fails on toolchains without
thread support:
CMake Error at /usr/share/cmake-3.28/Modules/FindPackageHandleStandardArgs.cmake:230 (message):
Could NOT find Threads (missing: Threads_FOUND)
Thread support is optional, so enable it only when the toolchain
provides threads.
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
The definition of BR2_PACKAGE_KODI_ARCH_SUPPORTS is incorrect, it
goes like this:
bool
default y if BR2_PACKAGE_FFMPEG_ARCH_SUPPORTS
default y if BR2_PACKAGE_HOST_OPENJDK_BIN_ARCH_SUPPORTS
so it means it would be "y" if either
BR2_PACKAGE_FFMPEG_ARCH_SUPPORTS *OR*
BR2_PACKAGE_HOST_OPENJDK_BIN_ARCH_SUPPORTS is true. While clearly what
we need is for both to be true: ffmpeg should be available for the
target architecture, and openjdk should be available for the host
architecture.
One option was to change to:
bool
default y if BR2_PACKAGE_FFMPEG_ARCH_SUPPORTS && BR2_PACKAGE_HOST_OPENJDK_BIN_ARCH_SUPPORTS
Or:
bool
default y if BR2_PACKAGE_FFMPEG_ARCH_SUPPORTS
depends on BR2_PACKAGE_HOST_OPENJDK_BIN_ARCH_SUPPORTS
But we preferred:
bool
default y
depends on BR2_PACKAGE_FFMPEG_ARCH_SUPPORTS
depends on BR2_PACKAGE_HOST_OPENJDK_BIN_ARCH_SUPPORTS
Fixes: b6a2f49429 ("package/kodi: depend on host-openjdk-bin instead of selecting BR2_NEEDS_HOST_JAVA")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Even though kodi itself has architecture dependencies (expressed
through BR2_PACKAGE_KODI_ARCH_SUPPORTS, the option
BR2_PACKAGE_KODI_MYSQL selects BR2_PACKAGE_MARIADB, which has its own
architecture dependencies as well. Make sure to propagate those to
BR2_PACKAGE_KODI_MYSQL, which doesn't require adding a Config.in
comment as these are purely architecture dependencies.
We haven't replicate all dependencies of BR2_PACKAGE_MARIADB because
all the others are covered by the top-level BR2_PACKAGE_KODI, and
propagating them would require adding a Config.in comment for
BR2_PACKAGE_KODI_MYSQL.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Since hidapi was introduced in commit
6267f34afd, it forgot to propagate some
dependencies of libgudev (which existed back then). Initially libgudev
was only needed when BR2_INIT_SYSTEMD=y, but still the dependencies
were not propagated for the systemd case.
Anyway, since e739dd5a11, libgudev is a
mandatory dependency of hidapi, independently from the selected init
system.
We make sure to propagate all dependencies of libgudev to hidapi, and
propagate them to the reverse dependencies of hidapi.
Fixes: 6267f34afd ("hidapi: new package")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
The backported fix is dropped, as 3.4 ships it:
0001-Fix-163-unterminated-username-used-with-getpwnam.patch
-> upstream commit d73777c2c356, released in 3.4
With the patch gone, LIBCONFUSE_IGNORE_CVES is no longer needed either.
3.4 also fixes three robustness defects that carry no CVE:
#180 isspace() argument fix, could crash the lexer
#182 stack exhaustion from deeply nested sections
#187 null dereference on an empty comment with CFGF_COMMENTS
Upstream release notes:
https://github.com/libconfuse/libconfuse/releases/tag/v3.4
Signed-off-by: Michael Fischer <mf@go-sys.de>
[Fiona: add link to upstream release notes]
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
The ARMv7 EABIhf toolchains can currently be selected with ARMv8 cores
selected, even when EABI is used due to how the condition is
constructed. This obviously fails as those toolchains are EABIhf,
causing the following build issue:
Incorrect ABI setting: EABI selected, but toolchain is incompatible
This is for example what happens with:
BR2_arm=y
BR2_cortex_a32=y
BR2_ARM_EABI=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_TOOLCHAIN_EXTERNAL_BOOTLIN_ARMV7_EABIHF_GLIBC_BLEEDING_EDGE=y
In order to address this, we adjust the script generating the Bootlin
toolchain package so that EABIhf is required for both ARMv7 and
ARMv8.
Please note that we already require BR2_arm for those toolchains, so
we are *only* talking about ARMv8 cores being used in 32-bit mode.
This will be needed to fix:
https://autobuild.buildroot.org/results/3ce1dbd480c71b782ef722c39034cb039a036523/
Reported-by: Julien Olivain <ju.o@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
BR2_PACKAGE_ARGP_STANDALONE is defined as follows:
config BR2_PACKAGE_ARGP_STANDALONE
depends on !BR2_TOOLCHAIN_USES_GLIBC
Some packages did:
select BR2_PACKAGE_ARGP_STANDALONE if !BR2_TOOLCHAIN_USES_GLIBC
while a number of others did:
select BR2_PACKAGE_ARGP_STANDALONE if BR2_TOOLCHAIN_USES_UCLIBC || BR2_TOOLCHAIN_USES_MUSL
This commit harmonizes the situation, by settling on the first
solution ("if !BR2_TOOLCHAIN_USES_GLIBC") as it matches how
BR2_PACKAGE_ARGP_STANDALONE is defined in the first place.
No functional change.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
BR2_PACKAGE_INTEL_VPL_GPU_RT selects BR2_PACKAGE_INTEL_MEDIADRIVER but
did not propagate "depends on BR2_TOOLCHAIN_GCC_AT_LEAST_8". This
commit fixes this issue, which was introduced in commit
ac65841def, when onevpl-intel-gpu was
introduced (it was later renamed to intel-vpl-gpu-rt).
Fixes: ac65841def ("package/onevpl-intel-gpu: new package")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
BR2_PACKAGE_INTEL_MEDIASDK selects BR2_PACKAGE_INTEL_MEDIADRIVER but
forgets to propagate the "depends on BR2_TOOLCHAIN_GCC_AT_LEAST_8".
This issue was introduced in commit
51b60c8acf, when "depends on
BR2_TOOLCHAIN_GCC_AT_LEAST_8" was added to mesa3d, propagated to
intel-mediadriver, but not intel-mediasdk.
Fixes: 51b60c8acf ("package/mesa3d: needs gcc >= 8")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
BR2_PACKAGE_PYTHON_MEMRAY selects BR2_PACKAGE_LIBUNWIND but forgot to
propagate "depends on BR2_TOOLCHAIN_GCC_AT_LEAST_4_9".
Fixes: c2df8bab97 ("package/python-memray: new package")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
BR2_PACKAGE_PYTHON_GRPCIO_REFLECTION selects
BR2_PACKAGE_PYTHON_PROTOBUF, but forgot to replicate "depends on
BR2_PACKAGE_HOST_PROTOBUF_ARCH_SUPPORTS".
Fixes: 3217fedcb8 ("package/python-grpcio-reflection: new package")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
BR2_PACKAGE_PYTHON_GOOGLEAPIS_COMMON_PROTOS selects
BR2_PACKAGE_PYTHON_PROTOBUF but did not propagate
BR2_PACKAGE_HOST_PROTOBUF_ARCH_SUPPORTS.
Fixes: d37766a886 ("package/python-googleapis-common-protos: new package")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Commit 66ddec89e8 ("package/udisks: bump
to version 2.92") mistakenly removed the BR2_USE_MMU dependency of
udisks when dropping "select BR2_PACKAGE_LVM2". Indeed, BR2_USE_MMU is
a dependency of many other packages selected by udisks.
Interestingly, the Config.in comments in the same file still had the
"depends on BR2_USE_MMU" dependencies.
Fixes: 66ddec89e8 ("package/udisks: bump to version 2.92")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Since bcc was introduced in commit
146498d13c, it lacked a dependency
propagation from clang for BR2_TOOLCHAIN_HAS_GCC_BUG_64735, this
commit fixes this mistake.
Fixes: 146498d13c ("package/bcc: new package")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
samba4 is available on !musl, but samba support in mpd is available
only with glibc. Turns out that samba support in mpd builds just fine
with uClibc-ng, and that samba4 no longer needs native RPC support: it
can use libtirpc when needed (it's handled in the samba4 package
itself).
Tested with the following defconfig:
BR2_aarch64=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_TOOLCHAIN_EXTERNAL_BOOTLIN=y
BR2_TOOLCHAIN_EXTERNAL_BOOTLIN_AARCH64_UCLIBC_BLEEDING_EDGE=y
BR2_PACKAGE_MPD=y
BR2_PACKAGE_MPD_LIBSMBCLIENT=y
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
BR2_PACKAGE_HOST_GO_TARGET_ARCH_SUPPORTS redefines the conditions to
determine if a host go compiler is available for the current host
architecture. Instead, make it explicit that those conditions are the
same by re-using BR2_PACKAGE_HOST_GO_HOST_ARCH_SUPPORTS.
No functional change.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
In commit
fef9cad1fe ("package/libxmlsec1: bump
version to 1.3.12"), libxmlsec1 was bumped, and alongside some
additional "depends on" were added.
However, these new "depends on" were not propagated to reverse
dependencies of libxmlsec1, i.e. openscap, causing Kconfig warnings:
WARNING: unmet direct dependencies detected for BR2_PACKAGE_LIBXMLSEC1
Depends on [n]: BR2_TOOLCHAIN_GCC_AT_LEAST_7 [=n] && BR2_TOOLCHAIN_HAS_ATOMIC [=n]
Selected by [y]:
- BR2_PACKAGE_OPENSCAP [=y] && BR2_PACKAGE_LIBGPG_ERROR_ARCH_SUPPORTS [=y] && !BR2_STATIC_LIBS [=n] && BR2_TOOLCHAIN_HAS_THREADS_NPTL [=y]
and potentially some build issues, even though we didn't check in the
autobuilders for potential failures.
This commit fixes that by properly propagating the new dependencies.
Cc: Alexis Lothoré <alexis.lothore@bootlin.com>
Cc: Julien Olivain <ju.o@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Acked-by: Alexis Lothoré <alexis.lothore@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Rebased patch 0001 and added Upstream: tag.
Removed patches which are included in this release.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
The demo programs, which have always been built and are still enabled by
default, gained a zlib dependency in imlib2 1.12.3: upstream commit
f8a451043871 ("imlib2_load: Add crc32 printout") started using zlib's
crc32() in imlib2_load, and 31006b425e11 ("imlib2_view: Optionally show
crc32 of image data") did the same for imlib2_view. Both hardcoded -lz.
Upstream commit b9555030dace ("autofoo: don't hardcode zlib flags"),
first released in 1.12.4, replaced -lz with $(ZLIB_LIBS) and added an
unconditional PKG_CHECK_MODULES(ZLIB, zlib) to the demo programs branch
of configure, turning the previously silent link-time requirement into a
configure failure:
checking for zlib... no
configure: error: Package requirements (zlib) were not met:
Package 'zlib' not found
As Buildroot went straight from 1.7.3 to 1.12.5 the intermediate state
was never packaged, but the dependency has in fact been missing since the
crc32 support landed.
Fixes: https://autobuild.buildroot.org/results/4e05404c353ef985d74757d0b1ec3eda2cdff523/
Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
https://github.com/softhsm/SoftHSMv2/releases/tag/2.7.0
Switched repo to new standalone repo:
2355064ce4
Also update the package home page in Config.in.
We need to use the github helper to download the code which also needs
autoreconf.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
[Julien: update the package home page in Config.in]
Signed-off-by: Julien Olivain <ju.o@free.fr>
https://github.com/apache/thrift/blob/v0.24.0/CHANGES.md
Please note that this bump includes CVE-2026-41608:
https://lists.apache.org/thread/vwsbcwqdpwdtp8qkjo11ol6rodbfm21f
which fixes a security bug in the python bindings that are not used by
buildroot.
Building this defconfig
BR2_aarch64=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_TOOLCHAIN_EXTERNAL_CUSTOM=y
BR2_TOOLCHAIN_EXTERNAL_DOWNLOAD=y
BR2_TOOLCHAIN_EXTERNAL_URL="http://toolchains.bootlin.com/downloads/releases/toolchains/aarch64--glibc--bleeding-edge-2017.05-toolchains-1-2.tar.bz2"
BR2_TOOLCHAIN_EXTERNAL_GCC_6=y
BR2_TOOLCHAIN_EXTERNAL_HEADERS_4_9=y
BR2_TOOLCHAIN_EXTERNAL_CUSTOM_GLIBC=y
BR2_TOOLCHAIN_EXTERNAL_CXX=y
BR2_PACKAGE_THRIFT=y
without this bump does not cause build errors.
The Gitlab pipelines detected a build error for this bump with the
defconfig bootlin-aarch64-glibc-old:
/builds/bkuhls/buildroot/br-test-pkg/bootlin-aarch64-glibc-old/build/thrift-0.24.0/lib/cpp/src/thrift/transport/TBufferTransports.h:110:32:
error: ‘ptrdiff_t’ does not name a type
To fix the problem we add a patch to include cstddef.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
