There are a small number of these that are out of place, so re-sort the
entire list to clean it up.
(From OE-Core rev: 76d681c4b17c9169238c571bae7e25df2040960c)
Signed-off-by: Trevor Gamblin <tgamblin@baylibre.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Add existing recipe from meta-python, since pytest lists it as a
dependency. This also puts it in maintainers.inc with myself as
maintainer.
[RP: Switch to use hatching directly]
(From OE-Core rev: 8c06239fdcf983dac759c7d80d6326807c993622)
Signed-off-by: Trevor Gamblin <tgamblin@baylibre.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Add a basic test for the toolchain switching code: set the
toolchain to GCC by default but Clang for a specific recipe, and verify
that two recipes are built with the expected compiler.
This works because before we strip the installed binaries there is a
.comment segment that contains the list of toolchains used.
(From OE-Core rev: 7988c32191927f1c6db422c959eab7a03dfeda04)
Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Discovered when looking for users of get_bb_var()'s postconfig argument,
this wrapper around runCmd() has a postconfig argument that has odd
behaviour: it _appends_ the new configuration to local.conf instead of
having them used for this specific run (unlike the other functions in
commands.py)
None of the eSDK tests use this functionality, so remove it. Future test
cases that need to write further configuration should do so directly.
(From OE-Core rev: 6196d096b95b36f8b72b4049d5479f8f23e7891d)
Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
ctest automatically optimizes the order of (parallel) test execution based on
historic test case runtime via the COST property (see [0]), which can have a
significant impact on overall test run times. Sadly this feature is broken in
CMake < 4.0.0 for test cases that have spaces in their name (see [1]).
This commit backports the upstream fix. As repeated test runs are expected to
mainly take place inside the SDK, the patch is only applied to 'nativesdk'
builds.
[0]: https://cmake.org/cmake/help/latest/prop_test/COST.html
[1]: https://gitlab.kitware.com/cmake/cmake/-/issues/26594
Reported-By: John Drouhard <john@drouhard.dev>
(From OE-Core rev: dcbaf42dd74cc0bda7254856589613718ed3f057)
Signed-off-by: Moritz Haase <Moritz.Haase@bmw.de>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Change enable-x11-forwarding PACKAGECONFIG to x11, while we can rely
directly on X11 DISTRO_FEATURE. Previously when enable-x11-forwarding was
used, the XAUTH_COMMAND failed due to missing xauth. So add xauth to
runtime recommends dependency and enable this behavior as default.
(From OE-Core rev: f0d2374b94a0e5d60ceed17998cd97f85b0486f6)
Signed-off-by: Andrej Valek <andrej.v@skyrain.eu>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If start-of-line is not enforced, then bogus tag 'gstreamer-sharp-1.13.0.1' would
match as 13.0.1.
(From OE-Core rev: 094cbd4f6a9f57eb146207355a921eaae44012ab)
Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This adds a new selftest case `test_sign_uboot_fit_image_without_spl` to verify
that the build can correctly generate and sign a U-Boot FIT image in a scenario
where no SPL is used.
Background:
- Some boards build only the U-Boot proper FIT image and do not require an SPL.
- The signing flow must handle this case gracefully: generate the ITS, sign
the FIT image, and skip signing/injecting a key into the SPL DTB.
What this test does:
1) Enables `UBOOT_FITIMAGE_ENABLE` and `SPL_SIGN_ENABLE` but explicitly sets
`SPL_DTB_BINARY` to an empty string to indicate that no SPL is present.
2) Verifies that the U-Boot ITS and FIT image are built successfully.
3) Confirms that the generated ITS file includes signature metadata as requested.
4) Dumps the FIT image to ensure that the signature nodes exist.
5) Confirms that the log for `do_uboot_assemble_fitimage` shows the expected
mkimage/mkimage_sign invocation.
This ensures that signing works correctly even when only the U-Boot proper is built,
which matches real-world configurations that do not require an SPL.
(From OE-Core rev: cdb4f4249a386113bebc65918a1b088c64e59182)
Signed-off-by: Jamin Lin <jamin_lin@aspeedtech.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This commit cleans up and modernizes the shell condition expressions in
`uboot-sign.bbclass` to follow best practices for portable and reliable shell usage.
Key changes:
- Replace legacy `[ -a ]` and `[ -o ]` with explicit `[ ] && [ ]` and `[ ] || [ ]`.
Modern POSIX and busybox sh recommend using `&&` and `||` instead of `-a` and `-o`
because `-a` and `-o` are less robust and can cause parsing ambiguities in some shells.
- Simplify `concat_dtb()` by moving the DTB existence check to the top and using
early `return` to avoid deep nesting.
- Remove redundant fallback `else` blocks; use clearer control flow with direct checks.
This improves maintainability, reduces shell syntax pitfalls, and aligns with
current shell scripting best practices.
References:
- POSIX recommends avoiding `-a` and `-o` in `[ ]` and using explicit `&&` and `||`:
https://pubs.opengroup.org/onlinepubs/9699919799/utilities/test.html
(From OE-Core rev: d2740e39800a044d557b620e38ca0ac1b8c6d030)
Signed-off-by: Jamin Lin <jamin_lin@aspeedtech.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Previously, the signing flow in "uboot-sign.bbclass" assumed that SPL was always
present and that the FIT signing process must inject the public key into the
SPL DTB. This made it inflexible for use cases where only the U-Boot proper
FIT image is built and signed, with no SPL binary at all.
This change introduces the following adjustments:
- The `SPL_DTB_BINARY` variable can be explicitly set to an empty string
to indicate that no SPL is present.
- The signing logic checks `SPL_DTB_BINARY` and skips injecting the
key or verifying the SPL DTB if it is empty.
- The FIT image generation and deployment are always performed if
`UBOOT_FITIMAGE_ENABLE` is enabled, regardless of the SPL settings.
- The deploy helper now uses a single check on `SPL_DTB_BINARY` to decide
whether to deploy the signed SPL DTB.
Now the sign step checks if SPL_DTB_BINARY is empty:
If present, it signs the FIT image and injects the public key into the SPL DTB,
then verifies both.
If empty, it only signs the FIT image and generates the ITS with the signature
node, but does not attempt to verify or add the key to a non-existent SPL DTB.
Key Behavior Explained
If SPL_DTB_BINARY is empty, we assume there is no SPL.
If UBOOT_FITIMAGE_ENABLE=1, we always create the FIT image and ITS.
If SPL_SIGN_ENABLE=1, we always sign the FIT image, but only inject the key into
the SPL DTB if it exists.
Example usage:
UBOOT_FITIMAGE_ENABLE = "1"
SPL_SIGN_ENABLE = "1"
SPL_DTB_BINARY = ""
This means:
- Generate and sign the FIT image.
- Do not attempt to sign or deploy an SPL DTB.
This aligns the implementation with real scenarios where some boards do not
require an SPL.
(From OE-Core rev: 7ad6acd8841752a5b75b8e2666bca5b609347cc1)
Signed-off-by: Jamin Lin <jamin_lin@aspeedtech.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Install libpng test-suite to run it as a ptest.
As the test-suite takes more than 30 seconds to run,
add libpng-ptest to PTESTS_SLOW in ptest-packagelists.inc
(From OE-Core rev: 8dca5305c950e6a06b3f344ffdbbb7386d802095)
Signed-off-by: Poonam Jadhav <poonam.jadhav@kpit.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We don't have a recipe for subunit in core (or any layer in the layer
index), and unfortunately a bug in the CMakeLists means that the native
build can detect subunit on the host (so host contamination) and only
half-enable it, so the build fails.
Take a patch from upstream's github to add an option for subunit, and
explicitly disable it.
(From OE-Core rev: 155f0f65fa0f08a4586d3c9a02976f43a10b08aa)
Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This patch is specific to the autotools build of libcheck, but we use
CMake now.
(From OE-Core rev: 98091331829bd88b57fef5b811a4117b8db16032)
Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When SPDX_INCLUDE_COMPILED_SOURCES is enabled, only include the
source code files that are used during compilation.
It uses debugsource information generated during do_package.
This enables an external tool to use the SPDX information to disregard
vulnerabilities that are not compiled.
As example, when used with the default config with linux-yocto, the spdx size is
reduced from 156MB to 61MB.
Tested with bitbake world on oe-core.
CC: Quentin Schulz <quentin.schulz@cherry.de>
CC: Joshua Watt <JPEWhacker@gmail.com>
CC: Peter Marko <peter.marko@siemens.com>
(From OE-Core rev: c6a2f1fca76fae4c3ea471a0c63d0b453beea968)
Signed-off-by: Daniel Turull <daniel.turull@ericsson.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
As currently written, the PREFERRED_PROVIDER entries don't work with multilib
since they are added after the manipulations by multilib_global at ConfigParsed
time. We therefore need to spell out the configs correctly for multilib.
To avoid variable overwritten warnings from the expansion, we then have to
expand out the nativesdk providers usng class overrides.
(From OE-Core rev: 1f69850b62d1e022055d0f3b93ad270c2980462e)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This change implements a toolchain selection mechanism. Selection is
made using a set of variables, primarily PREFERRED_TOOLCHAIN_TARGET which
defaults to gcc.
It uses the familiar name for toolchain e.g. "gcc" which selects GNU
compiler + binutils as default C/C++ toolchain or "clang" which will
use LLVM/Clang Compiler. Layers an add their own toolchain definitions
too.
There are also PREFERRED_TOOLCHAIN_NATIVE and PREFERRED_TOOLCHAIN_SDK
which will ulitmately allow selection of the toolchain used for the
native/cross and nativesdk/crosssdk compilers. This currently isn't
functional but is essential to the patch to ensure things are set
to the existing gcc support in those cases.
Users would most commonly want to set:
PREFERRED_TOOLCHAIN_TARGET ?= "clang"
in local.conf or other distro specific global configuration metadata.
It is also selectable at recipe scope, since not all packages are
buildable with either clang or gcc, a recipe can explicitly require
a given toolchain using the TOOLCAHIN variable, e.g. glibc can not
be built with clang therefore glibc recipe sets:
TOOLCHAIN = "gcc"
The TOOLCHAIN variable is distinct from the user preference so recipes
with specific requirements can be identified. This also allows different
polcies to be be specified for native/SDK cases in the future.
(From OE-Core rev: 45bdedd213aff8df3214b95ef2a8551c0abd93a0)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Use the new deferred class event to set the class overrides earlier.
This improves interaction of the override with PACKAGECONFIG values
that control conditional inherits (such as python support).
This also allows toolchain configuration in an easier and more user
friendly way.
(From OE-Core rev: d58f94d5684332bd4fac3747688558bb261b9c63)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This brings in the features we need to toolchain selection.
(From OE-Core rev: 78c1061dc3916677bd9c2825820a1bab7e75f28a)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Update the plugin names to account for the "-" to "_" plugin name change.
(From OE-Core rev: afa1b5c9f6ed17c021e37a54d0d6abee50a60bf9)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Changelog:
=============
- Fix duplicated dependencies added to the lock file when the same dependency with extras is requested.
- Stabilize order of the extras and dependency-groups fields in pylock output.
- Fix Windows 11 install pdm error, which is because of msgpack install failure.
- Change the return type of array_of_inline_tables to list[dict] from list[str]
- Ensure uv resolver to include hash for package files.
- Avoid infinite recursion when reading pyproject.toml with circular file dependencies.
- Support pylock as alternative lock format and make it opt-in by config.
- Search for package metadata in lock file first when reuse strategy is used.
(From OE-Core rev: 121c609e91dd7eb72670513eef8c31a5f2271c89)
Signed-off-by: Wang Mingyu <wangmy@fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
License-Update: Copyright year updated to 2025
(From OE-Core rev: 2ac6a959f32214d958a7a0cf1973a9bc66839a9b)
Signed-off-by: Wang Mingyu <wangmy@fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Changelog:
===========
- Fix a longstanding ordering issue with extracting type information
from properties which have a reference to another property. 'mac-mode'
is the one in the Linux kernel.
- Fix a false positive warning about missing
unevaluatedProperties/additionalProperties
(From OE-Core rev: e36601ca6454ec62ad2dd0db47724e2ad4c240cd)
Signed-off-by: Wang Mingyu <wangmy@fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Bugs fixed
----------
* Attribute lookups failed on the "bool" builtin type.
* Type checks on or-ed union types could incorrectly return false.
* Negative list indexing could accidentally wrap around twice in PyPy and the Limited API.
* Iterating over literal sequences with starred (unpacked) items could infer a wrong
type for the loop variable and fail to assign the values.
* Calls to C functions taking exception types failed to check for a 'None' argument.
* Fused functions had an incorrect "__module__" attribute.
* The type of Cython implemented functions had an incorrect "__module__" attribute.
* Errors while indexing into "bytearray" or "str" in "nogil" sections could crash.
* "bytearray.append()" could silently accept some invalid character numbers.
* The C++11 "<type_traits>" header was included regardless of the C++ version.
* "PyDict_GetItemStringRef()" was accidentally used in older Limited API versions.
* "abort()" was used but not always available in the Limited API.
* Some dependencies were missing from the "depfile".
* Embedded function signatures were not always separated from the existing docstring.
* "numpy.math" was missing from "Cython/Includes/" and could not be cimported.
* Some tests were adapted for NumPy 2.x.
* Some C compiler warnings were fixed.
* "Cython.Build" was not officially exposing the "cythonize" function.
(From OE-Core rev: ad9e2ba3d7c1af3f7084427eb9ddb0822460b108)
Signed-off-by: Wang Mingyu <wangmy@fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Changelog:
===========
* Fix missing override when CONFIG_COMPAT_OLD_SIGACTION is not defined
* fix: writeback: 'balance_dirty_pages' Respect 'CONFIG_CGROUP_WRITEBACK'
* Fix: scsi: RESERVE and RELEASE renamed in Linux v6.15-rc1
* Fix: del_timer[_sync] deleted in linux v6.15-rc1
* Fix: Use 'nonseekable_open' for proc files
* Fix: trace_balance_dirty_pages in Linux v6.14.2
* fix: version constraint for building lttng-probe-9p
(From OE-Core rev: 8f8b8e72ff4ea562630b428ff8ab7f163d0c7084)
Signed-off-by: Wang Mingyu <wangmy@fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Changelog:
===========
- selected bug fixes:
* improve transaction ordering by allowing more uninst->uninst
edges
* implement color filtering when adding update targets
- new features:
* support orderwithrequires dependencies in susedata.xml
(From OE-Core rev: 000d105135c6ae871bfa8699d9008a1f4324c7f7)
Signed-off-by: Wang Mingyu <wangmy@fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Changelog:
- Hardened the libjpeg API against hypothetical calling applications that may
erroneously change the value of the 'data_precision' field in
'jpeg_compress_struct' or 'jpeg_decompress_struct' after calling
'jpeg_start_compress()' or 'jpeg_start_decompress()'.
(From OE-Core rev: 49a69532bc2b4a7d6ae8136123e92b10b5e657f5)
Signed-off-by: Wang Mingyu <wangmy@fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
License-Update: Copyright year updated to 2025
not-win32.patch
refreshed for 3.5.1
(From OE-Core rev: 4c32496a8b89cb1355672d6b7159aff09c69797b)
Signed-off-by: Wang Mingyu <wangmy@fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Bugs fixed:
- Wrong behavior of GdkKeymap on macOS
- cups: NULL-terminate array of choices
- Fix the android build
- icontheme: Load the missing image icon from the theme
(From OE-Core rev: 1a31f269130429b7c690340c688c1f503f6473d3)
Signed-off-by: Wang Mingyu <wangmy@fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When using the license finder the caller might know some more license
hashes, for example if it is updating existing metadata.
Allow the caller to pass more hashes that can be used when identifying
licenses.
(From OE-Core rev: 9011bc307fcdccb144b75d77b36bbc5c8d4bd96d)
Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Rewrite the license checksum generation and loading of CSV files to be
clearer.
This also expands the scan of COMMON_LICENSE_DIR to include LICENSE_PATH,
which can be extended by layers to provide more license texts.
(From OE-Core rev: 417240ba7a9b3985530988940a222b079b503b64)
Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
