]> git.ipfire.org Git - thirdparty/openembedded/openembedded-core-contrib.git/log
thirdparty/openembedded/openembedded-core-contrib.git
2 weeks agoscripts/install-buildtools: Update to 5.2.3
Aleksandar Nikolic [Sun, 5 Oct 2025 20:05:16 +0000 (22:05 +0200)] 
scripts/install-buildtools: Update to 5.2.3

Update to the 5.2.3 release of the 5.2 series for buildtools

Signed-off-by: Aleksandar Nikolic <aleksandar.nikolic@zeiss.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agoselftest/meta_ide: source the environment first, then change to the sources directory
Alexander Kanavin [Tue, 7 Oct 2025 18:21:43 +0000 (20:21 +0200)] 
selftest/meta_ide: source the environment first, then change to the sources directory

This too used to work by coincidence: sourcing the environment
quietly failed without changing to the build directory, but
now that it works properly, things should be done in correct order.

Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agometa/classes-recipe/toolchain-scripts.bbclass: eliminate bash-ism
Alexander Kanavin [Tue, 7 Oct 2025 18:21:42 +0000 (20:21 +0200)] 
meta/classes-recipe/toolchain-scripts.bbclass: eliminate bash-ism

Sourcing a script with arguments is a non-standard bash extension
and doesn't work with other shells (e.g. dash, which is used on
Debian and derivatives). This used to work by coincidence when
running against integrated poky repo without having to separately
specify where bitbake is, but no longer does.

Using set is a POSIX standard.

Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agotoolchain-scripts.bbclass: fix bitbake-setup compatibility
Adrian Freihofer [Mon, 6 Oct 2025 19:10:55 +0000 (21:10 +0200)] 
toolchain-scripts.bbclass: fix bitbake-setup compatibility

Pass the BITBAKEDIR parameter to oe-init-build-env when called from the
environment-setup script. This fixes compatibility with bitbake-setup,
which places bitbake in a different directory structure than the
standard poky repository layout where bitbake is located alongside
oe-init-build-env.

The issue was discovered when running the oe-selftest test
DevtoolIdeSdkTests.test_devtool_ide_sdk_shared_sysroots, which failed
because the environment-setup script was not able to find bitbake.

File ".../openembedded-core/meta/lib/oeqa/selftest/cases/devtool.py",
  line 2955, in test_devtool_ide_sdk_shared_sysroots
  self.assertExists(cmake_native)
  ~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^
File ".../openembedded-core/meta/lib/oeqa/selftest/case.py", line 251,
  in assertExists
  raise self.failureException(msg)
AssertionError: 'Error: The bitbake directory (/tmp/devtoolqah9ndff2x/bitbake)
  does not exist!  Please ensure a copy of bitbake exists at this location or
  specify an alternative path on the command line\n
  .../build-st/tmp/sysroots/x86_64/usr/bin/cmake' does not exist

Signed-off-by: Adrian Freihofer <adrian.freihofer@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agoscripts/runqemu: raise an error when bitbake was not found
Richard Grünert [Wed, 27 Aug 2025 06:49:40 +0000 (08:49 +0200)] 
scripts/runqemu: raise an error when bitbake was not found

Running 'scrupts/runqemu' without bitbake in PATH causes the
following error:

```
Traceback (most recent call last):
  File "/home/rg/temp_stuff/oe_2/./scripts/runqemu", line 1807, in main
    config.check_args()
    ~~~~~~~~~~~~~~~~~^^
  File "/home/rg/temp_stuff/oe_2/./scripts/runqemu", line 624, in check_args
    s = re.search('^DEPLOY_DIR_IMAGE="(.*)"', self.bitbake_e, re.M)
  File "/usr/lib/python3.13/re/__init__.py", line 177, in search
    return _compile(pattern, flags).search(string)
           ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^
TypeError: expected string or bytes-like object, got 'NoneType'
```

This patch adds a more helpful error message to inform the user that
bitbake was not found, e.g. because oe-init-build-env was not sourced.

This is an example of the new error message after the patch:

```
runqemu - ERROR - In order for this script to dynamically infer paths
 kernels or filesystem images, you either need bitbake in your PATH
 or to source oe-init-build-env before running this script.

 Dynamic path inference can be avoided by passing a *.qemuboot.conf to
 runqemu, i.e. `runqemu /path/to/my-image-name.qemuboot.conf`

 Bitbake is needed to run 'bitbake -e', but it is not found in PATH. Please source the bitbake build environment.
```

CC: Richard Purdie <richard.purdie@linuxfoundation.org>
CC: Alexander Kanavin <alex.kanavin@gmail.com>
Signed-off-by: Richard Grünert <r.gruenert@pironex.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agoscripts/runqemu: remove the code block that works around the missing bitbake environment
Alexander Kanavin [Tue, 7 Oct 2025 12:46:52 +0000 (14:46 +0200)] 
scripts/runqemu: remove the code block that works around the missing bitbake environment

As confirmed by the previous patch this code path is never taken
and can be removed.

Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agorunqemu: ensure that bitbake environment is either returned, or an exception is raised
Alexander Kanavin [Tue, 7 Oct 2025 12:46:51 +0000 (14:46 +0200)] 
runqemu: ensure that bitbake environment is either returned, or an exception is raised

This eliminates the other remaining code path where environment getter
returns 'nothing'. This and the previous patch were tested in a-full,
and no errors occurred [1], which means the code paths that make
use of the function returning nothing are never actually executed
and can be cleaned up (in the following patch).

The rationale is that if environment getter cannot obtain the environment,
it should report that and not sweep the issue under the carpet;
it's up to the caller to handle that situation, or make pre-emptive
checks that avoid calling the environment getter when it is bound to fail.

[1] https://lists.openembedded.org/g/openembedded-core/message/223651

Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agoclang: consolidate LLVM_APPEND_VC_REV=OFF
Ross Burton [Wed, 8 Oct 2025 14:26:20 +0000 (15:26 +0100)] 
clang: consolidate LLVM_APPEND_VC_REV=OFF

Whilst the change to add TMPDIR to GIT_CEILING_DIRECTORIES should stop
LLVM from embedding git information into the recipes, also disable this
behaviour explicitly.

We do this because it's not just the sha of the source tree but also
the full URL of the repository, which would be an information leak if
an internal git mirror was being used.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agoclasses/toolchain/clang: depend on llvm-native for tools, not clang
Ross Burton [Tue, 7 Oct 2025 10:42:47 +0000 (11:42 +0100)] 
classes/toolchain/clang: depend on llvm-native for tools, not clang

The relevant tools (objdump, strip, etc) are part of llvm-native now,
so trim dependencies and depend on that directly instead of clang-cross.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agoclang: use llvm recipe
Ross Burton [Tue, 7 Oct 2025 10:42:46 +0000 (11:42 +0100)] 
clang: use llvm recipe

Change this recipe to build just clang and clang-tools-extra, using the
LLVM provided by the llvm recipe.

This adds an 'extra-tools' PACKAGECONFIG (enabled by default) that
controls whether to build the clang-tools-extra project. This includes
clang-tidy and clangd, but the compile time and size for these
components is not insignificant.

Add a patch from upstream to support using native prebuilt tools (such
as clang-tblgen) when building standalone.

Add a patch that is being worked on with upstream to not rebuild clang-
tblgen if it has already been provided. This saves a little build time,
but more importantly for us resolves a static linking/uninative problem.

Remove the dependency on clang-cross in non-native builds by just
depending on llvm-native and using the un-prefixed tools directly.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agollvm: add recipe for just the LLVM libraries
Ross Burton [Tue, 7 Oct 2025 10:42:45 +0000 (11:42 +0100)] 
llvm: add recipe for just the LLVM libraries

Whilst it's convenient to build all of the LLVM project in one big
recipe, that's not ideal when we may just need LLVM on target and not
the rest.

Bring back a LLVM recipe that can be used by both clang (shortly) and
Rust (in the future)

Set the build type to MinSizeRel and DEBUG_LEVELFLAG to -g1 (instead of
the default, -g): the LLVM debug symbols are very large (several
gigabytes) and this reduces them to hundreds of megabytes.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agoclang: globally disable build-time RPATHs for reproducibility
Ross Burton [Tue, 7 Oct 2025 10:42:44 +0000 (11:42 +0100)] 
clang: globally disable build-time RPATHs for reproducibility

Various bits of the LLVM project set the build RPATH in a way that means
we have non-deterministic binaries even though we should always be using
relative paths.

This clearly is not working as some of the binaries get rewritten on
install and have large string paddings that correlate with erased build
paths.

So that we don't have to disable RPATHs in every recipe, just do it once
in the common include file and remove the existing recipe-specific
assignments that are now not needed.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agobitbake.conf: add TMPDIR to GIT_CEILING_DIRECTORIES
Ross Burton [Wed, 8 Oct 2025 14:02:24 +0000 (15:02 +0100)] 
bitbake.conf: add TMPDIR to GIT_CEILING_DIRECTORIES

We export GIT_CEILING_DIRECTORIES=WORKDIR to ensure that git calls
inside the builds don't find oe-core when they're meant to be looking
for the git repository of the source code.

However, this breaks for recipes that use work-shared (such as llvm), as
their working directory is outside of WORKDIR.

Solve this by adding TMPDIR to the list as a final catch, but keeping
WORKDIR first so that git will stop sooner in the general case.

This solves reproduciblity problems in LLVM, where for example lld's
version string would contain the URL and commit hash of the poky repo
being built.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agopopulate_sdk_ext: Update to toolcfg.conf
Richard Purdie [Mon, 6 Oct 2025 23:05:36 +0000 (00:05 +0100)] 
populate_sdk_ext: Update to toolcfg.conf

Add handling for toolcfg to the eSDK generation. Take the opporunity to clean
up the code duplication too.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agobitbake.conf,lib/configfragments: Use a new dedicated toolcfg.conf file
Richard Purdie [Mon, 6 Oct 2025 13:21:12 +0000 (14:21 +0100)] 
bitbake.conf,lib/configfragments: Use a new dedicated toolcfg.conf file

Rather than using auto.conf which already has established use in CI, or
local.conf which users expect to own/control, start writing "tooling"
controlled settings to a toolcfg.conf.

This frees CI to handle auto.conf as it wants, but avoids the tooling
breaking users local.conf files.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agooeqa/selftest/buildhistory: Fix test if USER_CLASSES is unset
Richard Purdie [Mon, 6 Oct 2025 13:14:54 +0000 (14:14 +0100)] 
oeqa/selftest/buildhistory: Fix test if USER_CLASSES is unset

If USER_CLASSES is unset, the test was failing. Fix that.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agooeqa/selftest/oescripts: Ensure buildstats is present
Richard Purdie [Mon, 6 Oct 2025 13:14:53 +0000 (14:14 +0100)] 
oeqa/selftest/oescripts: Ensure buildstats is present

These tests need buildstats to be configured. Add that piece of
config in case it isn't by default.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agooeqa/selftest/bblock/fitimage/sstatetests: Fix changing MACHINE during the test
Richard Purdie [Mon, 6 Oct 2025 13:14:52 +0000 (14:14 +0100)] 
oeqa/selftest/bblock/fitimage/sstatetests: Fix changing MACHINE during the test

With config fragments, changing MACHINE in the test like this no
longer works. Use the forcevarable override to allow it to work.
This also needs a tweak to bitbake to work correctly, sent
seperately. Whilst ugly, this avoids the need to start changing
config fragments within oeqa right now.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agooeqa/selftest: Fix single threaded race issue
Richard Purdie [Mon, 6 Oct 2025 13:09:02 +0000 (14:09 +0100)] 
oeqa/selftest: Fix single threaded race issue

oe-selftest sets up separate build directories to run the tests in.
To to this, environment paths pointing at the previous build directory
are updated. In the multi-threaded case this is fine as the thread is
destroyed and the parent remains unchanged but in the single threaded
case, the environment is broken afterwards. This can mean we try and access
a directory which is in the process of being deleted (e.g. by clobberdir).

Restore the environment afterwards regardless to ensure the single threaded
case doesn't try and access the build directory which is now being deleted.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agoselftest/bblayers: maintain compatibility with integrated poky repo
Alexander Kanavin [Wed, 8 Oct 2025 12:42:53 +0000 (14:42 +0200)] 
selftest/bblayers: maintain compatibility with integrated poky repo

This amends the recently merged commit that assumes the transition
to separate repositories has already happened and re-instatates
support for integrated poky.

Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agooeqa/selftest/wic: fix PATH for wic.Wic2.test_extra_partition_plugin
Yoann Congal [Fri, 3 Oct 2025 21:12:48 +0000 (23:12 +0200)] 
oeqa/selftest/wic: fix PATH for wic.Wic2.test_extra_partition_plugin

Without importing PATH from the wic-tools recipes, the build host PATH
is used and this test may fail depending on tools (parted, dumpe2fs,
...) availability. This triggers build faillure on AB (e.g. [0])

To fix this, import PATH from wic-tools and ensure the original
environment is restored after.

Since this indent a block of code into a try/finally block, here is the
diff ignoring white spaces change:
  diff --git a/meta/lib/oeqa/selftest/cases/wic.py b/meta/lib/oeqa/selftest/cases/wic.py
  index bff3842305..bc99673d0d 100644
  --- a/meta/lib/oeqa/selftest/cases/wic.py
  +++ b/meta/lib/oeqa/selftest/cases/wic.py
  @@ -1680,0 +1681,4 @@ INITRAMFS_IMAGE = "core-image-initramfs-boot"
  +        oldpath = os.environ['PATH']
  +        os.environ['PATH'] = get_bb_var("PATH", "wic-tools")
  +
  +        try:
  @@ -1696,0 +1701,3 @@ INITRAMFS_IMAGE = "core-image-initramfs-boot"
  +        finally:
  +            os.environ['PATH'] = oldpath
  +

[0]: https://autobuilder.yoctoproject.org/valkyrie/#/builders/35/builds/2456

Signed-off-by: Yoann Congal <yoann.congal@smile.fr>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agooeqa/selftest/wic: fix PATH for wic.Wic2.test_extra_partition_space
Yoann Congal [Fri, 3 Oct 2025 21:12:47 +0000 (23:12 +0200)] 
oeqa/selftest/wic: fix PATH for wic.Wic2.test_extra_partition_space

Without importing PATH from the wic-tools recipes, the build host PATH
is used and this test may fail depending on tools (parted, dumpe2fs,
...) availability. This triggers build faillure on AB (e.g. [0])

To fix this, import PATH from wic-tools and ensure the original
environment is restored after.

Since this indent a block of code into a try/finally block, here is the
diff ignoring white spaces change:
  diff --git a/meta/lib/oeqa/selftest/cases/wic.py b/meta/lib/oeqa/selftest/cases/wic.py
  index b1c318bd4e..34d844b90b 100644
  --- a/meta/lib/oeqa/selftest/cases/wic.py
  +++ b/meta/lib/oeqa/selftest/cases/wic.py
  @@ -1331,0 +1332,4 @@
  +        oldpath = os.environ['PATH']
  +        os.environ['PATH'] = get_bb_var("PATH", "wic-tools")
  +
  +        try:
  @@ -1366,0 +1371,2 @@
  +        finally:
  +            os.environ['PATH'] = oldpath

[0]: https://autobuilder.yoctoproject.org/valkyrie/#/builders/35/builds/2456

Signed-off-by: Yoann Congal <yoann.congal@smile.fr>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agobusybox: patch CVE-2025-46394
Peter Marko [Fri, 3 Oct 2025 18:25:33 +0000 (20:25 +0200)] 
busybox: patch CVE-2025-46394

Pick commit mentioning this CVE.

Signed-off-by: Peter Marko <peter.marko@siemens.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agoudev-extraconf: Avoid slashes in mountpoint names
Robert Tiemann [Thu, 2 Oct 2025 09:59:25 +0000 (11:59 +0200)] 
udev-extraconf: Avoid slashes in mountpoint names

Devices with labels such as "Hello/World/Foo/Bar" cause mount.sh to
create the directory structure @MOUNT_BASE@/Hello/World/Foo/Bar. The
partition is mounted to the nested "Bar" directory. On device removal,
the directory structure is not cleaned up.

This commit replaces all forward slashes in partition labels by
underscores to avoid this edge case.

Signed-off-by: Robert Tiemann <rtie@gmx.de>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agoudev-extraconf: Speed up mount.sh
Robert Tiemann [Thu, 2 Oct 2025 09:59:24 +0000 (11:59 +0200)] 
udev-extraconf: Speed up mount.sh

On devices with many partitions, running blkid without parameters can
be rather slow because all block devices are inspected:

$ time /sbin/blkid
real    0m0.474s
user    0m0.026s
sys     0m0.172s

versus

$ time /sbin/blkid /dev/mmcblk0p10
real    0m0.027s
user    0m0.002s
sys     0m0.018s

Plugging in a device with 5 partitions means that mount.sh is going to
be executed 5 times, and so will be blkid. In the real-world case
outlined above, this adds up to an overhead of about 2.3 seconds for
blkid alone.

This commit changes mount.sh so that the block device of interest is
passed directly to blkid such that blkid inspects only that device,
leading to significant speedup.

Signed-off-by: Robert Tiemann <rtie@gmx.de>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agomesa: update 25.2.3 -> 25.2.4
Zoltán Böszörményi [Thu, 2 Oct 2025 15:12:26 +0000 (17:12 +0200)] 
mesa: update 25.2.3 -> 25.2.4

Announcement:
https://lists.freedesktop.org/archives/mesa-dev/2025-October/226554.html

Signed-off-by: Zoltán Böszörményi <zboszor@gmail.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agowic: extra partition plugin
Pierre-Loup GOSSE [Thu, 2 Oct 2025 14:45:34 +0000 (16:45 +0200)] 
wic: extra partition plugin

The extra_partition plugin allows populating an extra partition with
files listed in the new IMAGE_EXTRA_PARTITION_FILES variable. The
implementation is similar to the bootimg_partition plugin.

This plugin provides an easy way to install files that are not part of
the rootfs, from the deploy directory.

Signed-off-by: Pierre-Loup GOSSE <pierre-loup.gosse@smile.fr>
Reviewed-by: Yoann CONGAL <yoann.congal@smile.fr>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agomesa-demos: split info tools to a separate package
Dmitry Baryshkov [Thu, 2 Oct 2025 01:50:30 +0000 (04:50 +0300)] 
mesa-demos: split info tools to a separate package

While the rest of programs inside mesa-demos are really "demos", several
*info utilities have separate value as they allow gathering information
about the running system in a manner similar to clinfo or vulkaninfo.
Split them into a separate package in order to allow picking them info
the images without picking up the rest of "demos".

Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agotunes/arm: add support for newer arm64 cores
Jon Mason [Wed, 1 Oct 2025 17:58:14 +0000 (13:58 -0400)] 
tunes/arm: add support for newer arm64 cores

Add tunes for cores that have been present since GCC 14.1.0.
These have been present in meta-arm since mid-July, and have all been
verified with fvp-base and testimage.  Of those present in meta-arm,
arch-armv8-9a.inc, arch-armv9-4a.inc, and arch-armv9-5a.inc have been
excluded from this commit, as there are currently issues compiling some
packages (but still present in meta-arm for those that want to use them
despite those issues).

Signed-off-by: Jon Mason <jon.mason@arm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agogo: fix sigaction usage on i386 platforms
Randolph Sapp [Thu, 2 Oct 2025 00:28:24 +0000 (19:28 -0500)] 
go: fix sigaction usage on i386 platforms

In upstream the following commit [1] was submitted to resolve issues
with sigaction being used in linked libraries with cgo applications.

runtime: when using cgo on 386, call C sigaction function

This resolves potential segfaults with cgo applications that link to libraries
that intend to switch out signal handlers temporarily with sigaction.

[1] https://github.com/golang/go/commit/c5737dc21bbac9fbefc35ac9313e66291d66b382

Signed-off-by: Randolph Sapp <rs@ti.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agopython3-lxml: upgrade 6.0.1 -> 6.0.2
Khem Raj [Wed, 1 Oct 2025 18:18:39 +0000 (11:18 -0700)] 
python3-lxml: upgrade 6.0.1 -> 6.0.2

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agospirv-llvm-translator: Upgrade to 21.1.1
Khem Raj [Wed, 1 Oct 2025 18:18:38 +0000 (11:18 -0700)] 
spirv-llvm-translator: Upgrade to 21.1.1

Backports following fixes to 21 branch

29758b55 [Backport to llvm_release_210] Fix error propagation in parseSPIRV (#3365)
615cc15b Fix BFloat16 mangling (#3355)
9aae1acb [Backport to 21] Implement SPV_INTEL_bfloat16_arithmetic (#3290)
b50fc5ca [Backport to 21] Support for SPV_INTEL_shader_atomic_bfloat16 extension (#3343) (#3349)

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agoclang: Upgrade to 21.1.2 release
Deepesh Varatharajan [Wed, 1 Oct 2025 10:35:16 +0000 (03:35 -0700)] 
clang: Upgrade to 21.1.2 release

Brings following bugfixes on top of 21.1.1

b708aea0bc71 [SCEV] Don't perform implication checks with many predicates (#158652)
77a3b0eda361 [RISCV] Refactor RVV builtin code generation for reduce compilation time [NFC] (#154906)
f14551dbc4e8 [RISCV] Reduce ManualCodeGen for RVV intrinsics with rounding mode. NFC
dcc2c1c933be [RISCV] Reduce ManualCodeGen for segment load/store intrinsics. NFC
e625a781211e [LLVM] Update CUDA ELF flags for their new ABI (#149534)
3e93017936b5 MC: Better handle backslash-escaped symbols (#158780)
bc5e9a5e2009 [MC] Add parseSymbol() helper (NFC) (#158106)
e2e5eb2f1cd9 [Loads] Check for overflow when adding MaxPtrDiff + Offset.
661c387fc2f1 release/21.x: [VPlan] Don't narrow op multiple times in narrowInterleaveGroups.
f5c1b5206cbe [PowerPC] Avoid working on deleted node in ext bool trunc combine (#160050)
db70369f400e [Clang] Fix an error-recovery crash after d1a80dea (#159976)
9c8736f348e0 [ELF] -r/--emit-relocs: Fix crash when processing .rela.text before .text (#156354)
15a31832eab7 [RISCV] Re-work how VWADD_W_VL and similar _W_VL nodes are handled in combineOp_VLToVWOp_VL. (#159205)
f089fb21fffa [LoongArch] Fix MergeBaseOffset for constant pool index operand (#159336)
9eedaf5b1001 [VectorCombine] Fix scalarizeExtExtract for big-endian (#157962)
5af5cfb60d9e [clang-format] Handle C digit separators (#158418)
0174263ac214 Bump version to 21.1.2
ff039a98523f [RISCV] Support PreserveMost calling convention (#148214)
1a644bc50948 [X86] Fix assertion in AVX512 setcc combine due to invalid APInt mask width (#155775)
2c8cb316b5d2 [lldb][test] Only assert function name is in user-code on Darwin platforms
f5ba88341e7c [lldb][test] TestTsanBasic.py: fix function name assertion
faedeb1a2420 [lldb][Instrumentation] Set selected frame to outside sanitizer libraries (#133079)
54896838ca5e [lldb][Target] Clear selected frame index after a StopInfo::PerformAction (#133078)
f4907049285c [Clang][Cygwin] Use correct mangling rule (#158404)

Signed-off-by: Deepesh Varatharajan <Deepesh.Varatharajan@windriver.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agoselftest/bblayers: do not assume integrated poky repo
Alexander Kanavin [Mon, 6 Oct 2025 11:34:17 +0000 (13:34 +0200)] 
selftest/bblayers: do not assume integrated poky repo

The autobuilder is being transitioned to bitbake-setup which
sets up individual repositories that yocto is made of. Adjust
hardcoded revisions to match these repositories.

Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agodevtool: ide_sdk: pass BITBAKEDIR to oe-init-build-env
Adrian Freihofer [Sun, 5 Oct 2025 22:00:33 +0000 (00:00 +0200)] 
devtool: ide_sdk: pass BITBAKEDIR to oe-init-build-env

This fixes an issue where the generated install_and_deploy script is
unable to find the bitbake directory when run outside of the build
environment. This happens if the oe-selftest suite runs in a bitbake
environment that is bootstrapped by bitbake-setup.

oe-selftest -r devtool.DevtoolIdeSdkTests.test_devtool_ide_sdk_none_qemu

AssertionError: Command '.../build-st/workspace/ide-sdk/cmake-example/
  scripts/install_and_deploy_cmake-example-cortexa57' returned non-zero exit status 1:
Error: The bitbake directory (/tmp/devtoolqakq7kzgeo/bitbake) does not exist!
  Please ensure a copy of bitbake exists at this location or specify an
  alternative path on the command line
. /tmp/devtoolqakq7kzgeo/core-copy/oe-init-build-env
  /home/adrian/bitbake-builds/poky-master-poky-with-sstate-distro_poky-altcfg-machine_qemuarm64/build-st
  failed

Another reason this issue occurs with oe-selftests is that devtool
tests assume the full poky git repository is available. The setUpModule
function clones layer repositories, which for poky includes bitbake.
However, when using separate git repositories for bitbake and
openembedded-core, the bitbake directory is not preserved during layer
copying. While copying layers to allow modification during tests makes
sense, copying bitbake is less beneficial. Referring to the original
bitbake location is preferable, but cleaning up the devtool tests is
not part of this change.

Signed-off-by: Adrian Freihofer <adrian.freihofer@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agodevtool: ide-sdk: use /bin/sh instead of /bin/bash
Adrian Freihofer [Sun, 5 Oct 2025 22:00:32 +0000 (00:00 +0200)] 
devtool: ide-sdk: use /bin/sh instead of /bin/bash

When generating the install and deploy script for IDEs, use /bin/sh
instead of /bin/bash. While this is not addressing a known issue,
using the more portable /bin/sh shell is preferable and avoids
requiring bash to be installed.

Signed-off-by: Adrian Freihofer <adrian.freihofer@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agodevtool: ide_sdk: trivial alphabetical reorder
Adrian Freihofer [Sun, 5 Oct 2025 22:00:31 +0000 (00:00 +0200)] 
devtool: ide_sdk: trivial alphabetical reorder

Signed-off-by: Adrian Freihofer <adrian.freihofer@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
2 weeks agolib/bbconfigbuild/configfragments: disable the previous builtin fragment when enablin...
Alexander Kanavin [Thu, 2 Oct 2025 16:16:58 +0000 (18:16 +0200)] 
lib/bbconfigbuild/configfragments: disable the previous builtin fragment when enabling a new one

There was a flaw in the logic that allowed multiple builtin fragments with
the same prefix to be enabled at the same time. The correct behaviour
is that only one of them should be enabled, and when enabling it
all previously enabled fragments should be removed.

The issues that this caused are further explained in
https://bugzilla.yoctoproject.org/show_bug.cgi?id=15987

[YOCTO #15987]

Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agolinux-firmware: skip ldflags sanity check
Ross Burton [Mon, 29 Sep 2025 17:11:22 +0000 (18:11 +0100)] 
linux-firmware: skip ldflags sanity check

Some of the firmware blobs, for example qcom/apq8016/modem.mbn, are
actually ELF files. We don't want to hold these to our exacting standard
for link flags because we're not building them and they don't run on the
target directly.

Note that as this check parses the output of ${OBJDUMP} -p it behaves
differently with binutils vs llvm.  It looks like binutils bails early
as it doesn't know what the target architecture is, whereas LLVM is more
comprehensive in its support.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agopopulate_sdk_ext: do not require CONF_VERSION (e.g. local.conf version) to be set
Alexander Kanavin [Wed, 1 Oct 2025 11:45:40 +0000 (13:45 +0200)] 
populate_sdk_ext: do not require CONF_VERSION (e.g. local.conf version) to be set

The version checks for local.conf/site.conf/bblayers.conf are all optional,
and aren't enforced (by insane class) if the versions aren't set.

As bitbake-setup writes out a blank local.conf, it doesn't put a version in it
either. Also, esdk bundle has a fixed set of layers and is not at risk of
needing to update its own local.conf.

The same condition is already in place for esdk's bblayers.conf for similar reasons.

Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agoghostscript: upgrade 10.05.1 -> 10.06.0
Vijay Anusuri [Tue, 30 Sep 2025 15:39:01 +0000 (21:09 +0530)] 
ghostscript: upgrade 10.05.1 -> 10.06.0

Dropped 0001-Bug-708160-Fix-compatibility-with-C23-compilers.patch

Added 0001-Fix-32-bit-build.patch to fix 32-bit archs build

Added fPIC flag needed to prevent a linker error during the build process.

This upgrade addresses CVEs:
CVE-2025-59798, CVE-2025-59799, CVE-2025-59800, CVE-2025-59801

Changelog:
https://ghostscript.readthedocs.io/en/gs10.06.0/News.html

Signed-off-by: Vijay Anusuri <vanusuri@mvista.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agopackage.py: replace all files unconditionally when copy debug sources
Changqing Li [Tue, 30 Sep 2025 08:00:55 +0000 (16:00 +0800)] 
package.py: replace all files unconditionally when copy debug sources

This is for fixing reproducible issue for package like:
intel-speed-select-src. For intel-speed-select, one of the debug sources
is /usr/src/debug/intel-speed-select/1.0/include/linux/thermal.h,
file include/linux/thermal.h under ${S} (kernel-sources)
link file include/linux/thermal.h under ${B}, which link to ${S}/include/uapi/linux/thermal.h

During copy debug sources, sources under ${S} copied first, then sources
under ${B} is copied. mtime of ${S}/include/linux/thermal.h and
${S}/include/uapi/linux/thermal.h are decided by when it is fetched, so
it is not determinate, maybe same or different.

For cpio, if the in file is older than or the same as the exist file,
cpio will not replace the exist file with warning "cpio: xxx not created:
newer or same age version exists".  And this will cause
intel-speed-select-src maybe not reproducible.

And option '-u' for cpio, first, this will make the copied file
determistic. Second, source files under ${B} should have higher priority
then ${S}, it may be generated during build, the target is more likely
to use this file.

Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agolinux-firmware: Fix FILES for intel firmware
Ryan Eatmon [Thu, 25 Sep 2025 19:21:15 +0000 (14:21 -0500)] 
linux-firmware: Fix FILES for intel firmware

The 20250917 version is installing the firmware files into the
/usr/lib/firmware/intel directory and symbolic links to those files
under /usr/lib/firmware.  FILES needs to be updated to correctly package
all of the files into the right packages.  Without this we get a dependency
from the linux-firmware-iwlwifi-* and linux-firmware-qat packages back to
linux-firmware which means you have to install ALL of the drivers when you
depend on the specific package.

Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agoclang: build with MinSizeRel
Ross Burton [Mon, 29 Sep 2025 17:11:25 +0000 (18:11 +0100)] 
clang: build with MinSizeRel

This is the recommended build configuration upstream unless speed is the
ultimate goal. They say it's almost as fast as Release (-O3) but a lot
smaller on disk:

  clang-libclang-cpp: PKGSIZE changed from 71729568 to 49368816 (-31%)
  clang-libllvm: PKGSIZE changed from 83015559 to 56662823 (-32%)
  clang-tidy: PKGSIZE changed from 15861679 to 9956175 (-37%)
  clang-tools: PKGSIZE changed from 6865635 to 5374691 (-22%)
  clang: PKGSIZE changed from 29693308 to 22697548 (-24%)
  libclang: PKGSIZE changed from 40087018 to 26966218 (-33%)

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agoclang: fix reproducibility issue in LLVMConfig.cmake
Ross Burton [Mon, 29 Sep 2025 17:11:24 +0000 (18:11 +0100)] 
clang: fix reproducibility issue in LLVMConfig.cmake

The LLVMConfig.cmake file is mostly LLVM configuration, but it also
specifies if some Python modules (needed by the opt-viewer tool) were
present on the build host.

This is host contamination and a source of non-determinism, so remove it
from the installed file.

A ticket has been filed upstream to resolve this:

  https://github.com/llvm/llvm-project/issues/161199

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agoclang: use EXTRA_OECMAKE +=
Ross Burton [Mon, 29 Sep 2025 17:11:23 +0000 (18:11 +0100)] 
clang: use EXTRA_OECMAKE +=

Don't assign EXTRA_OECMAKE so that the include files can set defaults.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agogn: Add recipe
Khem Raj [Mon, 29 Sep 2025 15:36:47 +0000 (08:36 -0700)] 
gn: Add recipe

gn is a commonly used build tool to generate ninja files, used
in a lot of recipes e.g. chromium, qtwebengine, perfetto, hafnium etc.
these recipes come from different layers e.g. meta-qt6/meta-arm/meta-oe

since not all layers depend on meta-oe ( meta-arm does not ), its not
a possible option.

Given the usecases, putting it in core will benefit the ecosystem
and reduce some duplication. This recipe is taken from meta-arm

Disable maybe-initialized is a gcc specific option as error

Disable format warnings as errors, it adds -Wno-format explictly in
its build system

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agomusl: Update to tip of trunk
Khem Raj [Mon, 29 Sep 2025 15:33:19 +0000 (08:33 -0700)] 
musl: Update to tip of trunk

Brings
0ccaf057 printf: fix buffer overflow in floating point decimal formatting
0b86d60b riscv: fix setjmp assembly when compiling for ilp32f/lp64f.
f6944eb3 powerpc[64]: fix missing ctr and xer regs in syscall asm clobberlists
a6244de1 fix erroneous definition of IN6_IS_ADDR_V4COMPAT

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agocargo-c: update 0.10.15 -> 0.10.16
Deepesh Varatharajan [Mon, 29 Sep 2025 09:06:06 +0000 (02:06 -0700)] 
cargo-c: update 0.10.15 -> 0.10.16

Signed-off-by: Deepesh Varatharajan <Deepesh.Varatharajan@windriver.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agosanity: Update minimum gcc version to 10.1
Antonin Godard [Mon, 29 Sep 2025 07:31:24 +0000 (09:31 +0200)] 
sanity: Update minimum gcc version to 10.1

Since commit 19004950ad56 ("sanity: Check if the C++ toolchain supports
--std=gnu++20") host GCC must support "--std=gnu++20", which became
available in 10.1. This is already reflected in the documentation [1]
but not here.

Update the sanity check on GCC and raise the minimum version to 10.1.
Remove the dead link to www.softwarecollections.org.

[1]: https://git.yoctoproject.org/yocto-docs/commit/?id=0e538c102bfcb7184c76c2401e8cb878168c4434

Cc: Yoann Congal <yoann.congal@smile.fr>
Signed-off-by: Antonin Godard <antonin.godard@bootlin.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agolinux-firmware: backport the topology rename patch
Dmitry Baryshkov [Mon, 29 Sep 2025 05:37:46 +0000 (08:37 +0300)] 
linux-firmware: backport the topology rename patch

Backport the patch from the trunk, renaming topology file for one of
Qualcomm boards in order to reduce possible compatibility issues.

Cc: Vivek Puar <vpuar@qti.qualcomm.com>
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agobluez5: fix Bluetooth LE audio regression
Guðni Már Gilbert [Mon, 29 Sep 2025 04:33:52 +0000 (04:33 +0000)] 
bluez5: fix Bluetooth LE audio regression

Backport a fix for a regression introduced in 5.84 which breaks Bluetooth LE audio

Signed-off-by: Guðni Már Gilbert <gudni.m.g@gmail.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agoshadow: Increase the maximum group name length to 32 (the default)
Peter Kjellerstedt [Sun, 28 Sep 2025 16:37:33 +0000 (18:37 +0200)] 
shadow: Increase the maximum group name length to 32 (the default)

Back in 2014, the maximum group name length was increased from 16 (the
default) to 24. Since then, the default has increased to 32, and our
configuration is now actually a decrease in maximum group name length.

Remove the configuration to use the default instead, which aligns the
maximum group name length with the maximum user name length.

Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agomeson: upgrade to 1.9.1
Ross Burton [Sat, 27 Sep 2025 10:41:36 +0000 (11:41 +0100)] 
meson: upgrade to 1.9.1

Changes in this release:

751b09390 Bump versions to 1.9.1 for release
d0abb2133 Boost python must have a library component.
e6732c220 Check for header only Boost libraries.
08429813b get_llvm_tool_names: add llvm 21
f3b383217 docs: fix pre-1.8 order for per-subproject options
a2d395498 Revert "test cases: do not pass global option on command line"
d0f82544e Revert "tests: skip test common/223 in the -Ddefault_library=... jobs"
2abdb2f0b options: put back in place 1.7 ordering of opt=value vs subp:opt=value
83cbc65c2 Fix not passing user option args to scan-build build
e746db0b4 Document internal dep support in pkgconfig.generate `requires` arg
f6ab732b7 msetup: not-found subprojects do not have known options
5d21e653a utils: make .wraplock optional
d1e219cb8 Fix Cygwin test failure.
0fc77e2d7 coredata: do not write None to cmd_line.txt
533ba5cb3 options: do not raise exception for unknown options in -U command
ae822d1d4 test cases/common/32 has header: disable undef with libcxx
c58f0d951 rustdoc: skip --crate-type option
b8e0e9d3d fix transient failure on rust/12 bindgen
75cf757b7 coredata: check for per-subproject compiler and linker arguments
a747c4ea8 coredata: check for per-subproject compiler options

Drop the two backported patches.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agoxkeyboard-config: Turn absolute symlinks into relative
Khem Raj [Wed, 17 Sep 2025 20:41:10 +0000 (13:41 -0700)] 
xkeyboard-config: Turn absolute symlinks into relative

Native builds encode symlinks pointing to absolute paths into
build native sysroot which gets into sstate artifacts and an error
is rightly generated

ERROR: xkeyboard-config-native-2.45-r0 do_populate_sysroot: sstate found an absolute path symlink /mnt/b/yoe/master/build/tmp/work/x86_64-linux/xkeyboard-config-native/2.45/sysroot-destdir/mnt/b/yoe/master/build/tmp/work/x86_64-linux/xkeyboard-config-native/2.45/recipe-sysroot-native/usr/share/X11/xkb pointing at /mnt/b/yoe/master/build/tmp/work/x86_64-linux/xkeyboard-config-native/2.45/recipe-sysroot-native/usr/share/xkeyboard-config-2. Please replace this with a relative link.
ERROR: xkeyboard-config-native-2.45-r0 do_populate_sysroot: Failing task due to absolute path symlinks

Turn these symlinks to be relative inside D which will reflect
correclty without need for absolute path prefix

Suggested-by: Zoltán Böszörményi <zboszor@gmail.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agobuild-appliance-image: Update to master head revision
Richard Purdie [Wed, 1 Oct 2025 16:43:18 +0000 (17:43 +0100)] 
build-appliance-image: Update to master head revision

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agorust-target-config: Update the data layout for ppc64 targets.
Deepesh Varatharajan [Mon, 29 Sep 2025 13:07:59 +0000 (06:07 -0700)] 
rust-target-config: Update the data layout for ppc64 targets.

Rust commit for updating the data layouts:
https://github.com/rust-lang/rust/commit/0680155a171d47772302fc4da4cbbfa466378d81

Signed-off-by: Deepesh Varatharajan <Deepesh.Varatharajan@windriver.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
3 weeks agolinux-yocto/6.12: update CVE exclusions (6.12.49)
Bruce Ashfield [Mon, 29 Sep 2025 18:59:30 +0000 (14:59 -0400)] 
linux-yocto/6.12: update CVE exclusions (6.12.49)

Data pulled from: https://github.com/CVEProject/cvelistV5

    1/1 [
        Author: cvelistV5 Github Action
        Email: github_action@example.com
        Subject: 1 changes (1 new | 0 updated): - 1 new CVEs: CVE-2025-11135 - 0 updated CVEs:
        Date: Mon, 29 Sep 2025 01:40:51 +0000

    ]

Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agolinux-yocto/6.12: update to v6.12.49
Bruce Ashfield [Mon, 29 Sep 2025 18:59:29 +0000 (14:59 -0400)] 
linux-yocto/6.12: update to v6.12.49

Updating linux-yocto/6.12 to the latest korg -stable release that comprises
the following commits:

    da274362a7bd Linux 6.12.49
    6553fdf0f7d4 minmax.h: remove some #defines that are only expanded once
    cf5fe0b36f10 minmax.h: simplify the variants of clamp()
    5c2b06b31da6 minmax.h: move all the clamp() definitions after the min/max() ones
    5f6818002047 minmax.h: use BUILD_BUG_ON_MSG() for the lo < hi test in clamp()
    e94ce277b568 minmax.h: reduce the #define expansion of min(), max() and clamp()
    ab58f71b8fdf minmax.h: update some comments
    d9c5ccf6b69b minmax.h: add whitespace around operators and after commas
    63fd831aa5dc rtc: pcf2127: fix SPI command byte for PCF2131 backport
    207fa0d49522 xhci: dbc: Fix full DbC transfer ring after several reconnects
    490f1ca013b5 xhci: dbc: decouple endpoint allocation from initialization
    9b28ef1e4cc0 usb: xhci: remove option to change a default ring's TRB cycle bit
    e5051c055926 usb: xhci: introduce macro for ring segment list iteration
    c839be6df4da mptcp: pm: nl: announce deny-join-id0 flag
    34f351c0fa95 mm/gup: check ref_count instead of lru before migration
    096c5b1fde51 mm: add folio_expected_ref_count() for reference count calculation
    23da4e0bb2a3 vmxnet3: unregister xdp rxq info in the reset path
    ca8938704142 KVM: SVM: Set/clear SRSO's BP_SPEC_REDUCE on 0 <=> 1 VM count transitions
    e5a3331a2e98 x86/bugs: KVM: Add support for SRSO_MSR_FIX
    f9c6aec2a6dd x86/bugs: Add SRSO_USER_KERNEL_NO support
    0a3ac13d8686 platform/x86: asus-wmi: Re-add extra keys to ignore_key_wlan quirk
    7228ed156377 platform/x86: asus-wmi: Fix ROG button mapping, tablet mode on ASUS ROG Z13
    a89c34babc2e io_uring: fix incorrect io_kiocb reference in io_link_skb
    922338efaad6 smb: client: fix smbdirect_recv_io leak in smbd_negotiate() error path
    24c1106504c6 crypto: af_alg - Set merge to zero early in af_alg_sendmsg
    6ae90a2baf92 smb: client: let smbd_destroy() call disable_work_sync(&info->post_send_credits_work)
    b814660334bb smb: client: fix filename matching of deferred files
    ea5cbcecd54d drm/xe: Fix a NULL vs IS_ERR() in xe_vm_add_compute_exec_queue()
    f108c98c7005 drm: bridge: cdns-mhdp8546: Fix missing mutex unlock on error path
    0da73f782769 drm: bridge: anx7625: Fix NULL pointer dereference with early IRQ
    503de75db426 drm/xe/tile: Release kobject for the failure path
    22814abfd961 ASoC: Intel: catpt: Expose correct bit depth to userspace
    b51ded0f604e ASoC: SOF: Intel: hda-stream: Fix incorrect variable used in error message
    e5c10cec7a1b ASoC: wm8974: Correct PLL rate rounding
    987b1d5e39b9 ASoC: wm8940: Correct typo in control name
    ccba708502b6 ASoC: wm8940: Correct PLL rate rounding
    695673eb5711 io_uring/kbuf: drop WARN_ON_ONCE() from incremental length check
    93e9d0293d3b io_uring/msg_ring: kill alloc_cache for io_kiocb allocations
    e6b2b4a0ffd8 io_uring: include dying ring in task_work "should cancel" state
    d7a38ee4f0be io_uring: backport io_should_terminate_tw()
    9b71bfefc405 io_uring/cmd: let cmds to know about dying task
    9ffc5f132a35 ALSA: hda/realtek: Fix mute led for HP Laptop 15-dw4xx
    acab5c56a6fa selftests: mptcp: avoid spurious errors on TCP disconnect
    2b5b0674686f selftests: mptcp: connect: catch IO errors on listen side
    ca261278c6b7 mptcp: propagate shutdown to subflows when possible
    60b07b9f6093 rds: ib: Increment i_fastreg_wrs before bailing out
    21ba85d9d508 net: rfkill: gpio: Fix crash due to dereferencering uninitialized pointer
    232e74984061 drm/amd/display: Allow RX6xxx & RX7700 to invoke amdgpu_irq_get/put
    2ae6d79a769b KVM: SVM: Sync TPR from LAPIC into VMCB::V_TPR even if AVIC is active
    bd5524ec7155 mmc: mvsdio: Fix dma_unmap_sg() nents value
    6dbac7d814a7 ASoC: qcom: q6apm-lpass-dais: Fix missing set_fmt DAI op for I2S
    9c534dbfd172 ASoC: qcom: q6apm-lpass-dais: Fix NULL pointer dereference if source graph failed
    d1c96316e4c1 ASoC: qcom: audioreach: Fix lpaif_type configuration for the I2S interface
    bf46ed89bf24 btrfs: tree-checker: fix the incorrect inode ref size check
    cd92c8ab336c iommu/amd/pgtbl: Fix possible race while increase page table level
    c023b4f4444f iommu/vt-d: Fix __domain_mapping()'s usage of switch_to_super_page()
    1e68a5f046e9 LoongArch: Check the return value when creating kobj
    05a76baf2700 LoongArch: Align ACPI structures if ARCH_STRICT_ALIGN enabled
    382f5ff57010 LoongArch: vDSO: Check kcalloc() result in init_vdso()
    040f278cc1a5 LoongArch: Fix unreliable stack for live patching
    b6f29fa5f603 objtool/LoongArch: Mark special atomic instruction as INSN_BUG type
    89d40cc647da objtool/LoongArch: Mark types based on break immediate code
    1766f14c8f97 LoongArch: Update help info of ARCH_STRICT_ALIGN
    cbb8cd66d0bc mm: revert "mm: vmscan.c: fix OOM on swap stress test"
    f6e161f3fa99 gup: optimize longterm pin_user_pages() for large folio
    f8f64254bca5 dm-stripe: fix a possible integer overflow
    cb58eaad2235 dm-raid: don't set io_min and io_opt for raid1
    7061e566ce5d power: supply: bq27xxx: restrict no-battery detection to bq27000
    f91359651678 power: supply: bq27xxx: fix error return in case of no bq27000 hdq battery
    9aee87da5572 crypto: af_alg - Disallow concurrent writes in af_alg_sendmsg
    1adc72411f3d nilfs2: fix CFI failure when accessing /sys/fs/nilfs2/features/*
    9644798294c7 ksmbd: smbdirect: verify remaining_data_length respects max_fragmented_recv_size
    8be498fcbd5b ksmbd: smbdirect: validate data_offset and data_length field of smb_direct_data_transfer
    e97c45c770f5 perf/x86/intel: Fix crash in icl_update_topdown_event()
    ff27e23b311f octeontx2-pf: Fix use-after-free bugs in otx2_sync_tstamp()
    6e33a7eed587 cnic: Fix use-after-free bugs in cnic_delete_task
    acf8d06b8b97 net: liquidio: fix overflow in octeon_init_instr_queue()
    f07c925bb70e Revert "net/mlx5e: Update and set Xon/Xoff upon port speed set"
    208640e6225c tls: make sure to abort the stream if headers are bogus
    fa4749c06564 tcp: Clear tcp_sk(sk)->fastopen_rsk in tcp_disconnect().
    0c691ea3852c octeon_ep: fix VF MAC address lifecycle handling
    4c0bfb2dc6ab bonding: don't set oif to bond dev when getting NS target destination
    d1f3db4e7a3b net/mlx5e: Harden uplink netdev access against device unbind
    bec504867acc igc: don't fail igc_probe() on LED setup error
    610332f7ac20 i40e: remove redundant memory barrier when cleaning Tx descs
    80555adb5c89 ice: fix Rx page leak on multi-buffer frames
    1644ee7696f3 ice: store max_frame and rx_buf_len only in ice_rx_ring
    3e3be7bbe4a5 net: natsemi: fix `rx_dropped` double accounting on `netif_rx()` failure
    13e7a6e96076 selftests: mptcp: sockopt: fix error messages
    10e54bf7cb6e mptcp: tfo: record 'deny join id0' info
    bb7a3f09e9d4 selftests: mptcp: userspace pm: validate deny-join-id0 flag
    7f5b09cc84e0 mptcp: set remote_deny_join_id0 on SYN recv
    9a958802080c bonding: set random address only when slaves already exist
    660b2a8f5a30 qed: Don't collect too many protection override GRC elements
    5f445eb25990 net/tcp: Fix a NULL pointer dereference when using TCP-AO with TCP_REPAIR
    79320035973f dpaa2-switch: fix buffer pool seeding for control traffic
    3112c70b2e01 um: Fix FD copy size in os_rcv_fd_msg()
    00e98b5a6903 um: virtio_uml: Fix use-after-free after put_device in probe
    9c416e76a57f btrfs: fix invalid extref key setup when replaying dentry
    ded4d207a320 cgroup: split cgroup_destroy_wq into 3 workqueues
    eed66faed623 pcmcia: omap_cf: Mark driver struct with __refdata to prevent section mismatch
    8df33f4d4a0b wifi: mac80211: fix incorrect type for ret
    32adb020b0c3 wifi: mac80211: increase scan_ies_len for S1G
    814952c1b1ff ALSA: firewire-motu: drop EPOLLOUT from poll return values as write is not supported
    b146e0434feb nvme: fix PI insert on write
    2203ef417044 wifi: wilc1000: avoid buffer overflow in WID string configuration

Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agolinux-yocto/6.16: update CVE exclusions (6.16.9)
Bruce Ashfield [Mon, 29 Sep 2025 18:59:28 +0000 (14:59 -0400)] 
linux-yocto/6.16: update CVE exclusions (6.16.9)

Data pulled from: https://github.com/CVEProject/cvelistV5

    1/1 [
        Author: cvelistV5 Github Action
        Email: github_action@example.com
        Subject: 1 changes (1 new | 0 updated): - 1 new CVEs: CVE-2025-11135 - 0 updated CVEs:
        Date: Mon, 29 Sep 2025 01:40:51 +0000

    ]

Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agolinux-yocto/6.16: update to v6.16.9
Bruce Ashfield [Mon, 29 Sep 2025 18:59:27 +0000 (14:59 -0400)] 
linux-yocto/6.16: update to v6.16.9

Updating linux-yocto/6.16 to the latest korg -stable release that comprises
the following commits:

    b0d8acc71198 Linux 6.16.9
    f9fc28d1451b samples/damon/prcl: avoid starting DAMON before initialization
    40108f69c372 clk: sunxi-ng: mp: Fix dual-divider clock rate readback
    f9fa006e3527 samples/damon/mtier: avoid starting DAMON before initialization
    8f7dd196be59 samples/damon: change enable parameters to enabled
    1338fb5d576a samples/damon/prcl: fix boot time enable crash
    941452199941 dt-bindings: serial: 8250: move a constraint
    95c9489abfc4 dt-bindings: serial: 8250: spacemit: set clocks property as required
    ceee5585720a dt-bindings: serial: 8250: allow clock 'uartclk' and 'reg' for nxp,lpc1850-uart
    fb0b1ef7055b mptcp: pm: nl: announce deny-join-id0 flag
    46d6ff0da889 platform/x86: asus-wmi: Re-add extra keys to ignore_key_wlan quirk
    1aa91f3d4f16 platform/x86: asus-wmi: Fix ROG button mapping, tablet mode on ASUS ROG Z13
    50a98ce1ea69 io_uring: fix incorrect io_kiocb reference in io_link_skb
    0991418bf98f smb: client: fix smbdirect_recv_io leak in smbd_negotiate() error path
    9617c3ede9ff smb: client: fix file open check in __cifs_unlink()
    5ba113d0b049 io_uring/msg_ring: kill alloc_cache for io_kiocb allocations
    045ee26aa392 crypto: af_alg - Set merge to zero early in af_alg_sendmsg
    3fabb1236f2e smb: client: let smbd_destroy() call disable_work_sync(&info->post_send_credits_work)
    ac6fbc3d0030 smb: client: use disable[_delayed]_work_sync in smbdirect.c
    a39e32f03183 smb: client: fix filename matching of deferred files
    581fb78e0388 smb: client: let recv_done verify data_offset, data_length and remaining_data_length
    cbda551b90a7 smb: client: make use of struct smbdirect_recv_io
    089ea68aae42 smb: smbdirect: introduce struct smbdirect_recv_io
    86b4bddb8292 smb: client: make use of smbdirect_socket->recv_io.expected
    f2ffba55b686 smb: smbdirect: introduce smbdirect_socket.recv_io.expected
    dd1a415dcfd5 drm/xe/guc: Set RCS/CCS yield policy
    97207a4fed53 drm/xe/guc: Enable extended CAT error reporting
    56f34936bf8c drm/xe: Fix error handling if PXP fails to start
    09b473a80c1c ALSA: usb: qcom: Fix false-positive address space check
    b45cabfa717d drm/xe: Fix a NULL vs IS_ERR() in xe_vm_add_compute_exec_queue()
    e3fe0101463a drm: bridge: cdns-mhdp8546: Fix missing mutex unlock on error path
    1a7ea294d57f drm: bridge: anx7625: Fix NULL pointer dereference with early IRQ
    e6a1df8cf275 drm/xe/pf: Drop rounddown_pow_of_two fair LMEM limitation
    f32a0226e848 drm/xe/tile: Release kobject for the failure path
    65c5cfbd6d93 ASoC: amd: acp: Fix incorrect retrival of acp_chip_info
    f7229775e41d iommu/amd: Fix alias device DTE setting
    34f3a9e04e73 ASoC: Intel: catpt: Expose correct bit depth to userspace
    f6433733326b ASoC: SDCA: Fix return value in sdca_regmap_mbq_size()
    9ff967d7e8d9 ASoC: SOF: Intel: hda-stream: Fix incorrect variable used in error message
    cd59ca8f75db ASoC: codec: sma1307: Fix memory corruption in sma1307_setting_loaded()
    7c28b31b2209 ASoC: wm8974: Correct PLL rate rounding
    badf6143198e ASoC: wm8940: Correct typo in control name
    7a372ac1e890 ASoC: wm8940: Correct PLL rate rounding
    519b95c74f40 ALSA: hda/realtek: Fix mute led for HP Laptop 15-dw4xx
    ee8d393af37e selftests: mptcp: avoid spurious errors on TCP disconnect
    47f8d4403995 selftests: mptcp: connect: catch IO errors on listen side
    80d38ea27a1e mptcp: propagate shutdown to subflows when possible
    c62000bf27db rds: ib: Increment i_fastreg_wrs before bailing out
    bc509293c9d4 crypto: ccp - Always pass in an error pointer to __sev_platform_shutdown_locked()
    27d94a2a52cb gpiolib: acpi: initialize acpi_gpio_info struct
    21a39b958b4b net: rfkill: gpio: Fix crash due to dereferencering uninitialized pointer
    6db60106a07f io_uring: include dying ring in task_work "should cancel" state
    f757ab3db457 io_uring/io-wq: fix `max_workers` breakage and `nr_workers` underflow
    898aaf78480b drm/amd: Only restore cached manual clock settings in restore if OD enabled
    20f87640ebe7 drm/amd/display: Allow RX6xxx & RX7700 to invoke amdgpu_irq_get/put
    1ff89f5627ef drm/amdgpu: suspend KFD and KGD user queues for S0ix
    b477c5668ec6 drm/amdkfd: add proper handling for S0ix
    0e2db61cc589 KVM: SVM: Sync TPR from LAPIC into VMCB::V_TPR even if AVIC is active
    c0603b80436d x86/sev: Guard sev_evict_cache() with CONFIG_AMD_MEM_ENCRYPT
    79a9ba8da904 mmc: sdhci-uhs2: Fix calling incorrect sdhci_set_clock() function
    7650c994ced2 mmc: sdhci-pci-gli: GL9767: Fix initializing the UHS-II interface during a power-on
    7186d8e8bd96 mmc: sdhci: Move the code related to setting the clock from sdhci_set_ios_common() into sdhci_set_ios()
    d0b7ff384b7a mmc: mvsdio: Fix dma_unmap_sg() nents value
    66e6d1c92806 ASoC: qcom: q6apm-lpass-dais: Fix missing set_fmt DAI op for I2S
    cc336b242ea7 ASoC: qcom: q6apm-lpass-dais: Fix NULL pointer dereference if source graph failed
    59c4accddfeb ASoC: qcom: audioreach: Fix lpaif_type configuration for the I2S interface
    8276c97dccee ASoC: SDCA: Add quirk for incorrect function types for 3 systems
    417ed00d48e7 btrfs: tree-checker: fix the incorrect inode ref size check
    359613f2fa00 iommu/s390: Make attach succeed when the device was surprise removed
    17a58caf3863 iommu/s390: Fix memory corruption when using identity domain
    7d462bdecb7d iommu/amd/pgtbl: Fix possible race while increase page table level
    b0c0e231060a iommu/amd: Fix ivrs_base memleak in early_amd_iommu_init()
    7ff7d16649b2 iommu/vt-d: Fix __domain_mapping()'s usage of switch_to_super_page()
    1c731284374a LoongArch: KVM: Fix VM migration failure with PTW enabled
    960eedb14caf LoongArch: KVM: Avoid copy_*_user() with lock hold in kvm_pch_pic_regs_access()
    55ba91b4e04d LoongArch: KVM: Avoid copy_*_user() with lock hold in kvm_eiointc_sw_status_access()
    105605ca76e9 LoongArch: KVM: Avoid copy_*_user() with lock hold in kvm_eiointc_regs_access()
    291d4b01d3b1 LoongArch: KVM: Avoid copy_*_user() with lock hold in kvm_eiointc_ctrl_access()
    401363c839cb LoongArch: Handle jump tables options for RUST
    1967642780cf LoongArch: Make LTO case independent in Makefile
    db65fea5f0aa LoongArch: Check the return value when creating kobj
    5f2b63a398ed LoongArch: Align ACPI structures if ARCH_STRICT_ALIGN enabled
    a417571950f9 LoongArch: vDSO: Check kcalloc() result in init_vdso()
    2feeecd7c685 LoongArch: Fix unreliable stack for live patching
    5dbbc7b04c14 objtool/LoongArch: Mark special atomic instruction as INSN_BUG type
    e0aefa8f4612 objtool/LoongArch: Mark types based on break immediate code
    953138ff0ff6 LoongArch: Update help info of ARCH_STRICT_ALIGN
    1eda9ab8da6b mm: folio_may_be_lru_cached() unless folio_test_large()
    fb4e6d587a27 mm: revert "mm: vmscan.c: fix OOM on swap stress test"
    d0c8ba94cb70 mm/gup: local lru_add_drain() to avoid lru_add_drain_all()
    163843e8c8f3 gup: optimize longterm pin_user_pages() for large folio
    3958f9ec7251 mm: revert "mm/gup: clear the LRU flag of a page before adding to LRU batch"
    fdac0a3f58c0 mm/gup: check ref_count instead of lru before migration
    ee27658c239b dm-stripe: fix a possible integer overflow
    ba3a78db47ec dm-raid: don't set io_min and io_opt for raid1
    e8f496001e0c btrfs: initialize inode::file_extent_tree after i_mode has been set
    8ae09726773a Revert "sched_ext: Skip per-CPU tasks in scx_bpf_reenqueue_local()"
    a4ee54e68282 power: supply: bq27xxx: restrict no-battery detection to bq27000
    d18d7035ecb8 power: supply: bq27xxx: fix error return in case of no bq27000 hdq battery
    45bcf60fe49b crypto: af_alg - Disallow concurrent writes in af_alg_sendmsg
    7b7361da9e7f nilfs2: fix CFI failure when accessing /sys/fs/nilfs2/features/*
    ff750e9f2c4d zram: fix slot write race condition
    c64b915bb3d9 ksmbd: smbdirect: verify remaining_data_length respects max_fragmented_recv_size
    529b121b00a6 ksmbd: smbdirect: validate data_offset and data_length field of smb_direct_data_transfer
    5ca20bb7b4bd octeontx2-pf: Fix use-after-free bugs in otx2_sync_tstamp()
    0627e1481676 cnic: Fix use-after-free bugs in cnic_delete_task
    45f71f4ac4aa net: liquidio: fix overflow in octeon_init_instr_queue()
    6e1675c8b612 net: clear sk->sk_ino in sk_set_socket(sk, NULL)
    53222fc8ebbc Revert "net/mlx5e: Update and set Xon/Xoff upon port speed set"
    61ca2da5fb8f tls: make sure to abort the stream if headers are bogus
    ae313d14b45e tcp: Clear tcp_sk(sk)->fastopen_rsk in tcp_disconnect().
    0357a37b4a8b octeon_ep: fix VF MAC address lifecycle handling
    e2019c7d3a3e bonding: don't set oif to bond dev when getting NS target destination
    948381b58298 net/mlx5e: Add a miss level for ipsec crypto offload
    8df354eb2dd6 net/mlx5e: Harden uplink netdev access against device unbind
    805c7df4faca doc/netlink: Fix typos in operation attributes
    f05e82d85532 igc: don't fail igc_probe() on LED setup error
    3ce36b3b9c69 ixgbe: destroy aci.lock later within ixgbe_remove path
    a4a4b796dcf6 ixgbe: initialize aci.lock before it's used
    5ac700c61648 i40e: remove redundant memory barrier when cleaning Tx descs
    fcb5718ebfe7 ice: fix Rx page leak on multi-buffer frames
    4f21a0b3c8c2 net: natsemi: fix `rx_dropped` double accounting on `netif_rx()` failure
    5735f1fb81a8 selftests: mptcp: sockopt: fix error messages
    7f501faddb3c mptcp: tfo: record 'deny join id0' info
    fa9a7f272b01 selftests: mptcp: userspace pm: validate deny-join-id0 flag
    f679e3dc6c01 mptcp: set remote_deny_join_id0 on SYN recv
    f3439ed58636 bonding: set random address only when slaves already exist
    8b1dc0217f96 net: dst_metadata: fix IP_DF bit not extracted from tunnel headers
    70affe82e38f qed: Don't collect too many protection override GRC elements
    1c24b132c119 octeon_ep: Validate the VF ID
    71571e187106 rxrpc: Fix untrusted unsigned subtract
    8ac99c57029e rxrpc: Fix unhandled errors in rxgk_verify_packet_integrity()
    64e76fcb5311 dpll: fix clock quality level reporting
    993b734d31ab net/tcp: Fix a NULL pointer dereference when using TCP-AO with TCP_REPAIR
    d0c3f85a041a dpaa2-switch: fix buffer pool seeding for control traffic
    e1fa8f786c8d net/mlx5: Not returning mlx5_link_info table when speed is unknown
    4c2c59cc838c um: Fix FD copy size in os_rcv_fd_msg()
    c2ff91255e01 um: virtio_uml: Fix use-after-free after put_device in probe
    856e039ded02 smb: server: let smb_direct_writev() respect SMB_DIRECT_MAX_SEND_SGES
    09ea55d02c89 pcmcia: omap_cf: Mark driver struct with __refdata to prevent section mismatch
    0835c4592344 wifi: mac80211: fix incorrect type for ret
    0dbad5f5549e wifi: mac80211: increase scan_ies_len for S1G
    745418fc8229 wifi: mt76: do not add non-sta wcid entries to the poll list
    778a062c131e ALSA: firewire-motu: drop EPOLLOUT from poll return values as write is not supported
    0cdf320eb46c nvme: fix PI insert on write
    ae50f8562306 wifi: wilc1000: avoid buffer overflow in WID string configuration
    d06a83038bd6 perf maps: Ensure kmap is set up for all inserts
    314a92e5950f btrfs: zoned: fix incorrect ASSERT in btrfs_zoned_reserve_data_reloc_bg()
    eff2bb3b300b btrfs: fix invalid extref key setup when replaying dentry
    05e0b03447cf cgroup: split cgroup_destroy_wq into 3 workqueues

Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agooe-setup-build: fix dash support
Yoann Congal [Fri, 26 Sep 2025 14:53:05 +0000 (16:53 +0200)] 
oe-setup-build: fix dash support

Being minimalist, dash does not support the (non-POSIX) feature of
passing an argument while sourcing a script. Like in
  . <some path>/oe-init-build-env <build dir>

With dash, one must instead use:
  cd <some path>
  set <build dir>       # puts <build dir> in $1
  . ./oe-init-build-env # can only be called from its directory in dash

oe-setup-build generate a sourcable "init-build-env" script, this script
must use the above snippet to be used in dash.

Signed-off-by: Yoann Congal <yoann.congal@smile.fr>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agoexpat: upgrade to 2.7.3
Ross Burton [Fri, 26 Sep 2025 16:32:31 +0000 (17:32 +0100)] 
expat: upgrade to 2.7.3

Security fixes:
- Fix alignment of internal allocations for some non-amd64 architectures
  (e.g. sparc32); fixes up on the fix to CVE-2025-59375 from #1034 (of
  Expat 2.7.2 and related backports)

- Fix a class of false positives where input should have been rejected
  with error XML_ERROR_ASYNC_ENTITY; regression from CVE-2024-8176 fix
  pull request #973 (of Expat 2.7.0 and related backports). Please check
  the added unit tests for example documents.

Other changes:
- Prove and regression-proof absence of integer overflow from function
  expat_realloc
- Remove "harmless" cast that truncated a size_t to unsigned
- Autotools: Remove "ln -s" discovery
- docs: Be consistent with use of floating point around
  XML_SetAllocTrackerMaximumAmplification
- docs: Make it explicit that XML_GetCurrentColumnNumber starts at 0
- docs: Better integrate the effect of the activation thresholds
- docs: Fix an in-comment typo in expat.h
- docs: Fix a typo in README.md
- docs: Improve change log of release 2.7.2
- xmlwf: Resolve use of functions XML_GetErrorLineNumber and
  XML_GetErrorColumnNumber
- Windows: Normalize .bat files to CRLF line endings
- Version info bumped from 12:0:11 (libexpat*.so.1.11.0) to 12:1:11
  (libexpat*.so.1.11.1); see https://verbump.de/ for what these numbers
  do

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agoclasses/cmake: unset LDFLAGS in toolchain-native.bbclass
Ross Burton [Fri, 26 Sep 2025 14:33:39 +0000 (15:33 +0100)] 
classes/cmake: unset LDFLAGS in toolchain-native.bbclass

If a recipe is using toolchain-native.cmake to build native portion in a
non-native build, the target LDFLAGS from the environment will leak into
the native build.

This was noticed as building a SDK with clang means that LDFLAGS contains
a --dynamic-loader argument, so native binaries were trying to use the
target loader.

There are several variables that are set from LDFLAGS[1] so instead of
setting them all, we can simply unset the environment variable in the
toolchain.

[1] https://cmake.org/cmake/help/latest/envvar/LDFLAGS.html

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agooeqa/bblayers.py: add tests for adding new built-in fragments
Antonin Godard [Fri, 26 Sep 2025 09:46:53 +0000 (11:46 +0200)] 
oeqa/bblayers.py: add tests for adding new built-in fragments

Add discussed in [1], our best option for customizing built-in fragments
is to pass them from a layer configuration. In short, the reason is that
our statement must be parsed before the addfragments call is parsed.We
also have to use the :append override as using += would override the
original definition of OE_FRAGMENTS_BUILTIN (since it uses a ?=
assignment).

Provide a test case for customizing built-in fragments with
meta-selftest.

[1]: https://lore.kernel.org/yocto-docs/20250925-fragments-v1-0-c9f747361fb2@bootlin.com/T/#m9f7c9f110c084eba17e0f64d8b2ac7a88af3f38e

Cc: Alexander Kanavin <alex.kanavin@gmail.com>
Signed-off-by: Antonin Godard <antonin.godard@bootlin.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agotiff: ignore CVE-2025-8851
Peter Marko [Thu, 25 Sep 2025 16:12:05 +0000 (18:12 +0200)] 
tiff: ignore CVE-2025-8851

This is fixed in v4.7.0, however cve_check cannot match it as NVD says
"Up to (excluding) 2024-08-11".

Signed-off-by: Peter Marko <peter.marko@siemens.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agotiff: upgrade 4.7.0 -> 4.7.1
Peter Marko [Thu, 25 Sep 2025 16:12:04 +0000 (18:12 +0200)] 
tiff: upgrade 4.7.0 -> 4.7.1

Removed patches included in this new release

License-Update: BSD license added based on [1]

[1] https://gitlab.com/libtiff/libtiff/-/commit/a0b623c7809ea2aa4978d5d7b7bd10e519294c78

Signed-off-by: Peter Marko <peter.marko@siemens.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agopatchtest: Use raw strings for regex patterns
Paul Barker [Wed, 24 Sep 2025 19:57:47 +0000 (20:57 +0100)] 
patchtest: Use raw strings for regex patterns

This fixes several 'SyntaxWarning: invalid escape sequence' messages
printed when running patchtest.

Cc: Trevor Gamblin <tgamblin@baylibre.com>
Signed-off-by: Paul Barker <paul@pbarker.dev>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agobitbake-layers/create: Improve help message for layerid argument
Paul Barker [Wed, 24 Sep 2025 19:57:46 +0000 (20:57 +0100)] 
bitbake-layers/create: Improve help message for layerid argument

'Layer ID' isn't a commonly used term in our help messages or
documentation, so clarify that this is the identifier used for the new
layer in BBFILE_COLLECTIONS.

Also clarify that the default is the basename of the layerdir argument
if a layer ID is not separately provided.

Signed-off-by: Paul Barker <paul@pbarker.dev>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agounzip: remove invalid UPSTREAM_CHECK_REGEX
Changqing Li [Thu, 25 Sep 2025 08:43:42 +0000 (16:43 +0800)] 
unzip: remove invalid UPSTREAM_CHECK_REGEX

After SRC_URI is changed to sourceforge, this UPSTREAM_CHECK_REGEX
become invalid, just remove it.

Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agovim: upgrade 9.1.1652 -> 9.1.1683
Divya Chellam [Thu, 25 Sep 2025 06:07:12 +0000 (11:37 +0530)] 
vim: upgrade 9.1.1652 -> 9.1.1683

Handles CVE-2025-9389

Changes between 9.1.1652 -> 9.1.1683
====================================
https://github.com/vim/vim/compare/v9.1.1652...v9.1.1683

Signed-off-by: Divya Chellam <divya.chellam@windriver.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agobaremetal-helloworld: set UPSTREAM_CHECK_COMMITS
Changqing Li [Thu, 25 Sep 2025 08:44:20 +0000 (16:44 +0800)] 
baremetal-helloworld: set UPSTREAM_CHECK_COMMITS

set UPSTREAM_CHECK_COMMITS to make new commit is reported

Test result(change SRCREV to an old one):
$devtool check-upgrade-status baremetal-helloworld
baremetal-helloworld      0.1             new commits     Alejandro Hernandez <alejandro@enedino.org> db2bf750eaef7fc0832e13ada8291343bbcc3afe

Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agocwautomacros: delete the recipe
Alexander Kanavin [Thu, 25 Sep 2025 10:53:40 +0000 (12:53 +0200)] 
cwautomacros: delete the recipe

This is a long-obsolete set of custom autotools macros;
'which' was the last consumer in core or meta-oe.

Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
3 weeks agowhich: update 2.21 -> 2.23, build with meson
Alexander Kanavin [Thu, 25 Sep 2025 10:53:39 +0000 (12:53 +0200)] 
which: update 2.21 -> 2.23, build with meson

This (and 2.22) is a first release in many years. Rather than try
to use upstream's absurdly overblown, incompatible build system
(see below), I added a small-ish meson file. This means:

- drop tweaks and dependency on cwautomacros as that is no longer used

- drop patch as configure.ac has been rewritten, and the recipe is using meson anyway

- drop --disable-iberty for the same reason

In this realease, cwautomacros has been replaced by an equally custom, weird set
of macros, written by 'which' maintainer: https://github.com/CarloWood/cwm4

- one effect of that is that autoreconf isn't happy with which's configure.ac and won't run;
one is supposed to use a custom script instead: https://github.com/CarloWood/cwm4/blob/master/scripts/bootstrap.sh

- alas, that script is not shipped in tarballs; the maintainer wants
everyone to trust their 200k configure script (hello xz backdoor)

- building from git (where the script exists) is not impossible,
but that has no version tags

All this 'special handling' for what, exactly? Five .c files to produce one
single-function executable, and one manpage. Wich should all be in coreutils
to begin with. GNU's attachment to autotools defies reason.

Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agopiglit: rename virtual/opencl-icd to virtual/libopencl1
Ryan Eatmon [Tue, 23 Sep 2025 14:21:05 +0000 (09:21 -0500)] 
piglit: rename virtual/opencl-icd to virtual/libopencl1

A change in meta-openembedded that provided virtual/opencl-icd [1]
changed the name to virtual/libopencl1.

[1] https://git.openembedded.org/meta-openembedded/commit/?id=87b44d29b6942427bfb1fab02a4564e41dcc01a0

Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agometa/lib/oe/recipeutils.py: Handle multi-repo recipes for upgrade check
Uwe Kleine-König [Tue, 23 Sep 2025 15:47:02 +0000 (17:47 +0200)] 
meta/lib/oe/recipeutils.py: Handle multi-repo recipes for upgrade check

For a recipe that uses more than one git repo there isn't a single
SRCREV variable. For example for linux-yocto there is SRCREV_machine and
SRCREV_meta and rd.getVar("SRCREV") yields "INVALID".

Luckily bb.fetch2 already handles all the details and exposes the
currently used revision in ud. So just use that.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@baylibre.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agorust: Upgrade 1.89.0 -> 1.90.0
Yash Shinde [Wed, 24 Sep 2025 09:22:06 +0000 (02:22 -0700)] 
rust: Upgrade 1.89.0 -> 1.90.0

https://blog.rust-lang.org/2025/09/18/Rust-1.90.0/

* Recent changes in rustc require 'target-c-int-width' to be an integer,
  not a string. This fixes type consistency when generating target specs.
  https://github.com/rust-lang/rust/pull/142352.

* Rebase existing patches with v1.90.0.

* Drop merged patches with rust v1.90.0.
  - backport-fix-test-string-merging.patch
  https://github.com/rust-lang/rust/commit/2d51acd2fbcbadb6f30709c5dd305494d413d388

* The "remote-test-server" bin is now generated in stage1-tools-bin dir
  rather than stage2. Update the test suite accordingly.

  The tests/{assembly/codegen} dirs are renamed to
  tests/{assembly,codegen}-llvm. Update the test suite and
  patches accordingly.
  https://github.com/rust-lang/rust/pull/144249/commits

* Use "//@ ignore-riscv64" tag for tests failing on riscv-64 instead of previous
  "only-<target_arch>" tags.

  Test results summary:

   rust v1.90.0
   +-----------+--------+---------+
   | Machine   | Passed | Ignored |
   +-----------+--------+---------+
   | arm-32    | 29,517 |  1,529  |
   | arm-64    | 29,608 |  1,471  |
   | x86-32    | 29,508 |  1,507  |
   | x86-64    | 29,903 |  1,275  |
   | riscv-64  | 29,584 |  1,494  |
   +-----------+--------+---------+

   Test results difference (1.90 - 1.89):
   +-----------+--------+---------+
   | Machine   | Passed | Ignored |
   +-----------+--------+---------+
   | arm-32    |  +278  |   +61   |
   | arm-64    |  +279  |   +59   |
   | x86-32    |  +277  |   +58   |
   | x86-64    |  +419  |   +50   |
   | riscv-64  |  +280  |   +58   |
   +-----------+--------+---------+

Signed-off-by: Yash Shinde <Yash.Shinde@windriver.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agoapt: fix reproducibility issue
Hongxu Jia [Wed, 24 Sep 2025 07:43:43 +0000 (00:43 -0700)] 
apt: fix reproducibility issue

In oe-selftest of reproducibility, recipe apt has reproducibility issue:

...
2025-09-20 17:09:18,986 - oe-selftest - INFO - ======================================================================
2025-09-20 17:09:18,986 - oe-selftest - INFO - FAIL: test_reproducible_builds (reproducible.ReproducibleTests)
2025-09-20 17:09:18,986 - oe-selftest - INFO - ----------------------------------------------------------------------
2025-09-20 17:09:18,986 - oe-selftest - INFO - Traceback (most recent call last):
  File "/buildarea5/hjia/contrib/openembedded-core-contrib/meta/lib/oeqa/selftest/cases/reproducible.py", line 406, in test_reproducible_builds
    self.fail('\n'.join(fails))
AssertionError: The following deb packages are different and not in exclusion list:
build-st/reproducibleB-extended/tmp/deploy/deb/./x86-64-v3/apt-dbg_3.0.3-r0_amd64.deb
build-st/reproducibleB-extended/tmp/deploy/deb/./x86-64-v3/apt_3.0.3-r0_amd64.deb
The following ipk packages are different and not in exclusion list:
build-st/reproducibleB-extended/tmp/deploy/ipk/./x86-64-v3/apt-dbg_3.0.3-r0_x86-64-v3.ipk
build-st/reproducibleB-extended/tmp/deploy/ipk/./x86-64-v3/apt_3.0.3-r0_x86-64-v3.ipk
The following rpm packages are different and not in exclusion list:
build-st/reproducibleB-extended/tmp/deploy/rpm/./x86_64_v3/apt-3.0.3-r0.x86_64_v3.rpm
build-st/reproducibleB-extended/tmp/deploy/rpm/./x86_64_v3/apt-dbg-3.0.3-r0.x86_64_v3.rpm
...

Due to upstream apt commit [1], it calls `id -gn root' in cmake
to set ROOT_GROUP if not define. The output of `id -gn root' may
differ in the compared builds which caused reproducibility issue
...
|build-st/reproducibleB-extended/tmp/work/x86-64-v3-oe-linux/apt/
3.0.3/temp/log.do_configure:124:-- Found root group: wheel
...
|build-st/reproducibleA/tmp/work/x86-64-v3-oe-linux/apt/3.0.3/
temp/log.do_configure:124:-- Found root group: root
...

Explicitly set ROOT_GROUP = "root" for cmake other than call `id -gn root'
to support reproducibility

[1] https://salsa.debian.org/apt-team/apt/-/commit/6f1f3c9afdb6ade6a7be110b90c8fc9e603254cf

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agoqemuriscv: Use RVA23S64 profile for cpu
Khem Raj [Wed, 24 Sep 2025 06:24:50 +0000 (23:24 -0700)] 
qemuriscv: Use RVA23S64 profile for cpu

Enables RVV extensions uniformly across gcc and clang
as it is mandatory in RVA23 spec.

https://github.com/riscv/riscv-profiles/blob/main/src/rva23-profile.adoc

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agolibwpe: upgrade 1.16.2 -> 1.16.3
Moritz Haase [Wed, 24 Sep 2025 05:55:36 +0000 (07:55 +0200)] 
libwpe: upgrade 1.16.2 -> 1.16.3

Release notes are available at [0]. The new release contains a CMake-related
patch that we can now drop.

[0]: https://github.com/WebPlatformForEmbedded/libwpe/releases/tag/1.16.3

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>
4 weeks agosystemd: use if..then in prerm scriptlet
Daniel McGregor [Tue, 23 Sep 2025 19:01:46 +0000 (13:01 -0600)] 
systemd: use if..then in prerm scriptlet

Using the [ test ] && foo construct in the prerm scriptlet causes
the prerm scriptlet to fail if the final test condition fails, which
with rpm prevents the removal of the package. Switch to using
if;then;fi instead, since it doesn't cause the scriptlet to fail.

Signed-off-by: Daniel McGregor <daniel.mcgregor@vecima.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agopython3-dbusmock: upgrade 0.33.0 -> 0.37.0
Trevor Gamblin [Sun, 21 Sep 2025 14:08:40 +0000 (10:08 -0400)] 
python3-dbusmock: upgrade 0.33.0 -> 0.37.0

Upstream notably removed the setup.cfg/setup.py files in this period and
updated pyproject.toml. It still uses setuptools as the build backend,
and should be compatible with recent versions.

Changelog (https://github.com/martinpitt/python-dbusmock/releases):

0.37.0:

- templates: Drop iio-sensors-proxy (See #241)
- DBusTestCase: Drop obsolete {start,stop}_dbus() methods
- Fix dbus-daemon startup race condition (Debian#1109272)

0.36.0:

- mockobject: Fix _wrap_in_dbus_variant for Struct and Dict types (thanks Sebastian Wick)
- Drop setup.{cfg,py} and RHEL 9 support, move to pybuild (rhbz#2377609)
- Drop iio-sensor-proxy tests, the template is broken (see #241)

0.35.0:

- modemmanager: Add operator code (thanks Guido Günther)
- modemmanager: Allow to set CellBroadcast channel list (thanks Guido Günther)

0.34.3:

- tests: Relax libnotify expected format for libnotify 0.8.4

0.34.2:

- spec: Adjust %autosetup to PEP-625 tarball top-level directory

0.34.1:

- spec: Adjust Source URL to PEP-625 tarball name

0.34.0:

- modemmanager: Add cell broadcast messages (thanks Guido Günther)
- Move release builds to pybuild for PEP-625 compatibility

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>
4 weeks agopython3-pyparsing: upgrade 3.2.3 -> 3.2.4
Trevor Gamblin [Fri, 19 Sep 2025 16:48:02 +0000 (12:48 -0400)] 
python3-pyparsing: upgrade 3.2.3 -> 3.2.4

Release notes (https://github.com/pyparsing/pyparsing/releases):

- Barring any catastrophic bugs in this release, this will be the last
  release in the 3.2.x line. The next release, 3.3.0, will begin
  emitting DeprecationWarnings when the pre-PEP8 methods are used (see
  header notes above for more information, including available
  automation for converting any existing code using pyparsing with the
  old names).
- Fixed bug when using a copy of a Word expression (either by using the
  explicit copy() method, or attaching a results name), and setting a
  new expression name, a raised ParseException still used the original
  expression name. Also affected Regex expressions with as_match or
  as_group_list = True. Reported by Waqas Ilyas, in Issue #612 - good
  catch!
- Fixed type annotation for replace_with, to accept Any type. Fixes
  Issue #602, reported by esquonk.
- Added locking around potential race condition in
  ParserElement.reset_cache, as well as other cache-related methods.
  Fixes Issue #604, reported by CarlosDescalziIM.
- Substantial update to docstrings and doc generation in preparation for
  3.3.0, great effort by FeRD, thanks!
- Notable addition by FeRD to convert docstring examples to work with
  doctest! This was long overdue, thanks so much!

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>
4 weeks agopython3-hypothesis: upgrade 6.138.15 -> 6.139.2
Trevor Gamblin [Fri, 19 Sep 2025 16:48:01 +0000 (12:48 -0400)] 
python3-hypothesis: upgrade 6.138.15 -> 6.139.2

Changelog: (https://hypothesis.readthedocs.io/en/latest/changelog.html#changelog):

6.139.2:

- Internal refactoring for new lint rules.

6.139.1:

- Fixed another typo in error message around function-scoped fixtures.

6.139.0:

- Add get_current_profile_name(), which returns the name of the current settings profile.

6.138.17:

- Fixed typo in error message around function-scoped fixtures.

6.138.16:

- Improved error message for DeadlineExceeded.

Reproducibility OK.

ptests look OK:

|============================================================================
|Testsuite summary
|# TOTAL: 5
|# PASS: 5
|# SKIP: 0
|# XFAIL: 0
|# FAIL: 0
|# XPASS: 0
|# ERROR: 0
|DURATION: 1
|END: /usr/lib/python3-hypothesis/ptest
|2025-09-19T15:53
|STOP: ptest-runner
|TOTAL: 1 FAIL: 0

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>
4 weeks agopython3-trove-classifiers: upgrade 2025.9.8.13 -> 2025.9.11.17
Trevor Gamblin [Fri, 19 Sep 2025 16:48:00 +0000 (12:48 -0400)] 
python3-trove-classifiers: upgrade 2025.9.8.13 -> 2025.9.11.17

Release notes (https://github.com/pypa/trove-classifiers/releases):

2025.9.11.17:

- Add support for Odoo 19.0 in classifiers (#225)

2025.9.9.12:

- Add classifier: Framework InvenTree (#223)

* Add classifier: Framework InvenTree

* fix style issue

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>
4 weeks agopython3-numpy: upgrade 2.3.2 -> 2.3.3
Trevor Gamblin [Fri, 19 Sep 2025 16:47:59 +0000 (12:47 -0400)] 
python3-numpy: upgrade 2.3.2 -> 2.3.3

Release notes (https://github.com/numpy/numpy/releases/tag/v2.3.3):

The NumPy 2.3.3 release is a patch release split between a number of maintenance
updates and bug fixes. This release supports Python versions 3.11-3.14. Note
that the 3.14.0 final is currently expected in Oct, 2025. This release is based
on 3.14.0rc2.

Reproducibility seems OK.

ptests look OK:

|============================================================================
|Testsuite summary
|# TOTAL: 48954
|# PASS: 46490
|# SKIP: 2426
|# XFAIL: 33
|# FAIL: 0
|# XPASS: 5
|# ERROR: 0
|DURATION: 169
|END: /usr/lib/python3-numpy/ptest
|2025-09-19T15:20
|STOP: ptest-runner
|TOTAL: 1 FAIL: 0

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>
4 weeks agomaintainers.inc: add self for python recipes
Trevor Gamblin [Fri, 19 Sep 2025 16:47:58 +0000 (12:47 -0400)] 
maintainers.inc: add self for python recipes

Add myself as maintainer for some newly-unassigned Python recipes.
That'll allow them to be picked up on my upgrade checker.

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>
4 weeks agocmake: upgrade 4.1.0 -> 4.1.1
Moritz Haase [Tue, 23 Sep 2025 06:47:53 +0000 (08:47 +0200)] 
cmake: upgrade 4.1.0 -> 4.1.1

Release notes are available at [0].

[0]: https://cmake.org/cmake/help/v4.1/release/4.1.html#id1

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>
4 weeks agolibdnf: remove obsolete path patch
Ross Burton [Mon, 22 Sep 2025 14:20:31 +0000 (15:20 +0100)] 
libdnf: remove obsolete path patch

We needed to remove the build path from the source tree because it was
embedded into libdnf/dnf-repo.cpp.  However, this was replaced with an
environment variable in 0.67.0 onwards[1].

[1] libdnf 04d7fba0 ("Use environment variable in unittest instead of ugly hack in libdnf")

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agolibdnf: remove non-functional gtk-doc support
Ross Burton [Mon, 22 Sep 2025 14:20:30 +0000 (15:20 +0100)] 
libdnf: remove non-functional gtk-doc support

The gtk-doc integration doesn't actually work, as it isn't built by
default and the targets need to be triggered manually.

As clearly nobody was using it, and upstream appears to prefer doxygen
for API documentation, remove the gtk-doc support.

This also means adding gettext to the dependencies, as this was being
pulled in via gtk-doc before but is actually needed for the translations.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agolibdnf: remove obsolete gobject-introspection support
Ross Burton [Mon, 22 Sep 2025 14:20:29 +0000 (15:20 +0100)] 
libdnf: remove obsolete gobject-introspection support

The intention to remove G-I support was stated in [1] and the last few
pieces removed in [2], which were part of 0.15.0.

[1] libdnf a4abd42a ("Move libcheck dependency to tests/")
[2] libdnf e2f2862b ("[swdb]: C++ implementation with SWIG bindings.")

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agolibdnf: don't depend on libcheck
Ross Burton [Mon, 22 Sep 2025 14:20:28 +0000 (15:20 +0100)] 
libdnf: don't depend on libcheck

Apply a patch to make libcheck optional, so we don't need to depend on
it as we don't install the tests.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agoclang: no need to install tblgen binaries in nativesdk-clang
Ross Burton [Wed, 24 Sep 2025 12:08:34 +0000 (13:08 +0100)] 
clang: no need to install tblgen binaries in nativesdk-clang

These are only needed if cross-compiling clang, and if someone is
cross-compiling clang in a SDK they should build their own binaries as
these ones might not match the version of clang they're building.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agoclang: tell cross-builds of clang how to build natively
Ross Burton [Wed, 24 Sep 2025 12:08:33 +0000 (13:08 +0100)] 
clang: tell cross-builds of clang how to build natively

If we pass CROSS_TOOLCHAIN_FLAGS_NATIVE to all the builds, instead of
just nativesdk, then clang can correctly build its own native tools as
needed.

This means we can stop manually installing clang-tidy-confusable-chars-gen
as clang-tidy will build it.  We (currently) still build the tblgen
binaries once as they're used in several recipes so there's likely value
in building those once.

Also consolidate the LLVM_TABLEGEN variables as they're not specific to
a particular class.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agoclang: improve tablegen installation
Ross Burton [Wed, 24 Sep 2025 12:08:32 +0000 (13:08 +0100)] 
clang: improve tablegen installation

Backport a patch from upstream to install lldb-tblgen, and change the
llvm-tblgen-native recipe to use `cmake --install` instead of copying
the binaries.

This ensures that CMake does any post-install steps that are needed, and
we don't install binaries that have eg RPATHs pointing at the build tree.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agoclang: remove obsolete install steps
Ross Burton [Wed, 24 Sep 2025 12:08:31 +0000 (13:08 +0100)] 
clang: remove obsolete install steps

clang-pseudo and clang-rename have both been removed from upstream, so
don't try to install them anymore:

  llvm 40c45b6b4318 ("Remove clang-rename (#108988)")
  llvm ed8f78827895 ("Remove clang-pseudo (#109154)")

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agoclang: add option for llvm-exegesis
Ross Burton [Wed, 24 Sep 2025 12:08:30 +0000 (13:08 +0100)] 
clang: add option for llvm-exegesis

Quoting from the README:

    llvm-exegesis is a benchmarking tool that accepts or generates
    snippets and can measure characteristics of those snippets by
    executing it while keeping track of performance counters.

Potentially useful but not essential. This reduces the size of llvm-bin
by ~30MB and llvm-dbg by ~1100MB.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agoclang: disable all test targets
Ross Burton [Wed, 24 Sep 2025 12:08:29 +0000 (13:08 +0100)] 
clang: disable all test targets

We don't install the tests, or run them in the build tree, so tell cmake
not to generate the targets at all. This removes internal dependencies
which means we can turn off more components that we don't want.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agoclang: improve opt-viewer PACKAGECONFIG
Ross Burton [Wed, 24 Sep 2025 12:08:28 +0000 (13:08 +0100)] 
clang: improve opt-viewer PACKAGECONFIG

Instead of patching the python module detection, we can use the hidden
option LLVM_TOOL_OPT_VIEWER_BUILD.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 weeks agoclasses/cython: also process .cpp files
Ross Burton [Wed, 24 Sep 2025 12:07:39 +0000 (13:07 +0100)] 
classes/cython: also process .cpp files

Some Python packages, such as python3-frozenlist, generate .cpp files
with cython so we should also process those.

Frustratingly this doesn't actually solve the reproducible problem with
frozenlist as the path is a temporary directory...

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>