The kas utility to build bitbake based projects has been packaged for Gentoo Linux.

An improvement to the ARM_MMU_KEEP macro of the Linux kernel has been suggested. Additionally several changes to the Qt based kconfig tools have been proposed. Some of them became obsolete by an unrelated refactoring, but 3 of them have been applied. The other ones have been reworked and resubmitted. The signature of arch_spin_trylock() for the hexagon architecture was found to be different from all other architectures. The documentation of the initial values about the station table in the iwlwifi driver has been found wrong. It has also been suggested to annotate a trylock function in the i915 driver so that it's return value must always be checked.

New releases cominit 0.5.8 and crinit 0.14.12 have been released. Both releases fix some minor bugs.

An issue in the gstreamer documentation of videotestsrc has been found and a request for clarification has been filed.

When running CMake's test for the FindPython module some issues have been found when using newer Python versions with older compilers. Most of them were fixed by telling the compiler not to look for that unused code at all.

When monitoring security issues for our customers this often enough leads to the conclusion that the advisory itself is not entirely correct. One case that is often observed is incorrect version numbers. The Linux kernel is now it's own CNA and as such CVEs for Linux kernel are assigned from a team of Linux engineers. To be better safe than sorry in doubt all previous versions are listed as vulnerable in case no direct starting point can be identified. This leads to vulnerabilities being listed for older, e.g. LTS, versions that actually do not have the affected code because the driver has been introduced only at a later point. We have been tracking several of those CVEs internally for quite a while. When asked if these would be an acceptable addition to the kernel.org CVE entries itself this was welcomed. A typo in the process documentation was fixed and until the end of the year 28 commits have been accepted upstram, many of them annotating several CVEs at once.

Other security advirories that have been improved are the ones from Gstreamer, where the lists of affected versions have been made more consistent to ease consumption by both humans and automated tools.

Systemd's hostnamectl now respects the 16 color palette environment variable.

Issues with wrong Content-Type header handling in snipe-it have been identified. Some simple typos have already been fixed while some other issues remained initially but are now also resolved.

The Bitbake style checker oelint-adv has been improved again, this time to handle more checksum specifications.

When the meson build tool needs an exe_wrapper to run a binary compiled for a different platform it can now print out the name of the tool for easier debugging.