Why? Improve hardware support, e.g. - Intel UHD 750: remove the need for the workaround documented in 516a1762f9 - #19353+: some of these regressions should be fixed by upgrading the kernel; some others will also need a newer Mesa/libdrm stack, which we'll only get by upgrading to Bookworm in Tails 6.0 Process: https://tails.boum.org/contribute/Linux_kernel/ Specific issues to take into account: - [Kernel rfkill interface changes in 5.11 requiring userspace fixes](https://discourse.gnome.org/t/kernel-rfkill-interface-changes-in-5-11-requiring-userspace-fixes/6688) - I believe we don't need to do anything, because this only matters when building gnome-settings-daemon against recent kernel headers, while the source tree includes an old version of `rfkill.h` (without the breaking changes) and the Debian package does not build-depends on the Linux headers nor `linux-libc-dev` anyway (and even if it did, that would be Bullseye's headers, not 5.11+ ones). -- intrigeri - Test results: https://gitlab.tails.boum.org/tails/tails/-/merge_requests/996#note_201360 - [The 5.17 kernel will require some initrd generator changes for kms drivers](https://hansdegoede.livejournal.com/25948.html) - [security things in Linux v5.10](https://outflux.net/blog/archives/2022/04/04/security-things-in-linux-v5-10/) - Bundling these could save us some QA and paperwork: - #19023+ - #19345+ - #18858+