This pulls in new dependencies (flatpak, ostree-libs) which it did have before (see below). Also, it cannot be erased due to the dependency. Why are flatpak and ostree forced upon us?
dnf erase xdg-desktop-portal Error: Problem: The operation would result in removing the following protected packages: gnome-shell
Installing dependencies: flatpak-selinux noarch 1.10.1-1.fc33 updates 23 k flatpak-session-helper x86_64 1.10.1-1.fc33 updates 79 k ostree-libs x86_64 2020.8-1.fc33 updates 427 k Installing weak dependencies: flatpak x86_64 1.10.1-1.fc33 updates 1.7 M p11-kit-server x86_64 0.23.22-2.fc33 updates 199 k xdg-desktop-portal-wlr x86_64 0.1.0-2.fc33 fedora 34 k
Unfortunately, this update has been pushed to stable for F29 before even submitting an update for F30 at all. It breaks system-upgrade from F29 to F30 completely!
Since the previous version is not in updates any more users are forced to downgrade all the way down to the F29 base version, which is no good version for the dist-upgrade.
While the kernel installs and boots there is a reproducible segfault for dracut-install (https://bugzilla.redhat.com/show_bug.cgi?id=1685932) with the current dracut. It s not clear yet whether this a dracut issue or a kernel 5.0 issue. So +1 one for the working update, -1 for the segfault during dnf update.
Fedora 4.19 kernels before 4.19.5 and after show an acpi crash for me on an old HP/compaq 6715b laptop with an AMD Turion. Same crash with 4.19.8, whereas 4.19.5 did work. Low and behold: the koji build https://koji.fedoraproject.org/koji/taskinfo?taskID=31411873 ("atomacpi") works, the kernel boots as usual. acpitool -c says:
CPU type : AMD Turion(tm) 64 X2 Mobile Technology TL-60 Min/Max frequency : 800/2000 MHz Current frequency : 800 MHz Frequency governor : ondemand Freq. scaling driver : powernow-k8 Cache size : 512 KB Bogomips : 3989.80 Bogomips : 3989.80 Function Show_CPU_Info : could not read directory /proc/acpi/processor/ Make sure your kernel has ACPI processor support enabled.
This package has been pushed to F28 but not F29. The dependencies are messed up but I'm not sure whether the problem is in -3 or in the earlier versions. Rebuilding my package
impressive which depends on python2-pygame (unversioned) and reinstalling helped.
Maybe it is as simple as a missing explicit
Provides: python2-pygame for the subpackage.
In any case, having this in F28 forces maintainers of depending packages to bump their F28 versions for a fix - and not pushing it to F29 is not quite what we are supposed to do.
The way the kernel folks handled retpoline (gcc build in extra koji buildroot) was the best they could do to get those safer kernels out quickly and still have reproducible builds.
koji buildroots are available for everyone to do their scratch builds with. There's also a gcc-retpoline copr if you want to do builds there, using a retpoline gcc.