@jlebon Thanks for bringing my attention to this test! I wonder what test cases do this test include? Is it CoreOS-specific or can be run using tools like tmt?

Thanks for confirming it!

I've released kexec-tools-2.0.28-12.fc41. kexec-tools-2.0.28-10.fc41 should be used because it has the same problem as 2.0.28-10.fc40.

Thanks for correcting me!

Oh, this is a regression caused by the following commit. Sorry I didn't catch this issue when merging it.

commit e3a1eb06ac40d0d3c83ac15bb78c0094f8dc2b1a (kexec_sbin/bin-sbin-merge)
Author: Zbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl>
Date:   Sun Apr 14 19:06:17 2024 +0200

    Fix build when %_bindir==%_sbindir

    Preparation for https://fedoraproject.org/wiki/Changes/Unify_bin_and_sbin.
 # makedumpfile

@@ -249,8 +251,9 @@ fi
 %{_sbindir}/mkfadumprd
 %{_prefix}/lib/kernel/install.d/60-fadump.install
 %endif
+%{_bindir}/kdumpctl
 %{_sbindir}/mkdumprd
-%{_bindir}/*
+%{_sbindir}/vmcore-dmesg

The latest kdump-utils has already fixed this issue unintentionally. Since 1) this issue can be bypassed by updating kdump-utils to latest version 2) and only happens to rawhide 3) and you plan to fix it on your side, do you think "Conflicts: kdump-utils < 1.0.42" is still needed? Or what else can I do for you?

As you see from the file list kdump-utils doesn't have /usr/sbin/vmcore-dmesg. This file is always provided by kexec-tools. I imagine when kexec-tools gets updated, the old /usr/sbin/vmcore-dmesg should be replaced by the new version. So think the real problem should lie somewhere else.

Hi @jlebon,

Thanks for reporting the above test failure! But I'm not sure how to make sense of the above error! Can you help me understand it?