Created attachment 661992 [details]
Output of "journalctl --boot" after the failure occured during boot

Loading the kernel-module 'snd_aloop' via modules-load.d somehow propagates
into the initrd as soon as that is recreated.  However, loading this module
fails for some reason within the initrd.  It succeeds lateron, after the boot
process left the initrd.

To reproduce:
$ echo 'snd_aloop' > /etc/modules-load.d/snd_aloop.conf
$ mkinitrd
$ reboot
# then watch out for "[FAILED] Failed to start Load Kernel Modules."

dmesg shows lots of messages like
> snd_aloop: Unknown symbol snd_ctl_add (err 0)
> snd_aloop: Unknown symbol snd_pcm_lib_free_vmalloc_buffer (err 0)
> snd_aloop: Unknown symbol snd_pcm_new (err 0)
(see attached output of "journalctl --boot").

It's not clear to my why it (dracut?) tries to load the module inside the
initrd -- loading it after leaving the initrd would be sufficient.  I tried to
suppress it with the kernel command-line argument
"rd.driver.blacklist=snd_aloop", but to no effect.

The boot process itself is not interrupted by this failure, so this is merely a
cosmetic problem.  Yet, a service failure during early boot, prominently
visible while the luks passphrase is requested, is quite irritating for the
user.

BTW, loading the "lp" module with the same mechanism works fine.