Created attachment 516244 [details] fstab file Hey! I upgraded my working laptop today with last systemd-236-r5 and I noticed a new flag was needed (CONFIG_CGROUP_BPF=y), then after all reboot, the laptop goes to emergency shell, mount -a is working but there is not way to systemctl default. Maybe I am unlucky or I miss something but to be sure I downgraded to 233 and it works fine, so I masked the problematic (at least for me) version and come here to file a bug. Some information about my setup: profile: [20] default/linux/amd64/17.0/desktop/plasma/systemd (stable) * root and home are btrfs subvol The btrfs system is in a dm-crypt setup (but it still ask me the key fine). I did have a lot of information, but as the system is boot, I can access the journal, here the mounting problem: Jan 23 20:22:16 josekit470 systemd[1]: Mounted /boot. Jan 23 20:22:16 josekit470 systemd[1]: home.mount: Bound to unit dev-mapper-crypt.device, but unit isn't active. Jan 23 20:22:16 josekit470 systemd[1]: Dependency failed for /home. Jan 23 20:22:16 josekit470 systemd[1]: Dependency failed for Local File Systems. Jan 23 20:22:16 josekit470 systemd[1]: local-fs.target: Job local-fs.target/start failed with result 'dependency'. Jan 23 20:22:16 josekit470 systemd[1]: local-fs.target: Triggering OnFailure= dependencies. Jan 23 20:22:16 josekit470 systemd[1]: home.mount: Job home.mount/start failed with result 'dependency'. Jan 23 20:22:16 josekit470 systemd[1]: Starting Setup Virtual Console... The thing is there is no real error, so if there is someone that have a clue about why I got this error, that would be really helpfull !
Please attach the full journalctl output for the boot sequence that failed.
I also experienced a cryptsetup boot failure upon updating from systemd-233 (initramfs dracut-045) -> 236 (initramfs dracut-046) today. I was able to see something like "... luksx2d25e41d19x2d1580x2d4e7cx2d8875x2d134045008f33.service not found". At first I thought it was https://github.com/systemd/systemd/issues/6381 After some tryouts with things like `rd.luks.timeout=30` or whatnot, it didn't help, so it seemed like this older systemd-234 issue may not be related. Then I investigated dracut, and found this commit in current `master` branch https://github.com/dracutdevs/dracut/commit/0f6d93eb9d63695a64002ec8b0421fbc9fc8a7a3#diff-4b672151d9b37747d7241b418136ac98 ``` crypt: escape backslashes for systemd unit names b/c udev/initqueue/bash otherwise luks\x2d25e41d19\x2d1580\x2d4e7c\x2d8875\x2d134045008f33 turns to luksx2d25e41d19x2d1580x2d4e7cx2d8875x2d134045008f33 ``` This looked bad and screen heuristics seemed to support this being related. First, I tried to monkey-patch my dracut-046 system installation manually w/ adding `luksname="$(str_replace "$luksname" '\' '\\')"` as shown in the patch (one of the hunks is inapplicable, though). This didn't work. Then I ran a fresh `master` dracut copy from git checkout, and now I got a working initramfs that showed my crypted root device password prompt. Main thing I saw changed is that the password prompt now showed a nice human-readable device name, instead of the UUID like before.
I added a complete log
Created attachment 516332 [details] boot log
@lkraav: I don't known if it is the same issue because I can unlock the dm-crypt disk, my problem is later whem mounting the btrfs subvol
Created attachment 516338 [details] boot log
I try to switch back to systemd-236 this afternoon and the issue is still here, I added the new log (2018-01-24 13:17 UTC) but I think there is something wrong in this version.
(In reply to Leho Kraav (:macmaN @lkraav) from comment #2) I would be happy to backport that dracut patch, but please file a new bug for it.
I don't understand why genkernel is in the subject. Can somebody explain?
As the code of genkernel wasn't changed and the breaking factor was systemd version, I guess they changed something maybe the patch of dracut help I didn't test it. Actually I didn't experience the issue anymore, so unless this is a active problem for someone, I suggest to close this.