Lennart Poettering
Lennart Poettering
Hmm, what. libvirt puts the top-level or first-level delegated cgroup into threaded mode? it really shouldn't do that. I can do that for children further down the tree (two or...
> `cat /sys/fs/cgroup/machine.slice/machine-qemu\\x2d1\\x2dWin11\\x2dWin11.scope/cgroup.procs` -> `null` hmm, are you saying this literally returned the string "null" for you? that's not expected contents for this file. Is this a regular kernel, or...
@1wxyz any chance you can clarify this? Did you really see the string "null" in those cgroup.procs files?
> [ 14.156360] systemd[1]: Sending SIGKILL to PID 302 (6). > [ 14.156661] systemd[1]: Sending SIGKILL to PID 346 (5). > [ 14.156890] systemd[1]: Sending SIGKILL to PID 382 (sleep)....
there might be processes left over from generators that pid 1 forks off, which i think we should clean up. i think a more finegrained approach would be best here,...
As per the comments on https://github.com/systemd/systemd/pull/32235 I think we probably shouldn't exclude these from killing after all. But maybe downgrade the log messages about them.
So I take it that the problem this is addresses is if creds are used together with Type=simple and at least one ExecStartPost= line? I have the suspicion that ExecStartPost=...
My guess is that your TPM doesn't support "CFB" mode, but supports AES just fine (because it cannot be thaaaaat old)
Hmm, I don't get it? Why would you put /boot/ on the rootfs? I mean, the root fs is typically encrypted, verity-fied and what not. It likely uses a complex...