Bug 255231 - 13.0-RELEASE hangs on reboot after "Rebooting..." message on UEFI/QEMU/OpenStack
Summary: 13.0-RELEASE hangs on reboot after "Rebooting..." message on UEFI/QEMU/OpenStack
Status: Closed FIXED
Alias: None
Product: Base System
Classification: Unclassified
Component: kern (show other bugs)
Version: 13.0-RELEASE
Hardware: amd64 Any
: --- Affects Many People
Assignee: freebsd-bugs (Nobody)
Depends on:
Reported: 2021-04-19 17:08 UTC by vvv
Modified: 2022-06-17 17:19 UTC (History)
2 users (show)

See Also:

reboot screenshot (22.05 KB, image/png)
2021-04-19 17:08 UTC, vvv
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description vvv 2021-04-19 17:08:40 UTC
Created attachment 224260 [details]
reboot screenshot

Rebooting with 'reboot', 'shutdown -r now', Ctrl-Alt-Del starts shutdown sequence and hangs at the end. The issue only happens on OpenStack/QEMU environment when 13.0-RELEASE was started with UEFI firmware.
Comment 1 Andriy Gapon freebsd_committer 2021-04-20 14:01:18 UTC
Are you able to enter ddb using the magic sequence?
Or panic the system by injecting NMI?
Comment 2 vvv 2021-04-20 14:20:22 UTC
Have you any instruction how to do this using noVNC console?
Comment 3 Andriy Gapon freebsd_committer 2021-04-20 19:10:20 UTC
(In reply to vvv from comment #2)
<Enter> ~ Ctrl-b sequence should invoke ddb.
I recall that qemu has a "monitor" where some VM management commands can be issued.  I think that NMI can be injected from the monitor.
Comment 4 vvv 2021-04-20 21:40:39 UTC
I can inject NMI by the command:
virsh inject-nmi instance-id

It works when server is active. But after "Rebooting..." message injecting NMI does nothing.

It looks like incompatibility 13.0 with EFI firmware. Currently I use EFI firmware from https://www.kraxel.org/repos/jenkins/edk2/ repository. Replacing OVMF_CODE-pure-efi.fd with the same file from Fedora rawhide or Proxmox repository solves the problem.

Probably https://www.kraxel.org/repos/jenkins/edk2/ repository provides the newest files.
Comment 5 Vince Mulhollon 2022-06-17 15:42:38 UTC
Yesterday I installed a cloud image of 13.0 on a OpenStack "yoga-release" cluster and replicated the original reported symptoms, that the system halts but does not properly reboot automatically, although I could shut down the image manually and restart it in OpenStack at which point it works fine.

Then I upgraded to 13.1, repeated tests, and now it reboots normally/automatically using a "sudo shutdown -r now"

I believe you could consider this verified fixed in 13.1
Comment 6 vvv 2022-06-17 17:18:33 UTC
I confirm that the problem is fixed in 13.1.