Comment # 21 on bug 1047152 from Aleksa Sarai

(In reply to rob loranger from comment #20)
> This appears to be fixed in tumbleweed as of 4.12.9-1

The patch is still not present in 4.12.9. The reason why you cannot reproduce
the issue again is because of the nature of the bug. Effectively the issue is
that if there is a btrfs qgroup rescan running and the machine hard resets, the
flag indicating that "there is a rescan running" remains present on disk. The
flag is not cleared correctly (due to how systemd mounts the root filesystem),
resulting in a bunch of operations stalling indefinitely. My patch corrects the
codepath that systemd hits when mounting the root filesystem, causing the flag
to be cleared correctly. As a result, rebooting into a kernel without my patch
will still work (until you have the flag set and you hard reset the machine
again).

I spoke to one of the maintainers of the btrfs tree and it looks like it'll
land in 4.14, marked for stable -- so it will trickle into stable kernels as
soon as it's merged into mainline in 4.14-rc1 (in a few weeks).