Comment # 6 on bug 1047152 from Aleksa Sarai

I've figured out what the issue is. It's caused btrfs_qgroup_rescan_resume not
being called if you mount a filesystem as ro and then remount it as rw later.
It appears this is what systemd does during boot. I've prepared a patch which
I'm attaching here and I'll send to LKML.

I've tested it and it fixed the problem on my machine (clearing the flag
completely so when I rebooted into an old kernel the problem was resolved, and
crash_btrfs no longer caused the issue).