Sorry for the delay. My take on this is that the hang is not caused by nvme. The message is a red herring. Your current up to date userspace is doing reading /sys/class/nvme/nvme0/nvme0n1/uuid from the older kernel as well. The only difference is that the older kernel doesn't show the warning as it is silenced. The newer kernel is using a different 'printf' function which is now visible. On my system I see it behave exactly the same. That means we have to look at other things. Do you happen to gather any other logs?