Hi Eric,

[other lists removed intentionally]

This could be an upstream problem.

I tried to reproduce it on a test system with the same kernel with new 
created superblocks, there everything worked fine. Also after setting 
dev.cache.discard to yes.

Then I copied the complete partitions from the productive system and got 
the same bug there.

Long story short: It seems to be caused by some data in the the 
cache-device superblock. So even if it's an upstream problem, it's 
somehow caused by data in my superblock.

It can be found here: https://www.f0o.de/~sroesner/bcache-superblock.bz2

Maybe you can reproduce it, for testing I created a degraded raid on 
only one 500G partition and tried to register it to cause the bug.

Sebastian