> The fs-specific field solution from this patchset is much more efficient > than the rhashtable: efficient enough that we don't really have to worry > about it, regardless of fscrypt or fsverity. So I think it's a good > middle ground, and I'd like to just do it this way.
I agree.
