> We can simply add another bio flag to get back to the previous
> behavior. That being said thing to do in the end is to verify it
> at the top of the stack, and not the bottom eventuall. I can cook
> up a patch for that.
Yeah, the original code was careful about only adding the verification
hook to the top bio.
A bio flag is probably the path of least resistance. It already exists,
actually: BIP_BLOCK_INTEGRITY. But we'll need to make sure it gets
masked out when a bio is cloned. And then we can key off of that and
REQ_OP_READ in the endio function.
I prefer that approach to reverting Christoph's commit.
Martin K. Petersen Oracle Linux Engineering