Re: [PATCH v6 1/4] btrfs: test btrfs specific fsverity corruption

[Eric Biggers <ebiggers@kernel.org>]

On Mon, Feb 14, 2022 at 04:09:55PM -0800, Boris Burkov wrote:
> diff --git a/common/verity b/common/verity
> index 38eea157..eec8ae72 100644
> --- a/common/verity
> +++ b/common/verity
> @@ -3,6 +3,8 @@
>  #
>  # Functions for setting up and testing fs-verity
>  
> +. common/btrfs

Why does common/btrfs need to be included here?

> +# Check for userspace tools needed to corrupt verity data or metadata.
> +_require_fsverity_corruption()
> +{
> +	_require_xfs_io_command "fiemap"
> +	if [ $FSTYP == "btrfs" ]; then
> +		_require_btrfs_corrupt_block
> +	fi
> +}

Which function(s), specifically, is this function a prerequisite for?  Based on
the name, it sounds like it would be a prerequisite for calling
_fsv_scratch_corrupt_bytes() and _fsv_scratch_corrupt_merkle_tree().  But that's
apparently not the case, seeing as generic/576 calls
_fsv_scratch_corrupt_bytes() but doesn't call _require_fsverity_corruption(),
and your new test btrfs/290 calls _require_fsverity_corruption() but doesn't
call either _fsv_scratch_corrupt_bytes() or _fsv_scratch_corrupt_merkle_tree().

So, the purpose of this function is unclear.

Perhaps it should be a prerequisite for all _fsv_scratch_corrupt*() functions,
and btrfs/290 should check for btrfs-corrupt-block directly?

- Eric