Re: [PATCH v5 3/9] block: blk-crypto for Inline Encryption

["Theodore Y. Ts'o" <tytso@mit.edu>]

On Thu, Oct 31, 2019 at 10:57:13AM -0700, Christoph Hellwig wrote:
> On Mon, Oct 28, 2019 at 12:20:26AM -0700, Satya Tangirala wrote:
> > We introduce blk-crypto, which manages programming keyslots for struct
> > bios. With blk-crypto, filesystems only need to call bio_crypt_set_ctx with
> > the encryption key, algorithm and data_unit_num; they don't have to worry
> > about getting a keyslot for each encryption context, as blk-crypto handles
> > that. Blk-crypto also makes it possible for layered devices like device
> > mapper to make use of inline encryption hardware.
> > 
> > Blk-crypto delegates crypto operations to inline encryption hardware when
> > available, and also contains a software fallback to the kernel crypto API.
> > For more details, refer to Documentation/block/inline-encryption.rst.
> 
> Can you explain why we need this software fallback that basically just
> duplicates logic already in fscrypt?  As far as I can tell this fallback
> logic actually is more code than the actual inline encryption, and nasty
> code at that, e.g. the whole crypt_iter thing.

One of the reasons I really want this is so I (as an upstream
maintainer of ext4 and fscrypt) can test the new code paths using
xfstests on GCE, without needing special pre-release hardware that has
the ICE support.

Yeah, I could probably get one of those dev boards internally at
Google, but they're a pain in the tuckus to use, and I'd much rather
be able to have my normal test infrastructure using gce-xfstests and
kvm-xfstests be able to test inline-crypto.  So in terms of CI
testing, having the blk-crypto is really going to be helpful.

						- Ted