[dm-devel] [RFC PATCH 0/1] Add inline encryption support for dm-crypt

[Eric Biggers]

Hi Israel,

On Thu, Jan 13, 2022 at 08:09:00PM +0200, Israel Rukshin wrote:
> Hi,
> 
> I am working to add support for inline encryption/decryption
> at storage protocols like nvmf over RDMA. The HW that I am
> working with is ConnecX-6 Dx, which supports inline crypto
> and can send the data on the fabric at the same time.
> 
> This patchset is based on v5.16-rc4 with Eric Biggers patches
> of the HW wrapped keys.
> It can be retrieved from branch "wip-wrapped-keys" at
> https://git.kernel.org/pub/scm/fs/fscrypt/fscrypt.git
> 
> I tested this patch with modified nvme-rdma as the block device
> and created a DM crypt on top of it. I got performance enhancement
> compared to SW crypto. I tested the HW wrapped inline mode with
> the HW and the standard mode with a fallback algorithm.
> 
> Israel Rukshin (1):
>   dm crypt: Add inline encryption support
> 
>  block/blk-crypto.c    |   3 +
>  drivers/md/dm-crypt.c | 202 ++++++++++++++++++++++++++++++++++++------
>  2 files changed, 180 insertions(+), 25 deletions(-)

I appreciate the interest in my patchset that adds support for hardware-wrapped
inline encryption keys
(https://lore.kernel.org/linux-block/20211116033240.39001-1-ebiggers@kernel.org).
So far I've received very little feedback on it.

One of the challenges I've been having is that I still have no platform on which
I can actually test hardware-wrapped keys with the upstream kernel.  The feature
cannot be accepted upstream until there is a way to test it.  It's almost
working on the SM8350 SoC, but I am waiting for Qualcomm to fix some things.

It sounds like you've implemented a block device driver that exposes support for
hardware-wrapped keys.  Can you please post that driver?

Can you also elaborate on how wrapped keys work in your case?  In particular,
are they compatible with the whole design which I've documented in detail in my
patchset?  That would include implementing the ->import_key, ->generate_key,
->prepare_key, and ->derive_sw_secret operations.  All the different parts are
important.  If something needs to be optional, that's something I can consider
incorporating into the design, but it would restrict the use cases.

Also, will your driver only support wrapped keys, or will it support standard
keys too?

- Eric