On 11/27/18 1:10 PM, Mimi Zohar wrote:
On Tue, 2018-11-27 at 11:48 -0800, Dan Williams wrote:
> I was thinking that the generic-length *is* the format. This does not
> work for ecyptfs because it has that:
>
> payload_datalen = sizeof(struct ecryptfs_auth_tok);
>
> ...detail that is ecryptfs specific. For nvdimm the only detail of the
> format is the decrypted-data-length. However, I get the feeling I'm
> proposing a solution to a problem that does not exist yet. Let's just
> go with the "nvdimm" format for now.
Ah, that makes more sense now. Defining "Opt_nvdimm" or the generic
"Opt_enc32" is fine. Missing from this patch is the update to
Documentation/security/keys/trusted-encrypted.rst. Otherwise this
patch looks fine.
I'll go with enc32 and update the doc and resubmit with the series.
Thanks Mimi.
Mimi