On Tue, Feb 05, 2019 at 07:24:39AM -0500, Mimi Zohar wrote:

On Mon, 2019-02-04 at 14:30 -0800, Luis Chamberlain wrote:

quoted

On Mon, Feb 04, 2019 at 05:05:10PM -0500, Mimi Zohar wrote:

quoted

On Mon, 2019-02-04 at 12:38 -0800, Luis Chamberlain wrote:

quoted

quoted

I don't see a need for an additional LSM just for verifying kernel
module signatures.

But it is one, module signing was just spawned pre the boom of LSMs.

I do believe that treating the code as such would help with its reading
and long term maintenance.

Anyway, I had to try to convince you.

Perhaps, after IMA supports appended signatures (for kernel modules),
I could see making the existing kernel module appended signature
verification an LSM.

I don't see why wait.

For now, other than updating the comment, would you be willing to add
your Review/Ack to this patch?

But I don't particularly like the changes, I still believe trying to
LSM'ify kernel module signing would be a better start to help with
long term maintenace on this code.

Also, do we have selftests implemented to ensure we don't regress with
your changes?

  Luis