On Thu, 5 Sep 2019, Stephen Smalley wrote: > For the next and any future rounds of stacking support, I'm hoping we can be a > bit more rigorous in our code review and testing requirements. After the 5th iteration of the patchset and with everything having at least one trusted reviewer, I did ask the list if there were any objections and stated I would otherwise merge to v4.21: http://kernsec.org/pipermail/linux-security-module-archive/2018-December/010209.html Regardless, from now I'll also require signoffs from all of the major LSMs on these kinds of changes. I'd have to say I also have concerns about the direction of LSM stacking and what its final goal is. It seems to makes sense to be able to stack Apparmor inside SELinux or Smack, so that folk can use AA system containers on an SELinux or Smack host. We've not seen other proposed use-cases for full stacking materialize in a sustained and viable manner. Are there really any beyond AA inside a labeling LSM? Because if not I think this is a reasonable goal state for LSM stacking until we concretely know otherwise. -- James Morris <jmorris@xxxxxxxxx>
