[PATCH v2 0/3] Allow individual features to be locked down
Nicolas Bouchinet
nicolas.bouchinet at oss.cyber.gouv.fr
Thu Aug 14 10:51:20 UTC 2025
On Thu, Aug 14, 2025 at 01:02:36PM +0300, Nikolay Borisov wrote:
>
>
> On 8/14/25 11:59, Nicolas Bouchinet wrote:
> > Hi Nikolay,
> >
> > After discussing with Xiu, we have decided not to accept this patchset.
> >
> > The goal of Lockdown being to draw a clear line between ring-0 and uid-0,
> > having a more granular way to activate Lockdown will break it. Primarily
> > because most lockdown-reasons can be bypassed if used independently.
> >
> > Even if the goal of Lockdown were to be redefined, we would need to ensure the
> > security interdependence between different lockdown-reasons. This is highly
> > tied to where people calls the `security_locked_down` hook and thus is out of
> > our maintenance scope.
> >
> > Having coarse-grained lockdown reasons and integrity/confidentiality levels
> > allows us to ensure that all of the reasons are correctly locked down.
> >
> > Best regards,
> >
> > Nicolas
>
> Thanks for the feedback, to try and not have all this code go to waste, what
> about consdering patch 2 - kunits tests. Apart from
> lockdown_test_individual_level() the other tests are applicable to the
> existing lockdown implementation and can aid in future developments?
>
Yes of course, thanks a lot for those tests. Can you adapt them and send
them separately for review ?
i.e, the `lockdown_test_no_downgrade` should check for -EPERM.
FYI, I have a three week holiday starting today. I'll return on
Septembre the 8th. I let Xiu review those patches until then.
More information about the Linux-security-module-archive
mailing list