[RFC PATCH v3 3/3] devguard: added device guard for mknod in non-initial userns

[Christian Brauner]

On Fri, Dec 15, 2023 at 10:08:08AM -0800, Alexei Starovoitov wrote:
> On Fri, Dec 15, 2023 at 6:15 AM Christian Brauner <brauner at kernel.org> wrote:
> >
> > On Fri, Dec 15, 2023 at 02:26:53PM +0100, Michael Weiß wrote:
> > > On 15.12.23 13:31, Christian Brauner wrote:
> > > > On Wed, Dec 13, 2023 at 03:38:13PM +0100, Michael Weiß wrote:
> > > >> devguard is a simple LSM to allow CAP_MKNOD in non-initial user
> > > >> namespace in cooperation of an attached cgroup device program. We
> > > >> just need to implement the security_inode_mknod() hook for this.
> > > >> In the hook, we check if the current task is guarded by a device
> > > >> cgroup using the lately introduced cgroup_bpf_current_enabled()
> > > >> helper. If so, we strip out SB_I_NODEV from the super block.
> > > >>
> > > >> Access decisions to those device nodes are then guarded by existing
> > > >> device cgroups mechanism.
> > > >>
> > > >> Signed-off-by: Michael Weiß <michael.weiss at aisec.fraunhofer.de>
> > > >> ---
> > > >
> > > > I think you misunderstood me... My point was that I believe you don't
> > > > need an additional LSM at all and no additional LSM hook. But I might be
> > > > wrong. Only a POC would show.
> > >
> > > Yeah sorry, I got your point now.
> >
> > I think I might have had a misconception about how this works.
> > A bpf LSM program can't easily alter a kernel object such as struct
> > super_block I've been told.
> 
> Right. bpf cannot change arbitrary kernel objects,
> but we can add a kfunc that will change a specific bit in a specific
> data structure.
> Adding a new lsm hook that does:
>     rc = call_int_hook(sb_device_access, 0, sb);
>     switch (rc) {
>     case 0: do X
>     case 1: do Y
> 
> is the same thing, but uglier, since return code will be used
> to do this action.
> The 'do X' can be one kfunc
> and 'do Y' can be another.
> If later we find out that 'do X' is not a good idea we can remove
> that kfunc.

The reason I moved the SB_I_MANAGED_DEVICES here is that I want a single
central place where that is done for any possible LSM that wants to
implement device management. So we don't have to go chasing where that
bit is set for each LSM. I also don't want to have LSMs raise bits in
sb->s_iflags directly as that's VFS property.