On Sat, Dec 11, 2021 at 11:05:10AM -0500, James Bottomley wrote:
> On Sat, 2021-12-11 at 09:29 -0600, Serge E. Hallyn wrote:
> > On Fri, Dec 10, 2021 at 02:47:34PM -0500, Stefan Berger wrote:
> > > Use mac_admin_ns_capable() to check corresponding capability to
> > > allow
> > > read/write IMA policy without CAP_SYS_ADMIN but with CAP_MAC_ADMIN.
> > >
> > > Signed-off-by: Denis Semakin <denis.semakin@xxxxxxxxxx>
> > > Signed-off-by: Stefan Berger <stefanb@xxxxxxxxxxxxx>
> > > ---
> > > include/linux/capability.h | 6 ++++++
> > > security/integrity/ima/ima_fs.c | 2 +-
> > > 2 files changed, 7 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/include/linux/capability.h
> > > b/include/linux/capability.h
> > > index 65efb74c3585..991579178f32 100644
> > > --- a/include/linux/capability.h
> > > +++ b/include/linux/capability.h
> > > @@ -270,6 +270,12 @@ static inline bool
> > > checkpoint_restore_ns_capable(struct user_namespace *ns)
> > > ns_capable(ns, CAP_SYS_ADMIN);
> > > }
> > >
> > > +static inline bool mac_admin_ns_capable(struct user_namespace *ns)
> > > +{
> > > + return ns_capable(ns, CAP_MAC_ADMIN) ||
> > > + ns_capable(ns, CAP_SYS_ADMIN);
> > > +}
> > > +
> > > /* audit system wants to get cap info from files as well */
> > > int get_vfs_caps_from_disk(struct user_namespace *mnt_userns,
> > > const struct dentry *dentry,
> > > diff --git a/security/integrity/ima/ima_fs.c
> > > b/security/integrity/ima/ima_fs.c
> > > index a136d14f29ec..090ee85bfa3a 100644
> > > --- a/security/integrity/ima/ima_fs.c
> > > +++ b/security/integrity/ima/ima_fs.c
> > > @@ -440,7 +440,7 @@ static int ima_open_policy(struct inode *inode,
> > > struct file *filp)
> > > #else
> > > if ((filp->f_flags & O_ACCMODE) != O_RDONLY)
> > > return -EACCES;
> > > - if (!capable(CAP_SYS_ADMIN))
> > > + if (!mac_admin_ns_capable(user_ns))
> >
> > Sorry if I'm missing something. But I'm looking at your tree's
> > version of ima_update_policy() and failing to see where it adds
> > extra capability checks. Note that any unprivileged user can
> > unshare a user namespace, map its hostuid to nsuid 0, and pass
> > ns_capable(CAP_MAC_ADMIN).
> >
> > Likewise, a host uid 0 process which does not have CAP_MAC_ADMIN
> > can create a new user namespace, map hostuid 0 to nsuid 0, and
> > have CAP_MAC_ADMIN against the new userns.
> >
> > Somewhere you need to be checking for privilege against either
> > the parent ns or the init_user_ns. I'm not seeing where that's
> > being done. Can you point me to it?
>
> I think you're thinking of my patches, which used a single ima log for
> everything. With Stefan's patches, each namespace gets its own log and
> starts out with and empty "do nothing" policy. However, whatever
> policy was active in the parent namespace is still run for activities
> in the child, so any activity the parent requires to be
> measured/audited/appraised etc still happens, but all entries are made
> in the parent.
Oh. Stefan had previously said that one motivation was to do less logging
in the hostns, so I assumed that a child ns policy could make it so the
parent ns policy was no longer applied. If that's not the case, and not
planned, then that makes this a lot simpler.
