[PATCH 2/2] LSM: SafeSetID: Add setgroups() security policy handling

Micah Morton mortonm at chromium.org
Thu Jun 16 17:19:41 UTC 2022


On Tue, Jun 14, 2022 at 10:36 AM Micah Morton <mortonm at chromium.org> wrote:
>
> On Mon, Jun 13, 2022 at 4:46 PM Kees Cook <keescook at chromium.org> wrote:
> >
> > On Mon, Jun 13, 2022 at 02:00:03PM -0700, Micah Morton wrote:
> > > On Mon, Jun 13, 2022 at 1:28 PM Micah Morton <mortonm at chromium.org> wrote:
> > > [...]
> > > > +static int safesetid_task_fix_setgroups(struct cred *new, const struct cred *old)
> > > > +{
> > > > +       int i;
> > > > +
> > > > +       /* Do nothing if there are no setgid restrictions for our old RGID. */
> > > > +       if (setid_policy_lookup((kid_t){.gid = old->gid}, INVALID_ID, GID) == SIDPOL_DEFAULT)
> > > > +               return 0;
> > > > +
> > > > +       get_group_info(new->group_info);
> > > > +       for (i = 0; i < new->group_info->ngroups; i++) {
> > > > +               if (!id_permitted_for_cred(old, (kid_t){.gid = group_info->gid[i]}, GID)) {
> > >
> > > Oops, should be:
> > >
> > > !id_permitted_for_cred(old, (kid_t){.gid = new->group_info->gid[i]}, GID)
> > >
> > > Guess I won't send a whole new patch just for that one line
> >
> > This begs the question: are there self-tests for this LSM somewhere?
> > It'd be really nice to add them to tool/testing/selftests ...
>
> There actually is tools/testing/selftests/safesetid/ but I haven't
> updated it since v1 of SafeSetID that only accommodated UIDs. I've
> been relying on integration testing we have out of tree on the Chrome
> OS side but I suppose its reasonable to bring the selftest up to date
> as well :)
>
> Also both patches are a couple lines off from the ones I was finished
> developing/testing with.. some kind of screw up happened when I copied
> from my dev machine to another where I could get git-send-email
> working properly. I'll just resend these 2 patches along with the
> update to the selftest.

Just sent the updated patches and selftest patch.

>
> Thanks
>
> >
> > --
> > Kees Cook



More information about the Linux-security-module-archive mailing list