[PATCH] security: fix the logic in security_inode_getsecctx()

[Stephen Smalley]

On Fri, Jan 26, 2024 at 12:15 PM Ondrej Mosnacek <omosnace at redhat.com> wrote:
>
> On Fri, Jan 26, 2024 at 5:04 PM Stephen Smalley
> <stephen.smalley.work at gmail.com> wrote:
> >
> > On Fri, Jan 26, 2024 at 10:03 AM Stephen Smalley
> > <stephen.smalley.work at gmail.com> wrote:
> > >
> > > On Fri, Jan 26, 2024 at 5:44 AM Ondrej Mosnacek <omosnace at redhat.com> wrote:
> > > >
> > > > The inode_getsecctx LSM hook has previously been corrected to have
> > > > -EOPNOTSUPP instead of 0 as the default return value to fix BPF LSM
> > > > behavior. However, the call_int_hook()-generated loop in
> > > > security_inode_getsecctx() was left treating 0 as the neutral value, so
> > > > after an LSM returns 0, the loop continues to try other LSMs, and if one
> > > > of them returns a non-zero value, the function immediately returns with
> > > > said value. So in a situation where SELinux and the BPF LSMs registered
> > > > this hook, -EOPNOTSUPP would be incorrectly returned whenever SELinux
> > > > returned 0.
> > > >
> > > > Fix this by open-coding the call_int_hook() loop and making it use the
> > > > correct LSM_RET_DEFAULT() value as the neutral one, similar to what
> > > > other hooks do.
> > > >
> > > > Reported-by: Stephen Smalley <stephen.smalley.work at gmail.com>
> > > > Link: https://lore.kernel.org/selinux/CAEjxPJ4ev-pasUwGx48fDhnmjBnq_Wh90jYPwRQRAqXxmOKD4Q@mail.gmail.com/
> > > > Fixes: b36995b8609a ("lsm: fix default return value for inode_getsecctx")
> > > > Signed-off-by: Ondrej Mosnacek <omosnace at redhat.com>
> > > > ---
> > > >
> > > > I ran 'tools/nfs.sh' on the patch and even though it fixes the most
> > > > serious issue that Stephen reported, some of the tests are still
> > > > failing under NFS (but I will presume that these are pre-existing issues
> > > > not caused by the patch).
> > >
> > > Do you have a list of the failing tests? For me, it was hanging on
> > > unix_socket and thus not getting to many of the tests. I would like to
> > > triage the still-failing ones to confirm that they are in fact
> > > known/expected failures for NFS.
> >
> > Applying your patch and removing unix_socket from the tests to be run
> > (since it hangs), I get the following failures:
> > mac_admin/test            (Wstat: 0 Tests: 8 Failed: 2)
> >   Failed tests:  5-6
> > filesystem/ext4/test      (Wstat: 512 (exited 2) Tests: 76 Failed: 2)
> >   Failed tests:  1, 64
> >   Non-zero exit status: 2
> > filesystem/xfs/test       (Wstat: 512 (exited 2) Tests: 76 Failed: 2)
> >   Failed tests:  1, 64
> >   Non-zero exit status: 2
> > filesystem/jfs/test       (Wstat: 512 (exited 2) Tests: 83 Failed: 2)
> >   Failed tests:  1, 71
> >   Non-zero exit status: 2
> > filesystem/vfat/test      (Wstat: 512 (exited 2) Tests: 52 Failed: 2)
> >   Failed tests:  1, 46
> >   Non-zero exit status: 2
> > fs_filesystem/ext4/test   (Wstat: 512 (exited 2) Tests: 75 Failed: 2)
> >   Failed tests:  1, 63
> >   Non-zero exit status: 2
> > fs_filesystem/xfs/test    (Wstat: 512 (exited 2) Tests: 75 Failed: 2)
> >   Failed tests:  1, 63
> >   Non-zero exit status: 2
> > fs_filesystem/jfs/test    (Wstat: 512 (exited 2) Tests: 82 Failed: 2)
> >   Failed tests:  1, 70
> >   Non-zero exit status: 2
> > fs_filesystem/vfat/test   (Wstat: 512 (exited 2) Tests: 51 Failed: 2)
> >   Failed tests:  1, 45
> >   Non-zero exit status: 2
> > Files=77, Tests=1256, 308 wallclock secs ( 0.30 usr  0.10 sys +  6.84
> > cusr 21.78 csys = 29.02 CPU)
>
> I got the same ones (I, too, removed unix_socket to allow the rest to run).

unix_socket test is failing because type_transition rule is not being
applied to newly created server socket, leading to a denial when the
client tries to connect. I believe that once worked; will see if I can
find the last working kernel.