Re: [RFC PATCH ghak10 v4 0/2] audit: Log modifying adjtimex(2) calls

[Ondrej Mosnacek <omosnace@redhat.com>]

On Wed, Aug 22, 2018 at 11:27 PM Paul Moore <paul@paul-moore.com> wrote:
> On Tue, Aug 21, 2018 at 3:21 AM Miroslav Lichvar <mlichvar@redhat.com> wrote:
> > > On Mon, 20 Aug 2018, Ondrej Mosnacek wrote:
> > > > @John or other timekeeping/NTP folks: We had a discussion on the audit
> > > > ML on which of the internal timekeeping/NTP variables we should actually
> > > > log changes for. We are only interested in variables that can (directly
> > > > or indirectly) cause noticeable changes to the system clock, but since we
> > > > have only limited understanding of the NTP code, we would like to ask
> > > > you for advice on which variables are security relevant.
> >
> > I guess that mostly depends on whether you consider setting the clock
> > to run faster or slower than real time to be an important event for
> > the audit.
> >
> > > >   - NTP value adjustments:
> > > >     - time_offset (probably important)
> >
> > This can adjust the clock by up to 0.5 seconds per call and also speed
> > it up or slow down by up to about 0.05% (43 seconds per day).
>
> This seems worthwhile.
>
> > > >     - time_freq (maybe not important?)
> >
> > This can speed up or slow down by up to about 0.05%.
>
> This too.
>
> > > >     - time_status (likely important, can cause leap second injection)
> >
> > Yes, it can insert/delete leap seconds and it also enables/disables
> > synchronization of the hardware real-time clock.
>
> This one as well.
>
> > > >     - time_maxerror (maybe not important?)
> > > >     - time_esterror (maybe not important?)
> >
> > These two change the error estimates that are reported to applications
> > using ntp_gettime()/adjtimex(). If an application was periodically
> > checking that the clock is synchronized with some specified accuracy
> > and setting the maxerror to a larger value would cause the application
> > to abort, would it be an important event in the audit?
>
> Since these don't really affect the time, just the expected error, I'm
> not sure this is important.
>
> > > >     - time_constant (???)
> >
> > This controls the speed of the clock adjustments that are made when
> > time_offset is set. Probably not important for the audit.
>
> Agreed.  I think we can skip this.
>
> > > >     - time_adjust (sounds important)
> >
> > This is similar to time_freq. It can temporarily speed up or slow down
> > the clock by up to 0.05%.
>
> Like time_freq, we should probably log this too.
>
> > > >     - tick_usec (???)
> >
> > This is a more extreme version of time_freq. It can speed up or slow
> > down the clock by up to 10%.
>
> Let's audit this one too.

I agree with Paul on all counts. I will go ahead and prepare a
patchset that logs everything except maxerror, esterror, and constant.

Thank you, Miroslav, for the explanations!

--
Ondrej Mosnacek <omosnace at redhat dot com>
Associate Software Engineer, Security Technologies
Red Hat, Inc.