[PATCH v4] audit: log AUDIT_TIME_* records only from rules

[Paul Moore]

On Mon, Feb 14, 2022 at 10:23 AM Richard Guy Briggs <rgb at redhat.com> wrote:
>
> AUDIT_TIME_* events are generated when there are syscall rules present that are
> not related to time keeping.  This will produce noisy log entries that could
> flood the logs and hide events we really care about.
>
> Rather than immediately produce the AUDIT_TIME_* records, store the data in the
> context and log it at syscall exit time respecting the filter rules.
>
> Please see https://bugzilla.redhat.com/show_bug.cgi?id=1991919
>
> Fixes: 7e8eda734d30 ("ntp: Audit NTP parameters adjustment")
> Fixes: 2d87a0674bd6 ("timekeeping: Audit clock adjustments")
> Signed-off-by: Richard Guy Briggs <rgb at redhat.com>
> ---
> Changelog:
> v2:
> - rename __audit_ntp_log_ to audit_log_ntp
> - pre-check ntp before storing
> - move tk out of the context union and move ntp logging to the bottom of audit_show_special()
> - restructure logging of ntp to use ab and allocate more only if more
> - add Fixes lines
> v3
> - move tk into union
> - rename audit_log_ntp() to audit_log_time() and add tk
> - key off both AUDIT_TIME_* but favour NTP
> v4
> - drop tk goto in favour of ntp if clause
> - add comments to clarify calling function buffer expectations
>
>  kernel/audit.h   |  4 +++
>  kernel/auditsc.c | 86 +++++++++++++++++++++++++++++++++++++-----------
>  2 files changed, 70 insertions(+), 20 deletions(-)

...

> diff --git a/kernel/auditsc.c b/kernel/auditsc.c
> index fce5d43a933f..86f1c65ac933 100644
> --- a/kernel/auditsc.c
> +++ b/kernel/auditsc.c
> @@ -1340,6 +1340,53 @@ static void audit_log_fcaps(struct audit_buffer *ab, struct audit_names *name)
>                          from_kuid(&init_user_ns, name->fcap.rootid));
>  }
>
> +void audit_log_time(struct audit_context *context, struct audit_buffer **ab)
> +{
> +       const struct audit_ntp_data *ntp = &context->time.ntp_data;
> +       const struct timespec64 *tk = &context->time.tk_injoffset;
> +       const char *ntp_name[] = {
> +               "offset",
> +               "freq",
> +               "status",
> +               "tai",
> +               "tick",
> +               "adjust",
> +       };
> +       /* use up allocated ab from show_special before new one */
> +       int type, first = 1;
> +
> +       if (context->type == AUDIT_TIME_ADJNTPVAL) {
> +               for (type = 0; type < AUDIT_NTP_NVALS; type++) {
> +                       if (ntp->vals[type].newval != ntp->vals[type].oldval) {
> +                               if (first) {
> +                                       first = 0;
> +                               } else {
> +                                       audit_log_end(*ab);
> +                                       *ab = audit_log_start(context, GFP_KERNEL,
> +                                                             AUDIT_TIME_ADJNTPVAL);
> +                                       if (!*ab)
> +                                               return;
> +                               }
> +                               audit_log_format(*ab, "op=%s old=%lli new=%lli",
> +                                                ntp_name[type], ntp->vals[type].oldval,
> +                                                ntp->vals[type].newval);
> +                       }
> +               }
> +       }
> +       if (tk->tv_sec != 0 || tk->tv_nsec != 0) {
> +               if (!first) {
> +                       audit_log_end(*ab);
> +                       *ab = audit_log_start(context, GFP_KERNEL,
> +                                             AUDIT_TIME_INJOFFSET);
> +                       if (!*ab)
> +                               return;
> +               }
> +               audit_log_format(*ab, "sec=%lli nsec=%li",
> +                                (long long)tk->tv_sec, tk->tv_nsec);
> +       }
> +       /* allocated ab will be ended by show_special */
> +}

As discussed earlier, we can simplify this function above (e.g. get
rid of "first" and all the related work) by allowing audit_log_time()
to return a NULL audit_buffer.  I'd rather see that than the
audit_log_time() implementation above.

The audit_log_end() and most (all?) of the audit_log_*() functions can
safely handle a NULL audit_buffer, if they couldn't handle that
condition the kernel would die a terrible death in other areas.  We
might as well use that to our advantage here and clean up
audit_log_time().

-- 
paul-moore.com