Hello Ralf,
On Thu, Nov 05, 2020 at 07:59:18AM +0100, Ralf Horstmann wrote:
> Hi Sashan,
>
> * Alexandr Nedvedicky <[email protected]> [2020-11-05 00:25]:
> > Hello Ralf,
> >
> > I think we should rather give a try diff below. I would rather add
> > a return after m_freem(). Will diff below work for you too?
>
> Without understanding the full background I went with ignoring the return
> value
I'm sorry I have not provide better explanation in my first email.
this is the code after applying diff I've suggested earlier:
2277 void
2278 carp_transmit(struct carp_softc *sc, struct ifnet *ifp0, struct mbuf *m)
2279 {
2280 struct ifnet *ifp = &sc->sc_if;
2281
2282 #if NBPFILTER > 0
2283 {
2284 caddr_t if_bpf = ifp->if_bpf;
2285 if (if_bpf) {
2286 if (bpf_mtap_ether(if_bpf, m, BPF_DIRECTION_OUT)) {
2287 m_freem(m);
2288 return;
2289 }
2290 }
2291 }
2292 #endif /* NBPFILTER > 0 */
2293
2294 if (!ISSET(ifp0->if_flags, IFF_RUNNING)) {
2295 counters_inc(ifp->if_counters, ifc_oerrors);
2296 m_freem(m);
2297 return;
2298 }
the proposed change adds a 'return;' to line 2288, which is also consistent
with code, we can see at following line 2297.
poking to bpf_mtap(), we can see the function returns non-zero value
ordering
caller to drop packet. The bpf drop action has been added by dlg@ 4 years
ago:
Author: dlg <[email protected]>
Date: Tue Mar 29 10:38:27 2016 +0000
make bpf_mtap et al return whether the mbuf should be dropped
ok mpi@
> Anyway, I will give it a try to see whether it works.
>
thank you very much. you definitely earn credit for the hardest part:
finding a buggy line
regards
sashan
