1
0
mirror of https://git.FreeBSD.org/src.git synced 2026-06-02 11:24:32 +00:00

lagg: Handle a port count of zero

The sc_count check in lagg_transmit_ethernet() and
lagg_transmit_infiniband() is racy, as the lagg protocol handlers are
only synchronized by net_epoch.  Handle a count of 0 in each protocol
handler where it's needed, namely in the RR and LB handlers.

Reported by:	Yuxiang Yang, Yizhou Zhao, Xuewei Feng, Qi Li, and Ke Xu from Tsinghua University using GLM5.1 from Z.ai
Reviewed by:	pouria, zlei
MFC after:	2 weeks
Differential Revision:	https://reviews.freebsd.org/D56942
This commit is contained in:
Mark Johnston
2026-06-01 16:40:43 +00:00
parent 86326398b7
commit 49d90d9ddf
+23 -17
View File
@@ -1828,6 +1828,7 @@ lookup_snd_tag_port(struct ifnet *ifp, uint32_t flowid, uint32_t flowtype,
struct lagg_port *lp;
struct lagg_lb *lb;
uint32_t hash, p;
unsigned int count;
int err;
sc = ifp->if_softc;
@@ -1839,8 +1840,11 @@ lookup_snd_tag_port(struct ifnet *ifp, uint32_t flowid, uint32_t flowtype,
if ((sc->sc_opts & LAGG_OPT_USE_FLOWID) == 0 ||
flowtype == M_HASHTYPE_NONE)
return (NULL);
count = atomic_load_int(&sc->sc_count);
if (count == 0)
return (NULL);
p = flowid >> sc->flowid_shift;
p %= sc->sc_count;
p %= count;
lb = (struct lagg_lb *)sc->sc_psc;
lp = lb->lb_ports[p];
return (lagg_link_active(sc, lp));
@@ -2128,12 +2132,6 @@ lagg_transmit_ethernet(struct ifnet *ifp, struct mbuf *m)
if (m->m_pkthdr.csum_flags & CSUM_SND_TAG)
MPASS(m->m_pkthdr.snd_tag->ifp == ifp);
#endif
/* We need at least one port */
if (sc->sc_count == 0) {
m_freem(m);
if_inc_counter(ifp, IFCOUNTER_OERRORS, 1);
return (ENXIO);
}
ETHER_BPF_MTAP(ifp, m);
@@ -2150,12 +2148,6 @@ lagg_transmit_infiniband(struct ifnet *ifp, struct mbuf *m)
if (m->m_pkthdr.csum_flags & CSUM_SND_TAG)
MPASS(m->m_pkthdr.snd_tag->ifp == ifp);
#endif
/* We need at least one port */
if (sc->sc_count == 0) {
m_freem(m);
if_inc_counter(ifp, IFCOUNTER_OERRORS, 1);
return (ENXIO);
}
infiniband_bpf_mtap(ifp, m);
@@ -2421,13 +2413,20 @@ lagg_rr_start(struct lagg_softc *sc, struct mbuf *m)
{
struct lagg_port *lp;
uint32_t p;
unsigned int count;
count = atomic_load_int(&sc->sc_count);
if (__predict_false(count == 0)) {
if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
m_freem(m);
return (ENETDOWN);
}
p = atomic_fetchadd_32(&sc->sc_seq, 1);
p /= sc->sc_stride;
p %= sc->sc_count;
lp = CK_SLIST_FIRST(&sc->sc_ports);
p %= count;
while (p--)
lp = CK_SLIST_FIRST(&sc->sc_ports);
while (p-- && lp != NULL)
lp = CK_SLIST_NEXT(lp, lp_entries);
/*
@@ -2611,13 +2610,20 @@ lagg_lb_start(struct lagg_softc *sc, struct mbuf *m)
struct lagg_lb *lb = (struct lagg_lb *)sc->sc_psc;
struct lagg_port *lp = NULL;
uint32_t p = 0;
unsigned int count;
count = atomic_load_int(&sc->sc_count);
if (__predict_false(count == 0)) {
if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
m_freem(m);
return (ENETDOWN);
}
if ((sc->sc_opts & LAGG_OPT_USE_FLOWID) &&
M_HASHTYPE_GET(m) != M_HASHTYPE_NONE)
p = m->m_pkthdr.flowid >> sc->flowid_shift;
else
p = m_ether_tcpip_hash(sc->sc_flags, m, lb->lb_key);
p %= sc->sc_count;
p %= count;
lp = lb->lb_ports[p];
/*