summaryrefslogtreecommitdiff
path: root/sys/net/if_aggr.c
diff options
context:
space:
mode:
authorDavid Gwynne <dlg@cvs.openbsd.org>2020-07-22 02:16:03 +0000
committerDavid Gwynne <dlg@cvs.openbsd.org>2020-07-22 02:16:03 +0000
commite7539b190c92e728ca85bde26d2ba6e8599ec22c (patch)
tree855bd097524f5c275d95140541bec54bcee6faae /sys/net/if_aggr.c
parent2a91cae0189e06d8a9ef9b033ff30a43f9c53bb4 (diff)
deprecate interface input handler lists, just use one input function.
the interface input handler lists were originally set up to help us during the intial mpsafe network stack work. at the time not all the virtual ethernet interfaces (vlan, svlan, bridge, trunk, etc) were mpsafe, so we wanted a way to avoid them by default, and only take the kernel lock hit when they were specifically enabled on the interface. since then, they have been fixed up to be mpsafe. i could leave the list in place, but it has some semantic problems. because virtual interfaces filter packets based on the order they were attached to the parent interface, you can get packets taken away in surprising ways, especially when you reboot and netstart does something different to what you did by hand. by hardcoding the order that things like vlan and bridge get to look at packets, we can document the behaviour and get consistency. it also means we can get rid of a use of SRPs which were difficult to replace with SMRs. the interface input handler list is an SRPL, which we would like to deprecate. it turns out that you can sleep during stack processing, which you're not supposed to do with SRPs or SMRs, but SRPs are a lot more forgiving and it worked. lastly, it turns out that this code is faster than the input list handling, so lots of winning all around. special thanks to hrvoje popovski and aaron bieber for testing. this has been in snaps as part of a larger diff for over a week.
Diffstat (limited to 'sys/net/if_aggr.c')
-rw-r--r--sys/net/if_aggr.c39
1 files changed, 27 insertions, 12 deletions
diff --git a/sys/net/if_aggr.c b/sys/net/if_aggr.c
index 1707230cd96..aaceaf6f737 100644
--- a/sys/net/if_aggr.c
+++ b/sys/net/if_aggr.c
@@ -1,4 +1,4 @@
-/* $OpenBSD: if_aggr.c,v 1.32 2020/07/10 13:26:41 patrick Exp $ */
+/* $OpenBSD: if_aggr.c,v 1.33 2020/07/22 02:16:01 dlg Exp $ */
/*
* Copyright (c) 2019 The University of Queensland
@@ -332,6 +332,7 @@ struct aggr_port {
uint32_t p_mtu;
int (*p_ioctl)(struct ifnet *, u_long, caddr_t);
+ void (*p_input)(struct ifnet *, struct mbuf *);
int (*p_output)(struct ifnet *, struct mbuf *, struct sockaddr *,
struct rtentry *);
@@ -723,13 +724,14 @@ aggr_eh_is_slow(const struct ether_header *eh)
eh->ether_type == htons(ETHERTYPE_SLOW));
}
-static int
-aggr_input(struct ifnet *ifp0, struct mbuf *m, void *cookie)
+static void
+aggr_input(struct ifnet *ifp0, struct mbuf *m)
{
- struct ether_header *eh;
- struct aggr_port *p = cookie;
+ struct arpcom *ac0 = (struct arpcom *)ifp0;
+ struct aggr_port *p = ac0->ac_trunkport;
struct aggr_softc *sc = p->p_aggr;
struct ifnet *ifp = &sc->sc_if;
+ struct ether_header *eh;
int hlen = sizeof(*eh);
if (!ISSET(ifp->if_flags, IFF_RUNNING))
@@ -745,7 +747,7 @@ aggr_input(struct ifnet *ifp0, struct mbuf *m, void *cookie)
m = m_pullup(m, hlen);
if (m == NULL) {
/* short++ */
- return (1);
+ return;
}
eh = mtod(m, struct ether_header *);
}
@@ -756,7 +758,7 @@ aggr_input(struct ifnet *ifp0, struct mbuf *m, void *cookie)
case SLOWPROTOCOLS_SUBTYPE_LACP_MARKER:
if (mq_enqueue(&p->p_rxm_mq, m) == 0)
task_add(systq, &p->p_rxm_task);
- return (1);
+ return;
default:
break;
}
@@ -770,11 +772,10 @@ aggr_input(struct ifnet *ifp0, struct mbuf *m, void *cookie)
if_vinput(ifp, m);
- return (1);
+ return;
drop:
m_freem(m);
- return (1);
}
static int
@@ -1072,6 +1073,10 @@ aggr_add_port(struct aggr_softc *sc, const struct trunk_reqport *rp)
if (ifp0->if_type != IFT_ETHER)
return (EPROTONOSUPPORT);
+ error = ether_brport_isset(ifp0);
+ if (error != 0)
+ return (error);
+
if (ifp0->if_hardmtu < ifp->if_mtu)
return (ENOBUFS);
@@ -1103,6 +1108,7 @@ aggr_add_port(struct aggr_softc *sc, const struct trunk_reqport *rp)
CTASSERT(sizeof(p->p_lladdr) == sizeof(ac0->ac_enaddr));
memcpy(p->p_lladdr, ac0->ac_enaddr, sizeof(p->p_lladdr));
p->p_ioctl = ifp0->if_ioctl;
+ p->p_input = ifp0->if_input;
p->p_output = ifp0->if_output;
error = aggr_group(sc, p, SIOCADDMULTI);
@@ -1161,12 +1167,18 @@ aggr_add_port(struct aggr_softc *sc, const struct trunk_reqport *rp)
aggr_update_capabilities(sc);
+ /*
+ * use (and modification) of ifp->if_input and ac->ac_trunkport
+ * is protected by NET_LOCK.
+ */
+
ac0->ac_trunkport = p;
+
/* make sure p is visible before handlers can run */
membar_producer();
ifp0->if_ioctl = aggr_p_ioctl;
+ ifp0->if_input = aggr_input;
ifp0->if_output = aggr_p_output;
- if_ih_insert(ifp0, aggr_input, p);
aggr_mux(sc, p, LACP_MUX_E_BEGIN);
aggr_rxm(sc, p, LACP_RXM_E_BEGIN);
@@ -1385,10 +1397,13 @@ aggr_p_dtor(struct aggr_softc *sc, struct aggr_port *p, const char *op)
timeout_del(&p->p_current_while_timer);
timeout_del(&p->p_wait_while_timer);
- if_ih_remove(ifp0, aggr_input, p);
+ /*
+ * use (and modification) of ifp->if_input and ac->ac_trunkport
+ * is protected by NET_LOCK.
+ */
ac0->ac_trunkport = NULL;
-
+ ifp0->if_input = p->p_input;
ifp0->if_ioctl = p->p_ioctl;
ifp0->if_output = p->p_output;