Hi,
When testing parallel UDP input, I see contention on the netlock
from the protocol timer. There is a check to avoid loop traversal
if igmp and mld6 are not active. Move the shortcut to avoid grabbing
netlock. This should be save as variables igmp_timers_are_running
and mld_timers_are_running are read atomically. In case we miss
an fast timer due to MP races, we just run it next time.
ok?
bluhm
Index: netinet/igmp.c
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet/igmp.c,v
retrieving revision 1.78
diff -u -p -r1.78 igmp.c
--- netinet/igmp.c 28 Mar 2022 16:31:26 -0000 1.78
+++ netinet/igmp.c 21 Aug 2022 17:01:44 -0000
@@ -526,23 +526,21 @@ igmp_fasttimo(void)
{
struct ifnet *ifp;
- NET_LOCK();
-
/*
* Quick check to see if any work needs to be done, in order
* to minimize the overhead of fasttimo processing.
*/
if (!igmp_timers_are_running)
- goto out;
+ return;
+
+ NET_LOCK();
igmp_timers_are_running = 0;
TAILQ_FOREACH(ifp, &ifnet, if_list)
igmp_checktimer(ifp);
-out:
NET_UNLOCK();
}
-
void
igmp_checktimer(struct ifnet *ifp)
Index: netinet6/mld6.c
===================================================================
RCS file: /data/mirror/openbsd/cvs/src/sys/netinet6/mld6.c,v
retrieving revision 1.56
diff -u -p -r1.56 mld6.c
--- netinet6/mld6.c 30 Nov 2018 09:28:34 -0000 1.56
+++ netinet6/mld6.c 21 Aug 2022 17:01:26 -0000
@@ -327,20 +327,19 @@ mld6_fasttimeo(void)
{
struct ifnet *ifp;
- NET_LOCK();
-
/*
* Quick check to see if any work needs to be done, in order
* to minimize the overhead of fasttimo processing.
*/
if (!mld_timers_are_running)
- goto out;
+ return;
+
+ NET_LOCK();
mld_timers_are_running = 0;
TAILQ_FOREACH(ifp, &ifnet, if_list)
mld6_checktimer(ifp);
-out:
NET_UNLOCK();
}
