Skip to contentSkip to conversation

Patch "ipv6: prevent fib6_run_gc() contention" has been added to the 3.10-stable tree

PATCH [email protected] 1 participant 1 message Jun 30, 2015

Subsystems (3)
Networking Stable branch IPv4/IPv6
lore
Share & tools
lore

[email protected]

author maintainer Applied
#0 lore raw

This is a note to let you know that I've just added the patch titled

ipv6: prevent fib6_run_gc() contention

to the 3.10-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

The filename of the patch is:
ipv6-prevent-fib6_run_gc-contention.patch
and it can be found in the queue-3.10 subdirectory.

If you, or anyone else, feels it should not be added to the stable tree, please let <[email protected]> know about it.

Quoted 1 line

From 2ac3ac8f86f2fe065d746d9a9abaca867adec577 Mon Sep 17 00:00:00 2001

From: Michal Kubeček <[email protected]>
Date: Thu, 1 Aug 2013 10:04:14 +0200
Subject: ipv6: prevent fib6_run_gc() contention

From: Michal Kubeček <[email protected]>

commit 2ac3ac8f86f2fe065d746d9a9abaca867adec577 upstream.

On a high-traffic router with many processors and many IPv6 dst entries, soft lockup in fib6_run_gc() can occur when number of entries reaches gc_thresh.

This happens because fib6_run_gc() uses fib6_gc_lock to allow only one thread to run the garbage collector but ip6_dst_gc() doesn't update net->ipv6.ip6_rt_last_gc until fib6_run_gc() returns. On a system with many entries, this can take some time so that in the meantime, other threads pass the tests in ip6_dst_gc() (ip6_rt_last_gc is still not updated) and wait for the lock. They then have to run the garbage collector one after another which blocks them for quite long.

Resolve this by replacing special value ~0UL of expire parameter to fib6_run_gc() by explicit "force" parameter to choose between spin_lock_bh() and spin_trylock_bh() and call fib6_run_gc() with force=false if gc_thresh is reached but not max_size.

Signed-off-by Michal Kubecek <[email protected]>
Signed-off-by David S. Miller <[email protected]>
Cc Konstantin Khlebnikov <[email protected]>
Signed-off-by Greg Kroah-Hartman <[email protected]>
---
 include/net/ip6_fib.h |    2 +-
 net/ipv6/ip6_fib.c    |   19 ++++++++-----------
 net/ipv6/ndisc.c      |    4 ++--
 net/ipv6/route.c      |    4 ++--
 4 files changed, 13 insertions(+), 16 deletions(-)
include/net/ip6_fib.h +1 -1
@@ -301,7 +301,7 @@ extern void			inet6_rt_notify(int event,
 						struct nl_info *info);
 
 extern void			fib6_run_gc(unsigned long expires,
-					    struct net *net);
+					    struct net *net, bool force);
 
 extern void			fib6_gc_cleanup(void);
net/ipv6/ip6_fib.c +8 -11
@@ -1648,19 +1648,16 @@ static int fib6_age(struct rt6_info *rt,
 
 static DEFINE_SPINLOCK(fib6_gc_lock);
 
-void fib6_run_gc(unsigned long expires, struct net *net)
+void fib6_run_gc(unsigned long expires, struct net *net, bool force)
 {
-	if (expires != ~0UL) {
+	if (force) {
 		spin_lock_bh(&fib6_gc_lock);
-		gc_args.timeout = expires ? (int)expires :
-			net->ipv6.sysctl.ip6_rt_gc_interval;
-	} else {
-		if (!spin_trylock_bh(&fib6_gc_lock)) {
-			mod_timer(&net->ipv6.ip6_fib_timer, jiffies + HZ);
-			return;
-		}
-		gc_args.timeout = net->ipv6.sysctl.ip6_rt_gc_interval;
+	} else if (!spin_trylock_bh(&fib6_gc_lock)) {
+		mod_timer(&net->ipv6.ip6_fib_timer, jiffies + HZ);
+		return;
 	}
+	gc_args.timeout = expires ? (int)expires :
+			  net->ipv6.sysctl.ip6_rt_gc_interval;
 
 	gc_args.more = icmp6_dst_gc();
 

@@ -1677,7 +1674,7 @@ void fib6_run_gc(unsigned long expires,
 
 static void fib6_gc_timer_cb(unsigned long arg)
 {
-	fib6_run_gc(0, (struct net *)arg);
+	fib6_run_gc(0, (struct net *)arg, true);
 }
 
 static int __net_init fib6_net_init(struct net *net)
net/ipv6/ndisc.c +2 -2
@@ -1584,7 +1584,7 @@ static int ndisc_netdev_event(struct not
 	switch (event) {
 	case NETDEV_CHANGEADDR:
 		neigh_changeaddr(&nd_tbl, dev);
-		fib6_run_gc(~0UL, net);
+		fib6_run_gc(0, net, false);
 		idev = in6_dev_get(dev);
 		if (!idev)
 			break;

@@ -1594,7 +1594,7 @@ static int ndisc_netdev_event(struct not
 		break;
 	case NETDEV_DOWN:
 		neigh_ifdown(&nd_tbl, dev);
-		fib6_run_gc(~0UL, net);
+		fib6_run_gc(0, net, false);
 		break;
 	case NETDEV_NOTIFY_PEERS:
 		ndisc_send_unsol_na(dev);
net/ipv6/route.c +2 -2
@@ -1349,7 +1349,7 @@ static int ip6_dst_gc(struct dst_ops *op
 		goto out;
 
 	net->ipv6.ip6_rt_gc_expire++;
-	fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net);
+	fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, entries > rt_max_size);
 	net->ipv6.ip6_rt_last_gc = now;
 	entries = dst_entries_get_slow(ops);
 	if (entries < ops->gc_thresh)

@@ -2849,7 +2849,7 @@ int ipv6_sysctl_rtcache_flush(ctl_table
 	net = (struct net *)ctl->extra1;
 	delay = net->ipv6.sysctl.flush_delay;
 	proc_dointvec(ctl, write, buffer, lenp, ppos);
-	fib6_run_gc(delay <= 0 ? ~0UL : (unsigned long)delay, net);
+	fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
 	return 0;
 }

Patches currently in stable-queue which might be from [email protected] are

queue-3.10/ipv6-prevent-fib6_run_gc-contention.patch queue-3.10/ipv6-update-ip6_rt_last_gc-every-time-gc-is-run.patch