Skip to contentSkip to conversation

bridge: use the bridge IP addr as source addr of querier

PATCH In review Cong Wang 2 participants 10 messages Apr 29, 2013 – Apr 30, 2013

Subsystems (3)
lore
Share & tools
lore

Cong Wang

author
#0 lore raw

PATCH 1/3 bridge: use the bridge IP addr as source addr of querier

From: Cong Wang <[email protected]>

Quote from Adam:
"If it is believed that the use of 0.0.0.0
as the IP address is what is causing strange behaviour on other devices then is there a good reason that a bridge rather than a router shouldn't be the active querier? If not then using the bridge IP address and having the querier enabled by default may be a reasonable solution (provided that our querier obeys the election rules and shuts up if it sees a query from a lower IP address that isn't 0.0.0.0). Just because a device is the elected querier for IGMP doesn't appear to mean it is required to perform any other routing functions."

Cc Herbert Xu <[email protected]>
Cc Stephen Hemminger <[email protected]>
Cc "David S. Miller" <[email protected]>
Cc Adam Baker <[email protected]>
---
 net/bridge/br_multicast.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
net/bridge/br_multicast.c +2 -1
index 81f2389..6cb937c 100644
@@ -23,6 +23,7 @@
 #include <linux/skbuff.h>
 #include <linux/slab.h>
 #include <linux/timer.h>
+#include <linux/inetdevice.h>
 #include <net/ip.h>
 #if IS_ENABLED(CONFIG_IPV6)
 #include <net/ipv6.h>
@@ -381,7 +382,7 @@ static struct sk_buff *br_ip4_multicast_alloc_query(struct net_bridge *br,
 	iph->frag_off = htons(IP_DF);
 	iph->ttl = 1;
 	iph->protocol = IPPROTO_IGMP;
-	iph->saddr = 0;
+	iph->saddr = inet_select_addr(br->dev, 0, RT_SCOPE_LINK);
 	iph->daddr = htonl(INADDR_ALLHOSTS_GROUP);
 	((u8 *)&iph[1])[0] = IPOPT_RA;
 	((u8 *)&iph[1])[1] = 4;
Signature · 1 line
1.7.7.6
author
#1 lore raw

PATCH 2/3 bridge: only expire the mdb entry when query is received

From: Cong Wang <[email protected]>

Currently we arm the expire timer when the mdb entry is added, however, this causes problem when there is no querier sent out after that.

So we should only arm the timer when a corresponding query is received, as suggested by Herbert.

Cc Herbert Xu <[email protected]>
Cc Stephen Hemminger <[email protected]>
Cc "David S. Miller" <[email protected]>
Cc Adam Baker <[email protected]>
---
 net/bridge/br_multicast.c |   13 ++++++-------
 1 files changed, 6 insertions(+), 7 deletions(-)
net/bridge/br_multicast.c +6 -7
index 6cb937c..a821a4e 100644
@@ -616,8 +616,6 @@ rehash:
 
 	mp->br = br;
 	mp->addr = *group;
-	setup_timer(&mp->timer, br_multicast_group_expired,
-		    (unsigned long)mp);
 
 	hlist_add_head_rcu(&mp->hlist[mdb->ver], &mdb->mhash[hash]);
 	mdb->size++;
@@ -655,7 +653,6 @@ static int br_multicast_add_group(struct net_bridge *br,
 	struct net_bridge_mdb_entry *mp;
 	struct net_bridge_port_group *p;
 	struct net_bridge_port_group __rcu **pp;
-	unsigned long now = jiffies;
 	int err;
 
 	spin_lock(&br->multicast_lock);
@@ -670,7 +667,6 @@ static int br_multicast_add_group(struct net_bridge *br,
 
 	if (!port) {
 		mp->mglist = true;
-		mod_timer(&mp->timer, now + br->multicast_membership_interval);
 		goto out;
 	}
 
@@ -678,7 +674,7 @@ static int br_multicast_add_group(struct net_bridge *br,
 	     (p = mlock_dereference(*pp, br)) != NULL;
 	     pp = &p->next) {
 		if (p->port == port)
-			goto found;
+			goto out;
 		if ((unsigned long)p->port < (unsigned long)port)
 			break;
 	}
@@ -689,8 +685,6 @@ static int br_multicast_add_group(struct net_bridge *br,
 	rcu_assign_pointer(*pp, p);
 	br_mdb_notify(br->dev, port, group, RTM_NEWMDB);
 
-found:
-	mod_timer(&p->timer, now + br->multicast_membership_interval);
 out:
 	err = 0;
 
@@ -1130,6 +1124,9 @@ static int br_ip4_multicast_query(struct net_bridge *br,
 	if (!mp)
 		goto out;
 
+	setup_timer(&mp->timer, br_multicast_group_expired, (unsigned long)mp);
+	mod_timer(&mp->timer, now + br->multicast_membership_interval);
+
 	max_delay *= br->multicast_last_member_count;
 
 	if (mp->mglist &&
@@ -1204,6 +1201,8 @@ static int br_ip6_multicast_query(struct net_bridge *br,
 	if (!mp)
 		goto out;
 
+	setup_timer(&mp->timer, br_multicast_group_expired, (unsigned long)mp);
+	mod_timer(&mp->timer, now + br->multicast_membership_interval);
 	max_delay *= br->multicast_last_member_count;
 	if (mp->mglist &&
 	    (timer_pending(&mp->timer) ?
Signature · 1 line
1.7.7.6
author
#2 lore raw

Re: [Patch net-next 2/3] bridge: only expire the mdb entry when query is received

Quoted 8 lines · Cong Wang

From: Cong Wang <[email protected]>

Currently we arm the expire timer when the mdb entry is added, however, this causes problem when there is no querier sent out after that.

So we should only arm the timer when a corresponding query is received, as suggested by Herbert.

One problem with this solution is that the temp mdb entry will no longer be expired automatically any more, since no query no timer.

I am wondering if we should find other way to fix it, for example, touching the timer as long as there is multicast traffic (non-IGMP).

Thanks!

#3 lore raw

Re: [Patch net-next 2/3] bridge: only expire the mdb entry when query is received

Quoted 6 lines · Cong Wang

One problem with this solution is that the temp mdb entry will no longer be expired automatically any more, since no query no timer.

I am wondering if we should find other way to fix it, for example, touching the timer as long as there is multicast traffic (non-IGMP).

mdb entries represent group subscriptions. So what you're saying is that group subscriptions aren't expiring.

Which isn't a problem at all since if there is no querier then group subscriptions shouldn't expire. There has to be at least one querier in the network for this thing to work. Otherwise it just degenerates into a non-snooping switch, which is OK.

Cheers,

Signature · 3 lines
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
author
#4 lore raw

Re: [Patch net-next 2/3] bridge: only expire the mdb entry when query is received

Quoted 8 lines · Herbert Xu
Quoted 6 lines · Cong Wang

One problem with this solution is that the temp mdb entry will no longer be expired automatically any more, since no query no timer.

I am wondering if we should find other way to fix it, for example, touching the timer as long as there is multicast traffic (non-IGMP).

Quoted 7 lines

mdb entries represent group subscriptions. So what you're saying is that group subscriptions aren't expiring.

Which isn't a problem at all since if there is no querier then group subscriptions shouldn't expire. There has to be at least one querier in the network for this thing to work. Otherwise it just degenerates into a non-snooping switch, which is OK.

Makes sense. I will put this in the changelog.

Thanks.

author
#5 lore raw

PATCH 3/3 bridge: send query as soon as leave is received

In reply to Cong Wang

From: Cong Wang <[email protected]>

Continue sending queries when leave is received if the user marks it as a querier.

Cc Herbert Xu <[email protected]>
Cc Stephen Hemminger <[email protected]>
Cc "David S. Miller" <[email protected]>
Cc Adam Baker <[email protected]>
---
 net/bridge/br_multicast.c |   11 +++++++++++
 1 files changed, 11 insertions(+), 0 deletions(-)
net/bridge/br_multicast.c +11 -0
index a821a4e..03a8f2b 100644
@@ -1246,6 +1246,17 @@ static void br_multicast_leave_group(struct net_bridge *br,
 	if (!mp)
 		goto out;
 
+	if (br->multicast_querier) {
+		mod_timer(&br->multicast_query_timer, jiffies);
+		list_for_each_entry(port, &br->port_list, list) {
+			if (port->state == BR_STATE_DISABLED ||
+			    port->state == BR_STATE_BLOCKING)
+				continue;
+
+			__br_multicast_enable_port(port);
+		}
+	}
+
 	if (port && (port->flags & BR_MULTICAST_FAST_LEAVE)) {
 		struct net_bridge_port_group __rcu **pp;
 
Signature · 1 line
1.7.7.6
#6 lore raw

Re: [Patch net-next 3/3] bridge: send query as soon as leave is received

Quoted 4 lines · Cong Wang

From: Cong Wang <[email protected]>

Continue sending queries when leave is received if the user marks it as a querier.

If we're going to send a query on leave it should be group-specific and not general.

Thanks,

Signature · 3 lines
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
author
#7 lore raw

Re: [Patch net-next 3/3] bridge: send query as soon as leave is received

Quoted 9 lines · Herbert Xu
Quoted 4 lines · Cong Wang

From: Cong Wang <[email protected]>

Continue sending queries when leave is received if the user marks it as a querier.

If we're going to send a query on leave it should be group-specific and not general.

Right, I will fix it.

#8 lore raw
In reply to Cong Wang
Quoted 12 lines · Cong Wang

From: Cong Wang <[email protected]>

Quote from Adam:
"If it is believed that the use of 0.0.0.0
as the IP address is what is causing strange behaviour on other devices then is there a good reason that a bridge rather than a router shouldn't be the active querier? If not then using the bridge IP address and having the querier enabled by default may be a reasonable solution (provided that our querier obeys the election rules and shuts up if it sees a query from a lower IP address that isn't 0.0.0.0). Just because a device is the elected querier for IGMP doesn't appear to mean it is required to perform any other routing functions."

I don't think we should make this the default. Having it as an option though is fine with me. We may also want to allow the user to explicitly set an IP address.

Thanks,

Signature · 3 lines
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
author
#9 lore raw
Quoted 14 lines · Herbert Xu
Quoted 12 lines · Cong Wang

From: Cong Wang <[email protected]>

Quote from Adam:
"If it is believed that the use of 0.0.0.0
as the IP address is what is causing strange behaviour on other devices then is there a good reason that a bridge rather than a router shouldn't be the active querier? If not then using the bridge IP address and having the querier enabled by default may be a reasonable solution (provided that our querier obeys the election rules and shuts up if it sees a query from a lower IP address that isn't 0.0.0.0). Just because a device is the elected querier for IGMP doesn't appear to mean it is required to perform any other routing functions."

Quoted 3 lines

I don't think we should make this the default. Having it as an option though is fine with me. We may also want to allow the user to explicitly set an IP address.

Good point. I think we should at least make it a boolen option, I am not sure if allowing users to explicitly set it is overkill.

Thanks.