bonding: remap muticast addresses without using dev_close() and dev_open()

This patch fixes commit e36b9d16c6a6d0f59803b3ef04ff3c22c3844c10. The approach there is to call dev_close()/dev_open() whenever the device type is changed in order to remap the device IP multicast addresses to HW multicast addresses. This approach suffers from 2 drawbacks: *. It assumes tha the device is UP when calling dev_close(), or otherwise dev_close() has no affect. It is worth to mention that initscripts (Redhat) and sysconfig (Suse) doesn't act the same in this matter. *. dev_close() has other side affects, like deleting entries from the routing table, which might be unnecessary. The fix here is to directly remap the IP multicast addresses to HW multicast addresses for a bonding device that changes its type, and nothing else. Reported-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Moni Shoua <monis@voltaire.com> Signed-off-by: David S. Miller <davem@davemloft.net>

bonding: remap muticast addresses without using dev_close() and dev_open()
This patch fixes commit e36b9d16c6a6d0f59803b3ef04ff3c22c3844c10. The approach there is to call dev_close()/dev_open() whenever the device type is changed in order to remap the device IP multicast addresses to HW multicast addresses. This approach suffers from 2 drawbacks: *. It assumes tha the device is UP when calling dev_close(), or otherwise dev_close() has no affect. It is worth to mention that initscripts (Redhat) and sysconfig (Suse) doesn't act the same in this matter. *. dev_close() has other side affects, like deleting entries from the routing table, which might be unnecessary. The fix here is to directly remap the IP multicast addresses to HW multicast addresses for a bonding device that changes its type, and nothing else. Reported-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Moni Shoua <monis@voltaire.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Moni Shoua · David S. Miller
1 parent 481a819914
Showing 10 changed files with 82 additions and 6 deletions Side-by-side Diff
drivers/net/bonding/bond_main.c
include/linux/igmp.h
include/linux/netdevice.h
include/linux/notifier.h
include/net/addrconf.h
net/core/dev.c
net/ipv4/devinet.c
net/ipv4/igmp.c
net/ipv6/addrconf.c
net/ipv6/mcast.c
@@ -1211,7 +1211,7 @@
 			write_unlock_bh(&bond->curr_slave_lock);
 			read_unlock(&bond->lock);
  
-			netdev_bonding_change(bond->dev);
+			netdev_bonding_change(bond->dev, NETDEV_BONDING_FAILOVER);
  
 			read_lock(&bond->lock);
 			write_lock_bh(&bond->curr_slave_lock);
  
  
@@ -1469,14 +1469,17 @@
 	 */
 	if (bond->slave_cnt == 0) {
 		if (bond_dev->type != slave_dev->type) {
-			dev_close(bond_dev);
 			pr_debug("%s: change device type from %d to %d\n",
 				bond_dev->name, bond_dev->type, slave_dev->type);
+
+			netdev_bonding_change(bond_dev, NETDEV_BONDING_OLDTYPE);
+
 			if (slave_dev->type != ARPHRD_ETHER)
 				bond_setup_by_slave(bond_dev, slave_dev);
 			else
 				ether_setup(bond_dev);
-			dev_open(bond_dev);
+
+			netdev_bonding_change(bond_dev, NETDEV_BONDING_NEWTYPE);
 		}
 	} else if (bond_dev->type != slave_dev->type) {
 		pr_err(DRV_NAME ": %s ether type (%d) is different "
@@ -233,6 +233,8 @@
 extern void ip_mc_destroy_dev(struct in_device *);
 extern void ip_mc_up(struct in_device *);
 extern void ip_mc_down(struct in_device *);
+extern void ip_mc_unmap(struct in_device *);
+extern void ip_mc_remap(struct in_device *);
 extern void ip_mc_dec_group(struct in_device *in_dev, __be32 addr);
 extern void ip_mc_inc_group(struct in_device *in_dev, __be32 addr);
 extern void ip_mc_rejoin_group(struct ip_mc_list *im);
@@ -1873,7 +1873,8 @@
 extern int		dev_set_promiscuity(struct net_device *dev, int inc);
 extern int		dev_set_allmulti(struct net_device *dev, int inc);
 extern void		netdev_state_change(struct net_device *dev);
-extern void		netdev_bonding_change(struct net_device *dev);
+extern void		netdev_bonding_change(struct net_device *dev,
+					      unsigned long event);
 extern void		netdev_features_change(struct net_device *dev);
 /* Load a device via the kmod */
 extern void		dev_load(struct net *net, const char *name);
@@ -199,6 +199,8 @@
 #define NETDEV_FEAT_CHANGE	0x000B
 #define NETDEV_BONDING_FAILOVER 0x000C
 #define NETDEV_PRE_UP		0x000D
+#define NETDEV_BONDING_OLDTYPE  0x000E
+#define NETDEV_BONDING_NEWTYPE  0x000F
  
 #define SYS_DOWN	0x0001	/* Notify of system down */
 #define SYS_RESTART	SYS_DOWN
@@ -143,6 +143,8 @@
 extern int ipv6_dev_mc_dec(struct net_device *dev, const struct in6_addr *addr);
 extern void ipv6_mc_up(struct inet6_dev *idev);
 extern void ipv6_mc_down(struct inet6_dev *idev);
+extern void ipv6_mc_unmap(struct inet6_dev *idev);
+extern void ipv6_mc_remap(struct inet6_dev *idev);
 extern void ipv6_mc_init_dev(struct inet6_dev *idev);
 extern void ipv6_mc_destroy_dev(struct inet6_dev *idev);
 extern void addrconf_dad_failure(struct inet6_ifaddr *ifp);
@@ -1017,9 +1017,9 @@
 }
 EXPORT_SYMBOL(netdev_state_change);
  
-void netdev_bonding_change(struct net_device *dev)
+void netdev_bonding_change(struct net_device *dev, unsigned long event)
 {
-	call_netdevice_notifiers(NETDEV_BONDING_FAILOVER, dev);
+	call_netdevice_notifiers(event, dev);
 }
 EXPORT_SYMBOL(netdev_bonding_change);
  
@@ -1087,6 +1087,12 @@
 	case NETDEV_DOWN:
 		ip_mc_down(in_dev);
 		break;
+	case NETDEV_BONDING_OLDTYPE:
+		ip_mc_unmap(in_dev);
+		break;
+	case NETDEV_BONDING_NEWTYPE:
+		ip_mc_remap(in_dev);
+		break;
 	case NETDEV_CHANGEMTU:
 		if (inetdev_valid_mtu(dev->mtu))
 			break;
@@ -1298,6 +1298,28 @@
 	}
 }
  
+/* Device changing type */
+
+void ip_mc_unmap(struct in_device *in_dev)
+{
+	struct ip_mc_list *i;
+
+	ASSERT_RTNL();
+
+	for (i = in_dev->mc_list; i; i = i->next)
+		igmp_group_dropped(i);
+}
+
+void ip_mc_remap(struct in_device *in_dev)
+{
+	struct ip_mc_list *i;
+
+	ASSERT_RTNL();
+
+	for (i = in_dev->mc_list; i; i = i->next)
+		igmp_group_added(i);
+}
+
 /* Device going down */
  
 void ip_mc_down(struct in_device *in_dev)
@@ -137,6 +137,8 @@
 static void addrconf_join_anycast(struct inet6_ifaddr *ifp);
 static void addrconf_leave_anycast(struct inet6_ifaddr *ifp);
  
+static void addrconf_bonding_change(struct net_device *dev,
+				    unsigned long event);
 static int addrconf_ifdown(struct net_device *dev, int how);
  
 static void addrconf_dad_start(struct inet6_ifaddr *ifp, u32 flags);
@@ -2582,6 +2584,10 @@
 				return notifier_from_errno(err);
 		}
 		break;
+	case NETDEV_BONDING_OLDTYPE:
+	case NETDEV_BONDING_NEWTYPE:
+		addrconf_bonding_change(dev, event);
+		break;
 	}
  
 	return NOTIFY_OK;
@@ -2594,6 +2600,19 @@
 	.notifier_call = addrconf_notify,
 	.priority = 0
 };
+
+static void addrconf_bonding_change(struct net_device *dev, unsigned long event)
+{
+	struct inet6_dev *idev;
+	ASSERT_RTNL();
+
+	idev = __in6_dev_get(dev);
+
+	if (event == NETDEV_BONDING_NEWTYPE)
+		ipv6_mc_remap(idev);
+	else if (event == NETDEV_BONDING_OLDTYPE)
+		ipv6_mc_unmap(idev);
+}
  
 static int addrconf_ifdown(struct net_device *dev, int how)
 {
@@ -2249,6 +2249,25 @@
 	ma_put(ma);
 }
  
+/* Device changing type */
+
+void ipv6_mc_unmap(struct inet6_dev *idev)
+{
+	struct ifmcaddr6 *i;
+
+	/* Install multicast list, except for all-nodes (already installed) */
+
+	read_lock_bh(&idev->lock);
+	for (i = idev->mc_list; i; i = i->next)
+		igmp6_group_dropped(i);
+	read_unlock_bh(&idev->lock);
+}
+
+void ipv6_mc_remap(struct inet6_dev *idev)
+{
+	ipv6_mc_up(idev);
+}
+
 /* Device going down */
  
 void ipv6_mc_down(struct inet6_dev *idev)
...	...	@@ -1211,7 +1211,7 @@
1211	1211	write_unlock_bh(&bond->curr_slave_lock);
1212	1212	read_unlock(&bond->lock);
1213	1213
1214		- netdev_bonding_change(bond->dev);
	1214	+ netdev_bonding_change(bond->dev, NETDEV_BONDING_FAILOVER);
1215	1215
1216	1216	read_lock(&bond->lock);
1217	1217	write_lock_bh(&bond->curr_slave_lock);
1218	1218
1219	1219
...	...	@@ -1469,14 +1469,17 @@
1469	1469	*/
1470	1470	if (bond->slave_cnt == 0) {
1471	1471	if (bond_dev->type != slave_dev->type) {
1472		- dev_close(bond_dev);
1473	1472	pr_debug("%s: change device type from %d to %d\n",
1474	1473	bond_dev->name, bond_dev->type, slave_dev->type);
	1474	+
	1475	+ netdev_bonding_change(bond_dev, NETDEV_BONDING_OLDTYPE);
	1476	+
1475	1477	if (slave_dev->type != ARPHRD_ETHER)
1476	1478	bond_setup_by_slave(bond_dev, slave_dev);
1477	1479	else
1478	1480	ether_setup(bond_dev);
1479		- dev_open(bond_dev);
	1481	+
	1482	+ netdev_bonding_change(bond_dev, NETDEV_BONDING_NEWTYPE);
1480	1483	}
1481	1484	} else if (bond_dev->type != slave_dev->type) {
1482	1485	pr_err(DRV_NAME ": %s ether type (%d) is different "
...	...	@@ -233,6 +233,8 @@
233	233	extern void ip_mc_destroy_dev(struct in_device *);
234	234	extern void ip_mc_up(struct in_device *);
235	235	extern void ip_mc_down(struct in_device *);
	236	+extern void ip_mc_unmap(struct in_device *);
	237	+extern void ip_mc_remap(struct in_device *);
236	238	extern void ip_mc_dec_group(struct in_device *in_dev, __be32 addr);
237	239	extern void ip_mc_inc_group(struct in_device *in_dev, __be32 addr);
238	240	extern void ip_mc_rejoin_group(struct ip_mc_list *im);
...	...	@@ -1873,7 +1873,8 @@
1873	1873	extern int dev_set_promiscuity(struct net_device *dev, int inc);
1874	1874	extern int dev_set_allmulti(struct net_device *dev, int inc);
1875	1875	extern void netdev_state_change(struct net_device *dev);
1876		-extern void netdev_bonding_change(struct net_device *dev);
	1876	+extern void netdev_bonding_change(struct net_device *dev,
	1877	+ unsigned long event);
1877	1878	extern void netdev_features_change(struct net_device *dev);
1878	1879	/* Load a device via the kmod */
1879	1880	extern void dev_load(struct net net, const char name);
...	...	@@ -199,6 +199,8 @@
199	199	#define NETDEV_FEAT_CHANGE 0x000B
200	200	#define NETDEV_BONDING_FAILOVER 0x000C
201	201	#define NETDEV_PRE_UP 0x000D
	202	+#define NETDEV_BONDING_OLDTYPE 0x000E
	203	+#define NETDEV_BONDING_NEWTYPE 0x000F
202	204
203	205	#define SYS_DOWN 0x0001 /* Notify of system down */
204	206	#define SYS_RESTART SYS_DOWN
...	...	@@ -143,6 +143,8 @@
143	143	extern int ipv6_dev_mc_dec(struct net_device dev, const struct in6_addr addr);
144	144	extern void ipv6_mc_up(struct inet6_dev *idev);
145	145	extern void ipv6_mc_down(struct inet6_dev *idev);
	146	+extern void ipv6_mc_unmap(struct inet6_dev *idev);
	147	+extern void ipv6_mc_remap(struct inet6_dev *idev);
146	148	extern void ipv6_mc_init_dev(struct inet6_dev *idev);
147	149	extern void ipv6_mc_destroy_dev(struct inet6_dev *idev);
148	150	extern void addrconf_dad_failure(struct inet6_ifaddr *ifp);
...	...	@@ -1017,9 +1017,9 @@
1017	1017	}
1018	1018	EXPORT_SYMBOL(netdev_state_change);
1019	1019
1020		-void netdev_bonding_change(struct net_device *dev)
	1020	+void netdev_bonding_change(struct net_device *dev, unsigned long event)
1021	1021	{
1022		- call_netdevice_notifiers(NETDEV_BONDING_FAILOVER, dev);
	1022	+ call_netdevice_notifiers(event, dev);
1023	1023	}
1024	1024	EXPORT_SYMBOL(netdev_bonding_change);
1025	1025
...	...	@@ -1087,6 +1087,12 @@
1087	1087	case NETDEV_DOWN:
1088	1088	ip_mc_down(in_dev);
1089	1089	break;
	1090	+ case NETDEV_BONDING_OLDTYPE:
	1091	+ ip_mc_unmap(in_dev);
	1092	+ break;
	1093	+ case NETDEV_BONDING_NEWTYPE:
	1094	+ ip_mc_remap(in_dev);
	1095	+ break;
1090	1096	case NETDEV_CHANGEMTU:
1091	1097	if (inetdev_valid_mtu(dev->mtu))
1092	1098	break;
...	...	@@ -1298,6 +1298,28 @@
1298	1298	}
1299	1299	}
1300	1300
	1301	+/* Device changing type */
	1302	+
	1303	+void ip_mc_unmap(struct in_device *in_dev)
	1304	+{
	1305	+ struct ip_mc_list *i;
	1306	+
	1307	+ ASSERT_RTNL();
	1308	+
	1309	+ for (i = in_dev->mc_list; i; i = i->next)
	1310	+ igmp_group_dropped(i);
	1311	+}
	1312	+
	1313	+void ip_mc_remap(struct in_device *in_dev)
	1314	+{
	1315	+ struct ip_mc_list *i;
	1316	+
	1317	+ ASSERT_RTNL();
	1318	+
	1319	+ for (i = in_dev->mc_list; i; i = i->next)
	1320	+ igmp_group_added(i);
	1321	+}
	1322	+
1301	1323	/* Device going down */
1302	1324
1303	1325	void ip_mc_down(struct in_device *in_dev)
...	...	@@ -137,6 +137,8 @@
137	137	static void addrconf_join_anycast(struct inet6_ifaddr *ifp);
138	138	static void addrconf_leave_anycast(struct inet6_ifaddr *ifp);
139	139
	140	+static void addrconf_bonding_change(struct net_device *dev,
	141	+ unsigned long event);
140	142	static int addrconf_ifdown(struct net_device *dev, int how);
141	143
142	144	static void addrconf_dad_start(struct inet6_ifaddr *ifp, u32 flags);
...	...	@@ -2582,6 +2584,10 @@
2582	2584	return notifier_from_errno(err);
2583	2585	}
2584	2586	break;
	2587	+ case NETDEV_BONDING_OLDTYPE:
	2588	+ case NETDEV_BONDING_NEWTYPE:
	2589	+ addrconf_bonding_change(dev, event);
	2590	+ break;
2585	2591	}
2586	2592
2587	2593	return NOTIFY_OK;
...	...	@@ -2594,6 +2600,19 @@
2594	2600	.notifier_call = addrconf_notify,
2595	2601	.priority = 0
2596	2602	};
	2603	+
	2604	+static void addrconf_bonding_change(struct net_device *dev, unsigned long event)
	2605	+{
	2606	+ struct inet6_dev *idev;
	2607	+ ASSERT_RTNL();
	2608	+
	2609	+ idev = __in6_dev_get(dev);
	2610	+
	2611	+ if (event == NETDEV_BONDING_NEWTYPE)
	2612	+ ipv6_mc_remap(idev);
	2613	+ else if (event == NETDEV_BONDING_OLDTYPE)
	2614	+ ipv6_mc_unmap(idev);
	2615	+}
2597	2616
2598	2617	static int addrconf_ifdown(struct net_device *dev, int how)
2599	2618	{
...	...	@@ -2249,6 +2249,25 @@
2249	2249	ma_put(ma);
2250	2250	}
2251	2251
	2252	+/* Device changing type */
	2253	+
	2254	+void ipv6_mc_unmap(struct inet6_dev *idev)
	2255	+{
	2256	+ struct ifmcaddr6 *i;
	2257	+
	2258	+ /* Install multicast list, except for all-nodes (already installed) */
	2259	+
	2260	+ read_lock_bh(&idev->lock);
	2261	+ for (i = idev->mc_list; i; i = i->next)
	2262	+ igmp6_group_dropped(i);
	2263	+ read_unlock_bh(&idev->lock);
	2264	+}
	2265	+
	2266	+void ipv6_mc_remap(struct inet6_dev *idev)
	2267	+{
	2268	+ ipv6_mc_up(idev);
	2269	+}
	2270	+
2252	2271	/* Device going down */
2253	2272
2254	2273	void ipv6_mc_down(struct inet6_dev *idev)