Skip to contentSkip to conversation

ipv6: fix missing in6_ifa_put in addrconf

PATCH Applied John Fastabend 4 participants 5 messages Nov 16, 2010 – Nov 17, 2010

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ipv6: fix missing in6_ifa_put in addrconf

Fix ref count bug introduced by

commit 2de795707294972f6c34bae9de713e502c431296

Author: Lorenzo Colitti <[email protected]>
Date:   Wed Oct 27 18:16:49 2010 +0000

ipv6: addrconf: don't remove address state on ifdown if the address is being kept

Fix logic so that addrconf_ifdown() decrements the inet6_ifaddr refcnt correctly with in6_ifa_put().

---

 net/ipv6/addrconf.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)
net/ipv6/addrconf.c +3 -3
index b41ce0f..aaa3ca4 100644
@@ -2754,13 +2754,13 @@ static int addrconf_ifdown(struct net_device *dev, int how)
 			ifa->state = INET6_IFADDR_STATE_DEAD;
 			spin_unlock_bh(&ifa->state_lock);
 
-			if (state == INET6_IFADDR_STATE_DEAD) {
-				in6_ifa_put(ifa);
-			} else {
+			if (state != INET6_IFADDR_STATE_DEAD) {
 				__ipv6_ifa_notify(RTM_DELADDR, ifa);
 				atomic_notifier_call_chain(&inet6addr_chain,
 							   NETDEV_DOWN, ifa);
 			}
+
+			in6_ifa_put(ifa);
 			write_lock_bh(&idev->lock);
 		}
 	}
Acked-by on v1 1/1
#1 lore raw
Quoted 32 lines · John Fastabend

Fix ref count bug introduced by

commit 2de795707294972f6c34bae9de713e502c431296

Author: Lorenzo Colitti <[email protected]>
Date:   Wed Oct 27 18:16:49 2010 +0000

ipv6: addrconf: don't remove address state on ifdown if the address is being kept

Fix logic so that addrconf_ifdown() decrements the inet6_ifaddr refcnt correctly with in6_ifa_put().

---

 net/ipv6/addrconf.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)
net/ipv6/addrconf.c +0 -3
index b41ce0f..aaa3ca4 100644
@@ -2754,13 +2754,13 @@ static int addrconf_ifdown(struct net_device *dev, int how)
 			ifa->state = INET6_IFADDR_STATE_DEAD;
 			spin_unlock_bh(&ifa->state_lock);
 
-			if (state == INET6_IFADDR_STATE_DEAD) {
-				in6_ifa_put(ifa);
-			} else {

net/ipv6/addrconf.c static int addrconf_ifdown(struct net_device *dev, int how) · line 2766

Quoted 10 lines
net/ipv6/addrconf.c +3 -0
+			if (state != INET6_IFADDR_STATE_DEAD) {
 				__ipv6_ifa_notify(RTM_DELADDR, ifa);
 				atomic_notifier_call_chain(&inet6addr_chain,
 							   NETDEV_DOWN, ifa);
 			}
+
+			in6_ifa_put(ifa);
 			write_lock_bh(&idev->lock);
 		}
 	}

Thanks again John

#2 lore raw

Hi, I was hunting for a bug that a colleague discovered. When we exit a network namespace, the ns was hanging with dangling references to dev lo.

When I saw the patch below, I fetched the latest net-next from Daves repository.

And the problem is gone. Was this really the fix, since it seems to me that this is a fix to a recently added patch (dont remove address ....).

Did you fix anything else in the Ipv6 addrconf w.r.g dev_hold/dev_put?

netns still have problems with slab corruption during exit though:

 Nov 17 15:00:20 p04 user.err kernel: Slab corruption: size-4096
start=ffff88000612a000, len=4096
Nov 17 15:00:20 p04 user.err kernel: 020: 6b 6b 6b 6b 6b 6b
Nov 17 15:00:20 p04 user.info kernel:  6b 6b 00 00 00 00 00 00 00
Nov 17 15:00:20 p04 user.info kernel:  00

BR
Anders

Quoted 53 lines · Eric Dumazet
Quoted 32 lines · John Fastabend

Fix ref count bug introduced by

commit 2de795707294972f6c34bae9de713e502c431296

Author: Lorenzo Colitti <[email protected]>
Date:   Wed Oct 27 18:16:49 2010 +0000

ipv6: addrconf: don't remove address state on ifdown if the address is being kept

Fix logic so that addrconf_ifdown() decrements the inet6_ifaddr refcnt correctly with in6_ifa_put().

---

 net/ipv6/addrconf.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)
net/ipv6/addrconf.c +0 -3
index b41ce0f..aaa3ca4 100644
@@ -2754,13 +2754,13 @@ static int addrconf_ifdown(struct net_device *dev, int how)
 			ifa->state = INET6_IFADDR_STATE_DEAD;
 			spin_unlock_bh(&ifa->state_lock);
 
-			if (state == INET6_IFADDR_STATE_DEAD) {
-				in6_ifa_put(ifa);
-			} else {

net/ipv6/addrconf.c static int addrconf_ifdown(struct net_device *dev, int how) · line 2766

Quoted 10 lines
net/ipv6/addrconf.c +3 -0
+			if (state != INET6_IFADDR_STATE_DEAD) {
 				__ipv6_ifa_notify(RTM_DELADDR, ifa);
 				atomic_notifier_call_chain(&inet6addr_chain,
 							   NETDEV_DOWN, ifa);
 			}
+
+			in6_ifa_put(ifa);
 			write_lock_bh(&idev->lock);
 		}
 	}

Thanks again John

--
To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#3 lore raw
Quoted 14 lines · Anders Franzen

Hi, I was hunting for a bug that a colleague discovered. When we exit a network namespace, the ns was hanging with dangling references to dev lo.

When I saw the patch below, I fetched the latest net-next from Daves repository.

And the problem is gone. Was this really the fix, since it seems to me that this is a fix to a recently added patch (dont remove address ....).

Did you fix anything else in the Ipv6 addrconf w.r.g dev_hold/dev_put?

netns still have problems with slab corruption during exit though:

Quoted 10 lines
 Nov 17 15:00:20 p04 user.err kernel: Slab corruption: size-4096
start=ffff88000612a000, len=4096
Nov 17 15:00:20 p04 user.err kernel: 020: 6b 6b 6b 6b 6b 6b
Nov 17 15:00:20 p04 user.info kernel:  6b 6b 00 00 00 00 00 00 00
Nov 17 15:00:20 p04 user.info kernel:  00

BR
Anders

Hi Anders

Several patches of interest :

commit 9d82ca98f71fd686ef2f3017c5e3e6a4871b6e46
(ipv6: fix missing in6_ifa_put in addrconf)

commit 332dd96f7ac15e937088fe11f15cfe0210e8edd1 (net/dst: dst_dev_event() called after other notifiers)

commit ef885afbf8a37689afc1d9d545e2f3e7a8276c17 (net: use rcu_barrier() in rollback_registered_many)

maintainer Applied
#4 lore raw
In reply to John Fastabend

From: John Fastabend <[email protected]>
Date: Mon, 15 Nov 2010 22:29:21 -0800

Quoted 10 lines

Fix ref count bug introduced by

commit 2de795707294972f6c34bae9de713e502c431296

Author: Lorenzo Colitti <[email protected]>
Date:   Wed Oct 27 18:16:49 2010 +0000

ipv6: addrconf: don't remove address state on ifdown if the address is being kept

Quoted 5 lines

Fix logic so that addrconf_ifdown() decrements the inet6_ifaddr refcnt correctly with in6_ifa_put().

Applied, thanks a lot.