[PATCH] bonding: set device in RLB ARP packet handler

[PATCH] bonding: set device in RLB ARP packet handler

[Greg Edwards]

With commit 6146b1a4, the dev field in the RLB ARP packet handler was
set to NULL to wildcard and accommodate balancing VLANs on top of bonds.

This has the side-effect of the packet handler being called against
other, non RLB-enabled bonds, and a kernel oops results when it tries to
dereference rx_hashtbl in rlb_update_entry_from_arp(), which won't be
set for those bonds, e.g. active-backup.

With the __netif_receive_skb() changes from commit 1f3c8804, frames
received on VLANs correctly make their way to the bond's handler,
so we no longer need to wildcard the device.

Signed-off-by: Greg Edwards <greg.edwards@hp.com>
---
Jay,

The oops can be reproduced by:

modprobe bonding

echo active-backup > /sys/class/net/bond0/bonding/mode
echo 100 > /sys/class/net/bond0/bonding/miimon
ifconfig bond0 xxx.xxx.xxx.xxx netmask xxx.xxx.xxx.xxx
echo +eth0 > /sys/class/net/bond0/bonding/slaves
echo +eth1 > /sys/class/net/bond0/bonding/slaves

echo +bond1 > /sys/class/net/bonding_masters
echo balance-alb > /sys/class/net/bond1/bonding/mode
echo 100 > /sys/class/net/bond1/bonding/miimon
ifconfig bond1 xxx.xxx.xxx.xxx netmask xxx.xxx.xxx.xxx
echo +eth2 > /sys/class/net/bond1/bonding/slaves
echo +eth3 > /sys/class/net/bond1/bonding/slaves

Pass some traffic on bond0.  Boom.


 drivers/net/bonding/bond_alb.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/net/bonding/bond_alb.c b/drivers/net/bonding/bond_alb.c
index df48307..8d7dfd2 100644
--- a/drivers/net/bonding/bond_alb.c
+++ b/drivers/net/bonding/bond_alb.c
@@ -822,7 +822,7 @@ static int rlb_initialize(struct bonding *bond)
 
 	/*initialize packet type*/
 	pk_type->type = cpu_to_be16(ETH_P_ARP);
-	pk_type->dev = NULL;
+	pk_type->dev = bond->dev;
 	pk_type->func = rlb_arp_recv;
 
 	/* register to receive ARPs */
-- 
1.7.0.4

Re: [PATCH] bonding: set device in RLB ARP packet handler

[Andy Gospodarek]

On Thu, Jul 22, 2010 at 3:52 PM, Greg Edwards <greg.edwards@hp.com> wrote:
> With commit 6146b1a4, the dev field in the RLB ARP packet handler was
> set to NULL to wildcard and accommodate balancing VLANs on top of
> bonds.
>
> This has the side-effect of the packet handler being called against
> other, non RLB-enabled bonds, and a kernel oops results when it tries
> to
> dereference rx_hashtbl in rlb_update_entry_from_arp(), which won't be
> set for those bonds, e.g. active-backup.
>
> With the __netif_receive_skb() changes from commit 1f3c8804, frames
> received on VLANs correctly make their way to the bond's handler,
> so we no longer need to wildcard the device.
>

I see this problem as well, but I would propose to fix it another way to
not alter the receive path so close to the release of 2.6.35 and to
catch this for 802.3ad bonds as well.

> Signed-off-by: Greg Edwards <greg.edwards@hp.com>
> ---
> Jay,
>
> The oops can be reproduced by:
>
> modprobe bonding
>
> echo active-backup > /sys/class/net/bond0/bonding/mode
> echo 100 > /sys/class/net/bond0/bonding/miimon
> ifconfig bond0 xxx.xxx.xxx.xxx netmask xxx.xxx.xxx.xxx
> echo +eth0 > /sys/class/net/bond0/bonding/slaves
> echo +eth1 > /sys/class/net/bond0/bonding/slaves
>
> echo +bond1 > /sys/class/net/bonding_masters
> echo balance-alb > /sys/class/net/bond1/bonding/mode
> echo 100 > /sys/class/net/bond1/bonding/miimon
> ifconfig bond1 xxx.xxx.xxx.xxx netmask xxx.xxx.xxx.xxx
> echo +eth2 > /sys/class/net/bond1/bonding/slaves
> echo +eth3 > /sys/class/net/bond1/bonding/slaves
>
> Pass some traffic on bond0.  Boom.
>

bonding: make sure mode-specific handlers handle appropriate frames

This patch will exit out of rlb_arp_recv and bond_3ad_lacpdu_recv early
if the bond receiving the frame isn't using that mode.

Fixes problem reported by Greg Edwards <greg.edwards@hp.com>.

Signed-off-by: Andy Gospodarek <andy@greyhouse.net>

---
 drivers/net/bonding/bond_3ad.c |    3 ++-
 drivers/net/bonding/bond_alb.c |    6 +++---
 2 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/drivers/net/bonding/bond_3ad.c b/drivers/net/bonding/bond_3ad.c
index 822f586..cf7b08d 100644
--- a/drivers/net/bonding/bond_3ad.c
+++ b/drivers/net/bonding/bond_3ad.c
@@ -2463,7 +2463,8 @@ int bond_3ad_lacpdu_recv(struct sk_buff *skb, struct net_device *dev, struct pac
 	struct slave *slave = NULL;
 	int ret = NET_RX_DROP;
 
-	if (!(dev->flags & IFF_MASTER))
+	if (!(dev->flags & IFF_MASTER) ||
+	    (bond->params.mode != BOND_MODE_8023AD))
 		goto out;
 
 	read_lock(&bond->lock);
diff --git a/drivers/net/bonding/bond_alb.c b/drivers/net/bonding/bond_alb.c
index df48307..a82e38c 100644
--- a/drivers/net/bonding/bond_alb.c
+++ b/drivers/net/bonding/bond_alb.c
@@ -353,7 +353,7 @@ static void rlb_update_entry_from_arp(struct bonding *bond, struct arp_pkt *arp)
 
 static int rlb_arp_recv(struct sk_buff *skb, struct net_device *bond_dev, struct packet_type *ptype, struct net_device *orig_dev)
 {
-	struct bonding *bond;
+	struct bonding *bond = netdev_priv(bond_dev);
 	struct arp_pkt *arp = (struct arp_pkt *)skb->data;
 	int res = NET_RX_DROP;
 
@@ -361,7 +361,8 @@ static int rlb_arp_recv(struct sk_buff *skb, struct net_device *bond_dev, struct
 		bond_dev = vlan_dev_real_dev(bond_dev);
 
 	if (!(bond_dev->priv_flags & IFF_BONDING) ||
-	    !(bond_dev->flags & IFF_MASTER))
+	    !(bond_dev->flags & IFF_MASTER) ||
+	    (bond->params.mode != BOND_MODE_ALB))
 		goto out;
 
 	if (!arp) {
@@ -376,7 +377,6 @@ static int rlb_arp_recv(struct sk_buff *skb, struct net_device *bond_dev, struct
 
 	if (arp->op_code == htons(ARPOP_REPLY)) {
 		/* update rx hash table for this ARP */
-		bond = netdev_priv(bond_dev);
 		rlb_update_entry_from_arp(bond, arp);
 		pr_debug("Server received an ARP Reply from client\n");
 	}
-- 
1.6.2.5

Re: [PATCH] bonding: set device in RLB ARP packet handler

[Greg Edwards]

On Fri, Jul 23, 2010 at 07:34:56PM +0000, Andy Gospodarek wrote:
> On Thu, Jul 22, 2010 at 3:52 PM, Greg Edwards <greg.edwards@hp.com> wrote:
>> With commit 6146b1a4, the dev field in the RLB ARP packet handler was
>> set to NULL to wildcard and accommodate balancing VLANs on top of
>> bonds.
>>
>> This has the side-effect of the packet handler being called against
>> other, non RLB-enabled bonds, and a kernel oops results when it tries
>> to
>> dereference rx_hashtbl in rlb_update_entry_from_arp(), which won't be
>> set for those bonds, e.g. active-backup.
>>
>> With the __netif_receive_skb() changes from commit 1f3c8804, frames
>> received on VLANs correctly make their way to the bond's handler,
>> so we no longer need to wildcard the device.
>
> I see this problem as well, but I would propose to fix it another way to
> not alter the receive path so close to the release of 2.6.35 and to
> catch this for 802.3ad bonds as well.

Is the problem demonstrable with 802.3ad bonds?  bond_register_lacpdu()
sets pk_type->dev = bond->dev.

>> Signed-off-by: Greg Edwards <greg.edwards@hp.com>
>> ---
>> Jay,
>>
>> The oops can be reproduced by:
>>
>> modprobe bonding
>>
>> echo active-backup > /sys/class/net/bond0/bonding/mode
>> echo 100 > /sys/class/net/bond0/bonding/miimon
>> ifconfig bond0 xxx.xxx.xxx.xxx netmask xxx.xxx.xxx.xxx
>> echo +eth0 > /sys/class/net/bond0/bonding/slaves
>> echo +eth1 > /sys/class/net/bond0/bonding/slaves
>>
>> echo +bond1 > /sys/class/net/bonding_masters
>> echo balance-alb > /sys/class/net/bond1/bonding/mode
>> echo 100 > /sys/class/net/bond1/bonding/miimon
>> ifconfig bond1 xxx.xxx.xxx.xxx netmask xxx.xxx.xxx.xxx
>> echo +eth2 > /sys/class/net/bond1/bonding/slaves
>> echo +eth3 > /sys/class/net/bond1/bonding/slaves
>>
>> Pass some traffic on bond0.  Boom.
>>
>
> bonding: make sure mode-specific handlers handle appropriate frames
>
> This patch will exit out of rlb_arp_recv and bond_3ad_lacpdu_recv early
> if the bond receiving the frame isn't using that mode.

I had originally thought of doing something like this, but it didn't
seem as clean.  I don't have strong feelings one way or the other,
though.

Greg

Re: [PATCH] bonding: set device in RLB ARP packet handler

[Andy Gospodarek]

On Fri, Jul 23, 2010 at 01:59:47PM -0600, Greg Edwards wrote:
> On Fri, Jul 23, 2010 at 07:34:56PM +0000, Andy Gospodarek wrote:
> > On Thu, Jul 22, 2010 at 3:52 PM, Greg Edwards <greg.edwards@hp.com> wrote:
> >> With commit 6146b1a4, the dev field in the RLB ARP packet handler was
> >> set to NULL to wildcard and accommodate balancing VLANs on top of
> >> bonds.
> >>
> >> This has the side-effect of the packet handler being called against
> >> other, non RLB-enabled bonds, and a kernel oops results when it tries
> >> to
> >> dereference rx_hashtbl in rlb_update_entry_from_arp(), which won't be
> >> set for those bonds, e.g. active-backup.
> >>
> >> With the __netif_receive_skb() changes from commit 1f3c8804, frames
> >> received on VLANs correctly make their way to the bond's handler,
> >> so we no longer need to wildcard the device.
> >
> > I see this problem as well, but I would propose to fix it another way to
> > not alter the receive path so close to the release of 2.6.35 and to
> > catch this for 802.3ad bonds as well.
> 
> Is the problem demonstrable with 802.3ad bonds?  bond_register_lacpdu()
> sets pk_type->dev = bond->dev.
> 

Doh!  You are right.

Plus Jay just posted your patch again, so we can go with that solution.

> >> Signed-off-by: Greg Edwards <greg.edwards@hp.com>
> >> ---
> >> Jay,
> >>
> >> The oops can be reproduced by:
> >>
> >> modprobe bonding
> >>
> >> echo active-backup > /sys/class/net/bond0/bonding/mode
> >> echo 100 > /sys/class/net/bond0/bonding/miimon
> >> ifconfig bond0 xxx.xxx.xxx.xxx netmask xxx.xxx.xxx.xxx
> >> echo +eth0 > /sys/class/net/bond0/bonding/slaves
> >> echo +eth1 > /sys/class/net/bond0/bonding/slaves
> >>
> >> echo +bond1 > /sys/class/net/bonding_masters
> >> echo balance-alb > /sys/class/net/bond1/bonding/mode
> >> echo 100 > /sys/class/net/bond1/bonding/miimon
> >> ifconfig bond1 xxx.xxx.xxx.xxx netmask xxx.xxx.xxx.xxx
> >> echo +eth2 > /sys/class/net/bond1/bonding/slaves
> >> echo +eth3 > /sys/class/net/bond1/bonding/slaves
> >>
> >> Pass some traffic on bond0.  Boom.
> >>
> >
> > bonding: make sure mode-specific handlers handle appropriate frames
> >
> > This patch will exit out of rlb_arp_recv and bond_3ad_lacpdu_recv early
> > if the bond receiving the frame isn't using that mode.
> 
> I had originally thought of doing something like this, but it didn't
> seem as clean.  I don't have strong feelings one way or the other,
> though.
> 
> Greg
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html