[PATCH net-next] bonding: use WARN_ON instead of BUG in alb_upper_dev_walk

[PATCH net-next] bonding: use WARN_ON instead of BUG in alb_upper_dev_walk

[Zhengchao Shao]

If failed to allocate "tags" or could not find the final upper device from
start_dev's upper list in bond_verify_device_path(), only the loopback
detection of the current upper device should be affected, and the system is
no need to be panic.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
---
 drivers/net/bonding/bond_alb.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/net/bonding/bond_alb.c b/drivers/net/bonding/bond_alb.c
index dc2c7b979656..5519cc95b966 100644
--- a/drivers/net/bonding/bond_alb.c
+++ b/drivers/net/bonding/bond_alb.c
@@ -984,8 +984,10 @@ static int alb_upper_dev_walk(struct net_device *upper,
 	 */
 	if (netif_is_macvlan(upper) && !strict_match) {
 		tags = bond_verify_device_path(bond->dev, upper, 0);
-		if (IS_ERR_OR_NULL(tags))
-			BUG();
+		if (IS_ERR_OR_NULL(tags)) {
+			WARN_ON(1);
+			return 0;
+		}
 		alb_send_lp_vid(slave, upper->dev_addr,
 				tags[0].vlan_proto, tags[0].vlan_id);
 		kfree(tags);
-- 
2.34.1

Re: [PATCH net-next] bonding: use WARN_ON instead of BUG in alb_upper_dev_walk

[Jay Vosburgh]

Zhengchao Shao <shaozhengchao@huawei.com> wrote:

>If failed to allocate "tags" or could not find the final upper device from
>start_dev's upper list in bond_verify_device_path(), only the loopback
>detection of the current upper device should be affected, and the system is
>no need to be panic.
>
>Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
>---
> drivers/net/bonding/bond_alb.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
>diff --git a/drivers/net/bonding/bond_alb.c b/drivers/net/bonding/bond_alb.c
>index dc2c7b979656..5519cc95b966 100644
>--- a/drivers/net/bonding/bond_alb.c
>+++ b/drivers/net/bonding/bond_alb.c
>@@ -984,8 +984,10 @@ static int alb_upper_dev_walk(struct net_device *upper,
> 	 */
> 	if (netif_is_macvlan(upper) && !strict_match) {
> 		tags = bond_verify_device_path(bond->dev, upper, 0);
>-		if (IS_ERR_OR_NULL(tags))
>-			BUG();
>+		if (IS_ERR_OR_NULL(tags)) {
>+			WARN_ON(1);
>+			return 0;

	This seems reasonable enough, although I'd suggest the using
WARN_ON_ONCE instead of WARN_ON.  Alternatively, this could stay as
WARN_ON if the above also returns non-zero in order to terminate the
netdev_walk_all_upper_dev_rcu walk.  The intent here is to avoid
spamming the log if there's a lot of macvlans above the bond.  If the
allocation in bond_verify_device_path failed, trying again immediately
seems likely to fail as well.

	We could also arrange for whatever called alb_upper_dev_walk to
reschedule at a slightly later time, but I don't think that's worth the
trouble.  The bond will by default resend learning packets once per
second, so issues related to a lost learning packet should resolve
relatively quickly.

	-J

>+		}
> 		alb_send_lp_vid(slave, upper->dev_addr,
> 				tags[0].vlan_proto, tags[0].vlan_id);
> 		kfree(tags);
>-- 
>2.34.1

---
	-Jay Vosburgh, jay.vosburgh@canonical.com

Re: [PATCH net-next] bonding: use WARN_ON instead of BUG in alb_upper_dev_walk

[shaozhengchao]

On 2023/11/14 8:31, Jay Vosburgh wrote:
> Zhengchao Shao <shaozhengchao@huawei.com> wrote:
> 
>> If failed to allocate "tags" or could not find the final upper device from
>> start_dev's upper list in bond_verify_device_path(), only the loopback
>> detection of the current upper device should be affected, and the system is
>> no need to be panic.
>>
>> Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
>> ---
>> drivers/net/bonding/bond_alb.c | 6 ++++--
>> 1 file changed, 4 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/net/bonding/bond_alb.c b/drivers/net/bonding/bond_alb.c
>> index dc2c7b979656..5519cc95b966 100644
>> --- a/drivers/net/bonding/bond_alb.c
>> +++ b/drivers/net/bonding/bond_alb.c
>> @@ -984,8 +984,10 @@ static int alb_upper_dev_walk(struct net_device *upper,
>> 	 */
>> 	if (netif_is_macvlan(upper) && !strict_match) {
>> 		tags = bond_verify_device_path(bond->dev, upper, 0);
>> -		if (IS_ERR_OR_NULL(tags))
>> -			BUG();
>> +		if (IS_ERR_OR_NULL(tags)) {
>> +			WARN_ON(1);
>> +			return 0;
> 
> 	This seems reasonable enough, although I'd suggest the using
> WARN_ON_ONCE instead of WARN_ON.  Alternatively, this could stay as
> WARN_ON if the above also returns non-zero in order to terminate the
> netdev_walk_all_upper_dev_rcu walk.  The intent here is to avoid
> spamming the log if there's a lot of macvlans above the bond.  If the
> allocation in bond_verify_device_path failed, trying again immediately
> seems likely to fail as well.
Hi Jay:
	Thank you for your reply. I do agree with you. I will send v2.

Zhengchao Shao
> 
> 	We could also arrange for whatever called alb_upper_dev_walk to
> reschedule at a slightly later time, but I don't think that's worth the
> trouble.  The bond will by default resend learning packets once per
> second, so issues related to a lost learning packet should resolve
> relatively quickly.
> 
> 	-J
> 
>> +		}
>> 		alb_send_lp_vid(slave, upper->dev_addr,
>> 				tags[0].vlan_proto, tags[0].vlan_id);
>> 		kfree(tags);
>> -- 
>> 2.34.1
> 
> ---
> 	-Jay Vosburgh, jay.vosburgh@canonical.com