[PATCH net-next,v5] bonding: return -ENOMEM instead of BUG in alb_upper_dev_walk

[PATCH net-next,v5] bonding: return -ENOMEM 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.
So just return -ENOMEM in alb_upper_dev_walk to stop walking.

I also think that the following function calls
netdev_walk_all_upper_dev_rcu
---->>>alb_upper_dev_walk
---------->>>bond_verify_device_path
From this way, "end device" can eventually be obtained from "start device"
in bond_verify_device_path, IS_ERR(tags) could be instead of
IS_ERR_OR_NULL(tags) in alb_upper_dev_walk.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Acked-by: Jay Vosburgh <jay.vosburgh@canonical.com>
---
v5: drop print information, if the memory allocation fails, the mm layer 
    will emit a lot warning comprising the backtrace
v4: print information instead of warn
v3: return -ENOMEM instead of zero to stop walk
v2: use WARN_ON_ONCE instead of WARN_ON
---
 drivers/net/bonding/bond_alb.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/bonding/bond_alb.c b/drivers/net/bonding/bond_alb.c
index dc2c7b979656..7edf0fd58c34 100644
--- a/drivers/net/bonding/bond_alb.c
+++ b/drivers/net/bonding/bond_alb.c
@@ -985,7 +985,8 @@ 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();
+			return -ENOMEM;
+
 		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,v5] bonding: return -ENOMEM instead of BUG in alb_upper_dev_walk

[shaozhengchao]

On 2023/11/21 20:58, Zhengchao Shao 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.
> So just return -ENOMEM in alb_upper_dev_walk to stop walking.
> 
> I also think that the following function calls
> netdev_walk_all_upper_dev_rcu
> ---->>>alb_upper_dev_walk
> ---------->>>bond_verify_device_path
>>From this way, "end device" can eventually be obtained from "start device"
> in bond_verify_device_path, IS_ERR(tags) could be instead of
> IS_ERR_OR_NULL(tags) in alb_upper_dev_walk.
> 
> Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
> Acked-by: Jay Vosburgh <jay.vosburgh@canonical.com>
> ---
> v5: drop print information, if the memory allocation fails, the mm layer
>      will emit a lot warning comprising the backtrace
> v4: print information instead of warn
> v3: return -ENOMEM instead of zero to stop walk
> v2: use WARN_ON_ONCE instead of WARN_ON
> ---
>   drivers/net/bonding/bond_alb.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/bonding/bond_alb.c b/drivers/net/bonding/bond_alb.c
> index dc2c7b979656..7edf0fd58c34 100644
> --- a/drivers/net/bonding/bond_alb.c
> +++ b/drivers/net/bonding/bond_alb.c
> @@ -985,7 +985,8 @@ 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();
> +			return -ENOMEM;
> +
>   		alb_send_lp_vid(slave, upper->dev_addr,
>   				tags[0].vlan_proto, tags[0].vlan_id);
>   		kfree(tags);
Hi Paolo:
	I find that v4 has been merged into net-next. So v5 is not
needed. But should I send another patch to drop that print info?

Thanks

Zhengchao Shao

Re: [PATCH net-next,v5] bonding: return -ENOMEM instead of BUG in alb_upper_dev_walk

[Paolo Abeni]

On Wed, 2023-11-22 at 10:04 +0800, shaozhengchao wrote:
> 
> On 2023/11/21 20:58, Zhengchao Shao 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.
> > So just return -ENOMEM in alb_upper_dev_walk to stop walking.
> > 
> > I also think that the following function calls
> > netdev_walk_all_upper_dev_rcu
> > ---->>>alb_upper_dev_walk
> > ---------->>>bond_verify_device_path
> > > From this way, "end device" can eventually be obtained from "start device"
> > in bond_verify_device_path, IS_ERR(tags) could be instead of
> > IS_ERR_OR_NULL(tags) in alb_upper_dev_walk.
> > 
> > Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
> > Acked-by: Jay Vosburgh <jay.vosburgh@canonical.com>
> > ---
> > v5: drop print information, if the memory allocation fails, the mm layer
> >      will emit a lot warning comprising the backtrace
> > v4: print information instead of warn
> > v3: return -ENOMEM instead of zero to stop walk
> > v2: use WARN_ON_ONCE instead of WARN_ON
> > ---
> >   drivers/net/bonding/bond_alb.c | 3 ++-
> >   1 file changed, 2 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/net/bonding/bond_alb.c b/drivers/net/bonding/bond_alb.c
> > index dc2c7b979656..7edf0fd58c34 100644
> > --- a/drivers/net/bonding/bond_alb.c
> > +++ b/drivers/net/bonding/bond_alb.c
> > @@ -985,7 +985,8 @@ 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();
> > +			return -ENOMEM;
> > +
> >   		alb_send_lp_vid(slave, upper->dev_addr,
> >   				tags[0].vlan_proto, tags[0].vlan_id);
> >   		kfree(tags);
> Hi Paolo:
> 	I find that v4 has been merged into net-next. 

I'm sorry, that was due to PEBKAC here.

> So v5 is not
> needed. But should I send another patch to drop that print info?

Yes please, send a follow-up just dropping the print.

Thanks!

Paolo

Re: [PATCH net-next,v5] bonding: return -ENOMEM instead of BUG in alb_upper_dev_walk

[shaozhengchao]

On 2023/11/23 0:16, Paolo Abeni wrote:
> On Wed, 2023-11-22 at 10:04 +0800, shaozhengchao wrote:
>>
>> On 2023/11/21 20:58, Zhengchao Shao 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.
>>> So just return -ENOMEM in alb_upper_dev_walk to stop walking.
>>>
>>> I also think that the following function calls
>>> netdev_walk_all_upper_dev_rcu
>>> ---->>>alb_upper_dev_walk
>>> ---------->>>bond_verify_device_path
>>>>  From this way, "end device" can eventually be obtained from "start device"
>>> in bond_verify_device_path, IS_ERR(tags) could be instead of
>>> IS_ERR_OR_NULL(tags) in alb_upper_dev_walk.
>>>
>>> Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
>>> Acked-by: Jay Vosburgh <jay.vosburgh@canonical.com>
>>> ---
>>> v5: drop print information, if the memory allocation fails, the mm layer
>>>       will emit a lot warning comprising the backtrace
>>> v4: print information instead of warn
>>> v3: return -ENOMEM instead of zero to stop walk
>>> v2: use WARN_ON_ONCE instead of WARN_ON
>>> ---
>>>    drivers/net/bonding/bond_alb.c | 3 ++-
>>>    1 file changed, 2 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/net/bonding/bond_alb.c b/drivers/net/bonding/bond_alb.c
>>> index dc2c7b979656..7edf0fd58c34 100644
>>> --- a/drivers/net/bonding/bond_alb.c
>>> +++ b/drivers/net/bonding/bond_alb.c
>>> @@ -985,7 +985,8 @@ 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();
>>> +			return -ENOMEM;
>>> +
>>>    		alb_send_lp_vid(slave, upper->dev_addr,
>>>    				tags[0].vlan_proto, tags[0].vlan_id);
>>>    		kfree(tags);
>> Hi Paolo:
>> 	I find that v4 has been merged into net-next.
> 
> I'm sorry, that was due to PEBKAC here.
> 
>> So v5 is not
>> needed. But should I send another patch to drop that print info?
> 
> Yes please, send a follow-up just dropping the print.
> 
OK， I will do it today.
Thanks

Zhengchao Shao
> Thanks!
> 
> Paolo
> 
>