Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <320c2a05-e99a-88b4-2f67-11210ae37903@redhat.com>
Date:   Wed, 24 Aug 2022 22:28:58 -0400
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.12.0
Subject: Re: [PATCH net-next v2] bonding: Remove unnecessary check
Content-Language: en-US
To:     Sun Shouxin <sunshouxin@chinatelecom.cn>
Cc:     vfalico@gmail.com, andy@greyhouse.net, davem@davemloft.net,
        edumazet@google.com, kuba@kernel.org, pabeni@redhat.com,
        ast@kernel.org, daniel@iogearbox.net, hawk@kernel.org,
        john.fastabend@gmail.com, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org, bpf@vger.kernel.org,
        huyd12@chinatelecom.cn, Jay Vosburgh <j.vosburgh@gmail.com>
References: <20220824111712.5999-1-sunshouxin@chinatelecom.cn>
 <CAAoacNmKa5oM10J6DTLJ6PANmdS8k80Lcxygv_vXd_0DduXM4A@mail.gmail.com>
From:   Jonathan Toppins <jtoppins@redhat.com>
In-Reply-To: <CAAoacNmKa5oM10J6DTLJ6PANmdS8k80Lcxygv_vXd_0DduXM4A@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 8/24/22 14:07, Jay Vosburgh wrote:
> On 8/24/22, Sun Shouxin <sunshouxin@chinatelecom.cn> wrote:
>> This code is intended to support bond alb interface added to
>> Linux bridge by modifying MAC, however, it doesn't work for
>> one bond alb interface with vlan added to bridge.
>> Since commit d5410ac7b0ba("net:bonding:support balance-alb
>> interface with vlan to bridge"), new logic is adapted to handle
>> bond alb with or without vlan id, and then the code is deprecated.
> 
> I think this could still be clearer; the actual changes relate to the stack of
> interfaces (e.g., eth0 -> bond0 -> vlan123 -> bridge0), not what VLAN tags
> incoming traffic contains.
> 
> The code being removed here is specifically for the case of
> eth0 -> bond0 -> bridge0, without an intermediate VLAN interface
> in the stack (because, if memory serves, netif_is_bridge_port doesn't
> transfer through to the bond if there's a VLAN interface in between).
> 
> Also, this code is for incoming traffic, assigning the bond's MAC to
> traffic arriving on interfaces other than the active interface (which bears
> the bond's MAC in alb mode; the other interfaces have different MACs).
> Commit d5410ac7b0ba affects the balance assignments for outgoing ARP
> traffic.  I'm not sure that d5410 is an exact replacement for the code this
> patch removes.

I would be more comfortable with a change like this if it can be 
demonstrated that an example test case functions as expected before and 
after the change. Could a selftests test be written with veths to 
demonstrate this code is indeed redundant?

-Jon

> 
>>
>> Suggested-by: Hu Yadi <huyd12@chinatelecom.cn>
>> Signed-off-by: Sun Shouxin <sunshouxin@chinatelecom.cn>
>> ---
>>   drivers/net/bonding/bond_main.c | 13 -------------
>>   1 file changed, 13 deletions(-)
>>
>> diff --git a/drivers/net/bonding/bond_main.c
>> b/drivers/net/bonding/bond_main.c
>> index 50e60843020c..6b0f0ce9b9a1 100644
>> --- a/drivers/net/bonding/bond_main.c
>> +++ b/drivers/net/bonding/bond_main.c
>> @@ -1578,19 +1578,6 @@ static rx_handler_result_t bond_handle_frame(struct
>> sk_buff **pskb)
>>
>>   	skb->dev = bond->dev;
>>
>> -	if (BOND_MODE(bond) == BOND_MODE_ALB &&
>> -	    netif_is_bridge_port(bond->dev) &&
>> -	    skb->pkt_type == PACKET_HOST) {
>> -
>> -		if (unlikely(skb_cow_head(skb,
>> -					  skb->data - skb_mac_header(skb)))) {
>> -			kfree_skb(skb);
>> -			return RX_HANDLER_CONSUMED;
>> -		}
>> -		bond_hw_addr_copy(eth_hdr(skb)->h_dest, bond->dev->dev_addr,
>> -				  bond->dev->addr_len);
>> -	}
>> -
>>   	return ret;
>>   }
>>
>> --
>> 2.27.0
>>
>>
>