On 6/18/20 4:35 PM, Eric Dumazet wrote:
>
>
> On 6/18/20 4:19 PM, Randy Dunlap wrote:
>>
>> Please see lines 1250-1251.
>>
>>
>> case TIOCINQ:
>> lock_sock(sk);
>> skb = skb_peek(&scp->other_receive_queue);
>> if (skb) {
>> amount = skb->len;
>> } else {
>> skb_queue_walk(&sk->sk_receive_queue, skb) <<<<<
>> amount += skb->len; <<<<<
>> }
>> release_sock(sk);
>> err = put_user(amount, (int __user *)arg);
>> break;
>>
>>
>>
>> or is this some kind of GCC nested function magic?
>>
>
> I do not see a problem
>
> for (bla; bla; bla)
> amount += skb->len;
>
> Seems good to me.
>
OK, I get it (now). Thanks.
>>
>> commit bec571ec762a4cf855ad4446f833086fc154b60e
>> Author: David S. Miller <[email protected]>
>> Date: Thu May 28 16:43:52 2009 -0700
>>
>> decnet: Use SKB queue and list helpers instead of doing it by-hand.
>>
>>
>>
>> thanks.
>>
>
> Also decnet should not be any of our concerns in 2020 ?
Ack.
--
~Randy
