Re: ICMP packets - ll_temac with Microblaze

[Michael Wang <wangyun@linux.vnet.ibm.com>]

On 12/22/2011 06:46 PM, Eric Dumazet wrote:

> Le jeudi 22 décembre 2011 à 18:32 +0800, Michael Wang a écrit :
>> On 12/21/2011 06:11 PM, Michal Simek wrote:
>>
>>> Hi Eric and David,
>>>
>>> I have found one problem with ll_temac driver and
>>> this commit: net: more accurate skb truesize
>>> sha1: 87fb4b7b533073eeeaed0b6bf7c2328995f6c075
>>>
>>> The problem is only with icmp packets from the target. It is sent and
>>> driver receive it
>>> but it is not proceed to the application.
>>>
>>
>> Hi, Michal
>>
>> What's the type of icmp you are using? such as "EchoReps", we can find
>> the actually handler routine by this type.
>>
>> And you said the packet already received by driver, can you tell me the
>> way you used to confirm this?
>>
>> And have you checked when was the icmp packet being dropped, is it in
>> icmp_rcv or before or later?
> 
> Packet was dropped right before being queued in RAW socket
> receive_queue, because of low sk_rcvbuf setting (against skb->truesize)
> 

Hi, Eric

So do you mean the way is:

ip_local_deliver_finish --> raw_local_deliver --> raw_v4_input -->
raw_rcv --> raw_rcv_skb --> sock_queue_rcv_skb

And in sock_queue_rcv_skb:

        if (atomic_read(&sk->sk_rmem_alloc) + skb->truesize >=
            (unsigned)sk->sk_rcvbuf) {
                atomic_inc(&sk->sk_drops);
                trace_sock_rcvqueue_full(sk, skb);
                return -ENOMEM;				//drop?
        }

Is this the way the packet being dropped? Is this confirmed by some printk?

Sorry if the question is naive, I just want to make sure I am checking
at the right place.

Regards,
Michael Wang

> 
>