Re: [PATCH] brcmfmac: detect & reject faked packet generated by a firmware

[Arend van Spriel <arend.vanspriel@broadcom.com>]

On 1/31/2018 2:11 PM, Rafał Miłecki wrote:
>>>   void brcmf_netif_rx(struct brcmf_if *ifp, struct sk_buff *skb)
>>>   {
>>> +    if (!brcmf_is_valid_skb(skb)) {
>>> +        brcmu_pkt_buf_free_skb(skb);
>>
>> Maybe we should add a driver stat for this although I better have a
>> look into the root cause of this.
>
> It seems there are following stats fields we can use:
> 1) rx_errors
> 2) rx_dropped
> 3) rx_length_errors
> 4) rx_over_errors
> 5) rx_crc_errors
> 6) rx_frame_errors
> 7) rx_fifo_errors
> 8) rx_missed_errors
>
> Which one do you think may fit this case the best?

Those are actually netdev stats, right? Not sure I want to have those, 
but if any I would say 3) fits best.

Regards,
Arend