On 2019/4/7 22:30, David Ahern wrote:
> On 4/6/19 11:46 PM, linmiaohe wrote:
>> diff --git a/drivers/net/vrf.c b/drivers/net/vrf.c
>> index 7c1430ed0244..4d35d37f225a 100644
>> --- a/drivers/net/vrf.c
>> +++ b/drivers/net/vrf.c
>> @@ -43,6 +43,12 @@
>>
>>  #define FIB_RULE_PREF  1000       /* default preference for FIB rules */
>>
>> +/* The MTU is really irrelevant for VRF except for odd cases, so limit it
>> + * to a reasonable range which both works for IPV4 and IPV6.
>> + */
>> +#define VRF_MIN_MTU	1280      /* same as IPV6_MIN_MTU */
>> +#define VRF_MAX_MTU	0xFFFFU   /* same as IP_MAX_MTU */
> 
> Since VRF does not care about MTU, there is no reason for VRF specific
> macros.
> 
>> +
>>  static unsigned int vrf_net_id;
>>
>>  struct net_vrf {
>> @@ -1274,8 +1280,8 @@ static void vrf_setup(struct net_device *dev)
>>  	/* default to no qdisc; user can add if desired */
>>  	dev->priv_flags |= IFF_NO_QUEUE;
>>
>> -	dev->min_mtu = 0;
>> -	dev->max_mtu = 0;
>> +	dev->min_mtu = VRF_MIN_MTU;
>> +	dev->max_mtu = VRF_MAX_MTU;
> 
> Set these to IPV6_MIN_MTU and ETH_MAX_MTU and add a comment that VRF
> devices do not care about MTU, but if the MTU is set too low then the
> ipv4 and ipv6 protocols are disabled which breaks networking.
> 
> 

Thanks a lot for your advice. I will send v3 latter.