On Thu, Oct 5, 2017 at 6:59 AM, David Laight <david.lai...@aculab.com> wrote:
> From: William Tu
>> Sent: 05 October 2017 01:14
>> Similarly to early patch for erspan_xmit(), the ARPHDR_ETHER device
>> is the length of the whole ether packet. So skb->len should subtract
>> the dev->hard_header_len.
>>
>> Fixes: 1a66a836da63 ("gre: add collect_md mode to ERSPAN tunnel")
>> Signed-off-by: William Tu <u9012...@gmail.com>
>> Cc: Xin Long <lucien....@gmail.com>
>> ---
>> net/ipv4/ip_gre.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
>> index b279c325c7f6..10b21fe5b3a6 100644
>> --- a/net/ipv4/ip_gre.c
>> +++ b/net/ipv4/ip_gre.c
>> @@ -579,7 +579,7 @@ static void erspan_fb_xmit(struct sk_buff *skb, struct
>> net_device *dev,
>> if (gre_handle_offloads(skb, false))
>> goto err_free_rt;
>>
>> - if (skb->len > dev->mtu) {
>> + if (skb->len - dev->hard_header_len > dev->mtu) {
>
> Can you guarantee that skb->len > dev_hard_header_len?
> It is probably safer to check skb->len > dev->hard_header_len + dev->mtu
> since that addition isn't going to overflow.
Sure, I will fix it.
>
>> pskb_trim(skb, dev->mtu);
>> truncate = true;
>
> Is that pskb_trim() now truncating to the correct size?
You're right, now I should truncate to (dev->mtu + dev_hard_header_len)
Thanks
William
