Re: [PATCH] 1/1 net: packet: Keep 802.1Q VLAN tag in packet on SOCK_DGRAM socket - resend

[Eric Dumazet <eric.dumazet@gmail.com>]

Le 04/01/2010 10:09, Milan Dadok a écrit :
> Keep 802.1Q VLAN tag on non HW vlan accelerated network card received to SOCK_DGRAM socket.
> 
> Signed-off-by: Milan Dadok <milan@dadok.name>
> 
> ---
> diff -urN af_packet.c.orig af_packet.c
> --- af_packet.c.orig    2009-12-26 12:34:15.000000000 +0100
> +++ af_packet.c 2009-12-28 14:31:14.000000000 +0100
> @@ -57,6 +57,8 @@
>  #include <linux/inet.h>
>  #include <linux/netdevice.h>
>  #include <linux/if_packet.h>
> +#include <linux/if_vlan.h>
> +#include "../8021q/vlan.h"

This is ugly, please move is_vlan_dev() in if_vlan.h instead ?

>  #include <linux/wireless.h>
>  #include <linux/kernel.h>
>  #include <linux/kmod.h>
> @@ -680,8 +682,28 @@
>                 if (sk->sk_type != SOCK_DGRAM)
>                         skb_push(skb, skb->data - skb_mac_header(skb));
>                 else if (skb->pkt_type == PACKET_OUTGOING) {
> -                       /* Special case: outgoing packets have ll header at head */
> -                       skb_pull(skb, skb_network_offset(skb));
> +                       /* Special case: outgoing packets have ll header at head
> +                        * but we must leave 802.1Q encapsulation etc. (only for non HW vlan accelerated)
> +                        * encasulation len = actual header_len minus hard_header_len
> +                        * packet outgoing from vlan1@eth0 on eth0 have skb_network_offset=18, hard_header_len=14
> +                        */
> +                       int hard_header_len;
> +                       struct net_device *pdev;
> +                       hard_header_len = dev->hard_header_len;
> +                       pdev = dev;
> +                       /* if dev is vlan device, hard_header_len contains 802.1Q encap, subtract it, recursively
> +                        * ie. vlan3@vlan2@vlan1@eth0
> +                        */
> +                       while (is_vlan_dev(pdev)) {
> +                               struct net_device *real_dev = vlan_dev_info(pdev)->real_dev;
A new line after variable definition is welcomed.
> +                               hard_header_len -= pdev->hard_header_len - real_dev->hard_header_len;
> +                               pdev = real_dev;
> +                       }
> +
> +                       skb_pull(skb, skb_network_offset(skb) -
> +                            (skb_network_offset(skb) - hard_header_len>0 ? skb_network_offset(skb) - hard_header_len : 0));
>                 }
>         }

I find this logic a bit hard to read, you might use this instead :

if (skb_network_offset(skb) > hard_header_len)
	skb_pull(skb, hard_header_len);
else
	skb_pull(skb, skb_network_offset(skb));




Also, could you please read Documentation/SubmittingPatches and
Documentation/email-clients.txt

Your patch has at least two problems :

1) It is not an unified diff
2) It has garbled tabulations (they are replaced by spaces)

It is not applicable as is.

Thank you