netdev - RE: replace hooks in __netif_receive

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <0199E0D51A61344794750DC57738F58E70B17D377A@GVW1118EXC.americas.hpqcorp.net>
Date:	Tue, 1 Jun 2010 16:13:51 +0000
From:	"Fischer, Anna" <anna.fischer@...com>
To:	Stephen Hemminger <shemminger@...tta.com>,
	Jiri Pirko <jpirko@...hat.com>,
	"davem@...emloft.net" <davem@...emloft.net>
CC:	"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
	"kaber@...sh.net" <kaber@...sh.net>,
	"eric.dumazet@...il.com" <eric.dumazet@...il.com>
Subject: RE: replace hooks in __netif_receive_skb (v4)

> Subject: net: replace hooks in __netif_receive_skb (v4)
> 
> 
> From: Jiri Pirko <jpirko@...hat.com>
> 
> What this patch does is it removes two receive frame hooks (for bridge
> and for
> macvlan) from __netif_receive_skb. These are replaced them with a
> single
> hook for both. It only supports one hook per device because it makes no
> sense to do bridging and macvlan on the same device.
> 
> Then a network driver (of virtual netdev like macvlan or bridge) can
> register
> an rx_handler for needed net device.


I think the idea of this is really good, and it has been long required to get rid of the bridging hook and the "hack" for macvlan to get into the network stack. 

However, I wonder, if this is to be used as a generic interface, then why the restriction of only having a single hook per device? Yes, it makes sense to do this for the bridge/macvlan combination, but in general there could be other cases where you would want to allow multiple receivers per device. 


> @@ -2792,6 +2778,7 @@ EXPORT_SYMBOL(__skb_bond_should_drop);
>  static int __netif_receive_skb(struct sk_buff *skb)
>  {
>  	struct packet_type *ptype, *pt_prev;
> +	rx_callback_func_t *rh;
>  	struct net_device *orig_dev;
>  	struct net_device *master;
>  	struct net_device *null_or_orig;
> @@ -2855,12 +2842,16 @@ static int __netif_receive_skb(struct sk
>  ncls:
>  #endif
> 
> -	skb = handle_bridge(skb, &pt_prev, &ret, orig_dev);
> -	if (!skb)
> -		goto out;
> -	skb = handle_macvlan(skb, &pt_prev, &ret, orig_dev);
> -	if (!skb)
> -		goto out;
> +	/* Handle special case of bridge or macvlan */
> +	if ((rh = rcu_dereference(skb->dev->rx_handler)) != NULL) {
> +		if (pt_prev) {
> +			ret = deliver_skb(skb, pt_prev, orig_dev);

What happens with 'ret' here? It is completely ignored, isn't it? Because you assume that there is only a single receiver per device? I think it would be much better to have return codes that indicate whether a packet has been consumed by a receive handler, or whether it is supposed to be processed further. The same actually applies for the packet handlers that are processed a bit further down in netif_receive_skb() - the return code is ignored and so a handler has no control over further processing of the packet.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html