Re: [PATCH net v2 1/1] ip6_tunnel: snapshot encap in xmit

[Ido Schimmel <idosch@nvidia.com>]

On Sat, Aug 08, 2026 at 04:40:49PM +0800, Ren Wei wrote:
> @@ -1202,6 +1204,15 @@ int ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev, __u8 dsfield,
>  				     t->parms.name);
>  		goto tx_err_dst_release;
>  	}
> +
> +	/* Can tear, but hlen and build_header() use the same snapshot. */
> +	ipencap = data_race(t->encap);
> +	encap_hlen = ip6_encap_hlen(&ipencap);
> +	if (unlikely(encap_hlen < 0))
> +		goto tx_err_dst_release;
> +	psh_hlen = sizeof(struct ipv6hdr) + encap_hlen;
> +	max_headroom = psh_hlen;
> +
>  	mtu = dst6_mtu(dst) - eth_hlen - psh_hlen - t->tun_hlen;
>  	if (encap_limit >= 0) {
>  		max_headroom += 8;

[...]

> @@ -1272,10 +1283,10 @@ int ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev, __u8 dsfield,
>  	 * needed_headroom if necessary.
>  	 */
>  	max_headroom = LL_RESERVED_SPACE(tdev) + sizeof(struct ipv6hdr)
> -			+ dst->header_len + t->hlen;
> +			+ dst->header_len + t->tun_hlen + encap_hlen;

Sashiko is correct that 'hlen = tun_hlen + encap_hlen' doesn't hold for
ip6erspan which accounts for the ERSPAN header length separately (not
part of tun_hlen).

There is no need to touch this line since it's irrelevant to the fix.
The needed headroom was already calculated correctly earlier
('psh_hlen') and guaranteed to be available in the skb by
skb_realloc_headroom(), when needed.

>  	ip_tunnel_adj_headroom(dev, max_headroom);
>  
> -	err = ip6_tnl_encap(skb, t, &proto, fl6);
> +	err = ip6_tnl_encap(skb, &ipencap, &proto, fl6);
>  	if (err)
>  		return err;
>  
> -- 
> 2.34.1