On 10/03/17 02:54, David Ahern wrote:

On 3/7/17 5:46 PM, Robert Shearman wrote:

quoted

@@ -78,6 +70,29 @@ static int mpls_xmit(struct sk_buff *skb)

 	tun_encap_info = mpls_lwtunnel_encap(dst->lwtstate);

+	/* Obtain the ttl */
+	if (dst->ops->family == AF_INET) {
+		if (tun_encap_info->ttl_propagate == MPLS_TTL_PROP_DISABLED)
+			ttl = tun_encap_info->default_ttl;
+		else if (tun_encap_info->ttl_propagate == MPLS_TTL_PROP_DEFAULT &&
+			 !net->mpls.ip_ttl_propagate)
+			ttl = net->mpls.default_ttl;
+		else
+			ttl = ip_hdr(skb)->ttl;

After staring at that for a while, an explanation above this if {} else
{} section on the ttl selection will be very helpful.

Ok, would a comment like the following improve things sufficiently?

	/* Obtain the ttl using the following set of rules.
	 *
	 * LWT ttl propagation setting:
	 *  - disabled => use default TTL value from LWT
	 *  - enabled  => use TTL value from IPv4/IPv6 header
	 *  - default  =>
	 *   Global ttl propagation setting:
	 *    - disabled => use default TTL value from global setting
	 *    - enabled => use TTL value from IPv4/IPv6 header
	 */

Thanks,
Rob