2014-10-11 13:55, Jijiang Liu:
> Support VxLAN Tx checksum offload, which include
> - outer L3(IP) checksum offload
> - inner L3(IP) checksum offload
> - inner L4(UDP, TCP and SCTP) checksum offload
[...]
> +
> + /* fields to support tunnelling packet TX offloads */
I know that previous comment is "fields to support TX offloads",
but I'd prefer "for TX offloading of tunnels".
Maybe that "encapsulation" is better than "tunnel".
Just my opinion.
> + union {
> + /**< combined inner l2/l3 lengths as single var */
> + uint16_t inner_l2_l3_len;
> +
> + struct {
> + /**< inner L3 (IP) Header Length. */
> + uint16_t inner_l3_len:9;
> +
> + /**< L2 (MAC) Header Length. */
> + uint16_t inner_l2_len:7;
> + };
> + };
I would like to highlight that you are using 2 bytes in the second cache line
of the mbuf.
It deserves at least a line in the commit log.
Actually I'd prefer a separate patch for mbuf modifications.
Thanks
--
Thomas
