Again, just some quick comments after a glimpse.
On Sat, Jul 01, 2017 at 07:08:42PM +0800, Jiayu Hu wrote:
> + for (i = 0; i < nb_pkts; i++) {
> + if (RTE_ETH_IS_IPV4_HDR(pkts[i]->packet_type) &&
> + (pkts[i]->packet_type & RTE_PTYPE_L4_TCP)) {
> + ret = gro_tcp4_reassemble(pkts[i],
> + &tcp_tbl,
> + param->max_packet_size,
> + current_time);
> + if (ret > 0)
> + /* merge successfully */
> + nb_after_gro--;
> + else if (ret < 0)
> + unprocess_pkts[unprocess_num++] =
> + pkts[i];
Even it's just one statement, if the statement is spawned more than
one line, including the comment, the {} should be used.
Section 1.6.2. Control Statements and Loops of:
http://dpdk.org/doc/guides/contributing/coding_style.html
> + } else
> + unprocess_pkts[unprocess_num++] =
> + pkts[i];
Besides, why breaking it to two lines, judging that it can be fit into
one line smaller than 80 chars.
> + }
> +
> + /* re-arrange GROed packets */
> + if (nb_after_gro < nb_pkts) {
> + i = gro_tcp4_tbl_timeout_flush(&tcp_tbl, 0,
> + pkts, nb_pkts);
> + if (unprocess_num > 0)
> + memcpy(&pkts[i], unprocess_pkts,
> + sizeof(struct rte_mbuf *) *
> + unprocess_num);
Ditto.
> +void *gro_tcp4_tbl_create(uint16_t socket_id,
> + uint16_t max_flow_num,
> + uint16_t max_item_per_flow)
> +{
> + size_t size;
> + uint32_t entries_num;
> + struct gro_tcp4_tbl *tbl;
> +
> + entries_num = max_flow_num * max_item_per_flow;
> + entries_num = entries_num > GRO_TCP4_TBL_MAX_ITEM_NUM ?
> + GRO_TCP4_TBL_MAX_ITEM_NUM : entries_num;
> +
> + if (entries_num == 0)
> + return NULL;
> +
> + tbl = (struct gro_tcp4_tbl *)rte_zmalloc_socket(
> + __func__,
> + sizeof(struct gro_tcp4_tbl),
> + RTE_CACHE_LINE_SIZE,
> + socket_id);
Again, the cast (from void *) is unnessary and should be dropped.
> + memcpy(&(tbl->keys[key_idx].key),
> + &key, sizeof(struct tcp4_key));
Again, I believe they two can be fit into one single line.
--yliu
