bnxt_en: Enable XPS by default on driver load

Enable XPS on default during NIC open. The choice of
Tx queue is based on the CPU executing the thread that
submits the Tx request. The pool of Tx queues will be
spread evenly across both device-attached NUMA nodes(local)
and remote NUMA nodes.
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
Link: https://lore.kernel.org/r/20240402093753.331120-3-pavan.chebbi@broadcom.comSigned-off-by: Jakub Kicinski <kuba@kernel.org>

drivers/net/ethernet/broadcom/bnxt/bnxt.c

@@ -11804,6 +11804,46 @@ static void bnxt_cfg_usr_fltrs(struct bnxt *bp)
 		bnxt_cfg_one_usr_fltr(bp, usr_fltr);
 }
 
+static int bnxt_set_xps_mapping(struct bnxt *bp)
+{
+	int numa_node = dev_to_node(&bp->pdev->dev);
+	unsigned int q_idx, map_idx, cpu, i;
+	const struct cpumask *cpu_mask_ptr;
+	int nr_cpus = num_online_cpus();
+	cpumask_t *q_map;
+	int rc = 0;
+
+	q_map = kcalloc(bp->tx_nr_rings_per_tc, sizeof(*q_map), GFP_KERNEL);
+	if (!q_map)
+		return -ENOMEM;
+
+	/* Create CPU mask for all TX queues across MQPRIO traffic classes.
+	 * Each TC has the same number of TX queues. The nth TX queue for each
+	 * TC will have the same CPU mask.
+	 */
+	for (i = 0; i < nr_cpus; i++) {
+		map_idx = i % bp->tx_nr_rings_per_tc;
+		cpu = cpumask_local_spread(i, numa_node);
+		cpu_mask_ptr = get_cpu_mask(cpu);
+		cpumask_or(&q_map[map_idx], &q_map[map_idx], cpu_mask_ptr);
+	}
+
+	/* Register CPU mask for each TX queue except the ones marked for XDP */
+	for (q_idx = 0; q_idx < bp->dev->real_num_tx_queues; q_idx++) {
+		map_idx = q_idx % bp->tx_nr_rings_per_tc;
+		rc = netif_set_xps_queue(bp->dev, &q_map[map_idx], q_idx);
+		if (rc) {
+			netdev_warn(bp->dev, "Error setting XPS for q:%d\n",
+				    q_idx);
+			break;
+		}
+	}
+
+	kfree(q_map);
+
+	return rc;
+}
+
 static int __bnxt_open_nic(struct bnxt *bp, bool irq_re_init, bool link_re_init)
 {
 	int rc = 0;
@@ -11866,8 +11906,12 @@ static int __bnxt_open_nic(struct bnxt *bp, bool irq_re_init, bool link_re_init)
 		}
 	}
 
-	if (irq_re_init)
+	if (irq_re_init) {
 		udp_tunnel_nic_reset_ntf(bp->dev);
+		rc = bnxt_set_xps_mapping(bp);
+		if (rc)
+			netdev_warn(bp->dev, "failed to set xps mapping\n");
+	}
 
 	if (bp->tx_nr_rings_xdp < num_possible_cpus()) {
 		if (!static_key_enabled(&bnxt_xdp_locking_key))