igc: switch to napi_build_skb()
authorAlexander Lobakin <alexandr.lobakin@intel.com>
Tue, 23 Nov 2021 17:18:38 +0000 (18:18 +0100)
committerTony Nguyen <anthony.l.nguyen@intel.com>
Tue, 28 Dec 2021 17:42:33 +0000 (09:42 -0800)
napi_build_skb() reuses per-cpu NAPI skbuff_head cache in order
to save some cycles on freeing/allocating skbuff_heads on every
new Rx or completed Tx.
igc driver runs Tx completion polling cycle right before the Rx
one and uses napi_consume_skb() to feed the cache with skbuff_heads
of completed entries, so it's never empty and always warm at that
moment. Switch to the napi_build_skb() to relax mm pressure on
heavy Rx.

Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Nechama Kraus <nechamax.kraus@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
drivers/net/ethernet/intel/igc/igc_main.c

index 142c57b..8af2675 100644 (file)
@@ -1729,7 +1729,7 @@ static struct sk_buff *igc_build_skb(struct igc_ring *rx_ring,
        net_prefetch(xdp->data_meta);
 
        /* build an skb around the page buffer */
-       skb = build_skb(xdp->data_hard_start, truesize);
+       skb = napi_build_skb(xdp->data_hard_start, truesize);
        if (unlikely(!skb))
                return NULL;