Improves cache efficiency by batching rx skb processing. Small
performance improvement on RX.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Reviewed-by: Shannon Nelson <shannon.nelson@amd.com>
---
 drivers/net/ethernet/lantiq_etop.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/lantiq_etop.c b/drivers/net/ethernet/lantiq_etop.c
index 3c289bfe0a09..94b37c12f3f7 100644
--- a/drivers/net/ethernet/lantiq_etop.c
+++ b/drivers/net/ethernet/lantiq_etop.c
@@ -122,8 +122,7 @@ ltq_etop_alloc_skb(struct ltq_etop_chan *ch)
 	return 0;
 }
 
-static void
-ltq_etop_hw_receive(struct ltq_etop_chan *ch)
+static void ltq_etop_hw_receive(struct ltq_etop_chan *ch, struct list_head *lh)
 {
 	struct ltq_etop_priv *priv = netdev_priv(ch->netdev);
 	struct ltq_dma_desc *desc = &ch->dma.desc_base[ch->dma.desc];
@@ -143,7 +142,7 @@ ltq_etop_hw_receive(struct ltq_etop_chan *ch)
 
 	skb_put(skb, len);
 	skb->protocol = eth_type_trans(skb, ch->netdev);
-	netif_receive_skb(skb);
+	list_add_tail(&skb->list, lh);
 }
 
 static int
@@ -151,6 +150,7 @@ ltq_etop_poll_rx(struct napi_struct *napi, int budget)
 {
 	struct ltq_etop_chan *ch = container_of(napi,
 				struct ltq_etop_chan, napi);
+	LIST_HEAD(rx_list);
 	int work_done = 0;
 
 	while (work_done < budget) {
@@ -158,9 +158,12 @@ ltq_etop_poll_rx(struct napi_struct *napi, int budget)
 
 		if ((desc->ctl & (LTQ_DMA_OWN | LTQ_DMA_C)) != LTQ_DMA_C)
 			break;
-		ltq_etop_hw_receive(ch);
+		ltq_etop_hw_receive(ch, &rx_list);
 		work_done++;
 	}
+
+	netif_receive_skb_list(&rx_list);
+
 	if (work_done < budget) {
 		napi_complete_done(&ch->napi, work_done);
 		ltq_dma_ack_irq(&ch->dma);
-- 
2.46.2

On Tue, Oct 01, 2024 at 11:45:58AM -0700, Rosen Penev wrote:
> Improves cache efficiency by batching rx skb processing. Small
> performance improvement on RX.

Benchmark numbers would be good.

> 
> Signed-off-by: Rosen Penev <rosenp@gmail.com>
> Reviewed-by: Shannon Nelson <shannon.nelson@amd.com>
> ---
>  drivers/net/ethernet/lantiq_etop.c | 11 +++++++----
>  1 file changed, 7 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/net/ethernet/lantiq_etop.c b/drivers/net/ethernet/lantiq_etop.c
> index 3c289bfe0a09..94b37c12f3f7 100644
> --- a/drivers/net/ethernet/lantiq_etop.c
> +++ b/drivers/net/ethernet/lantiq_etop.c
> @@ -122,8 +122,7 @@ ltq_etop_alloc_skb(struct ltq_etop_chan *ch)
>  	return 0;
>  }
>  
> -static void
> -ltq_etop_hw_receive(struct ltq_etop_chan *ch)
> +static void ltq_etop_hw_receive(struct ltq_etop_chan *ch, struct list_head *lh)

Please don't put the return type on the same line. If you look at this
driver, it is the coding style to always have it on a separate
line. You broken the coding style.


    Andrew

---
pw-bot: cr

On Tue, Oct 1, 2024 at 3:40 PM Andrew Lunn <andrew@lunn.ch> wrote:
>
> On Tue, Oct 01, 2024 at 11:45:58AM -0700, Rosen Penev wrote:
> > Improves cache efficiency by batching rx skb processing. Small
> > performance improvement on RX.
>
> Benchmark numbers would be good.
>
> >
> > Signed-off-by: Rosen Penev <rosenp@gmail.com>
> > Reviewed-by: Shannon Nelson <shannon.nelson@amd.com>
> > ---
> >  drivers/net/ethernet/lantiq_etop.c | 11 +++++++----
> >  1 file changed, 7 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/net/ethernet/lantiq_etop.c b/drivers/net/ethernet/lantiq_etop.c
> > index 3c289bfe0a09..94b37c12f3f7 100644
> > --- a/drivers/net/ethernet/lantiq_etop.c
> > +++ b/drivers/net/ethernet/lantiq_etop.c
> > @@ -122,8 +122,7 @@ ltq_etop_alloc_skb(struct ltq_etop_chan *ch)
> >       return 0;
> >  }
> >
> > -static void
> > -ltq_etop_hw_receive(struct ltq_etop_chan *ch)
> > +static void ltq_etop_hw_receive(struct ltq_etop_chan *ch, struct list_head *lh)
>
> Please don't put the return type on the same line. If you look at this
> driver, it is the coding style to always have it on a separate
> line. You broken the coding style.
I'm using git clang-format HEAD~1 on my commits. Something to improve I guess.
>
>
>     Andrew
>
> ---
> pw-bot: cr
>