i40e: optimize AF_XDP Tx completion path

Improve the performance of the AF_XDP zero-copy Tx completion
path. When there are no XDP buffers being sent using XDP_TX or
XDP_REDIRECT, we do not have go through the SW ring to clean up any
entries since the AF_XDP path does not use these. In these cases, just
fast forward the next-to-use counter and skip going through the SW
ring. The limit on the maximum number of entries to complete is also
removed since the algorithm is now O(1). To simplify the code path, the
maximum number of entries to complete for the XDP path is therefore
also increased from 256 to 512 (the default number of Tx HW
descriptors). This should be fine since the completion in the XDP path
is faster than in the SKB path that has 256 as the maximum number.

This patch provides around 4% throughput improvement for the l2fwd
application in xdpsock on my machine.

Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

1 files changed, 23 insertions, 20 deletions

diff --git a/drivers/net/ethernet/intel/i40e/i40e_xsk.c b/drivers/net/ethernet/intel/i40e/i40e_xsk.c
index 8e489ed54138..744861dc7acd 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_xsk.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_xsk.c
@@ -378,6 +378,7 @@ int i40e_clean_rx_irq_zc(struct i40e_ring *rx_ring, int budget)
  **/
 static bool i40e_xmit_zc(struct i40e_ring *xdp_ring, unsigned int budget)
 {
+	unsigned int sent_frames = 0, total_bytes = 0;
 	struct i40e_tx_desc *tx_desc = NULL;
 	struct i40e_tx_buffer *tx_bi;
 	bool work_done = true;
@@ -408,6 +409,9 @@ static bool i40e_xmit_zc(struct i40e_ring *xdp_ring, unsigned int budget)
 				   | I40E_TX_DESC_CMD_EOP,
 				   0, desc.len, 0);
 
+		sent_frames++;
+		total_bytes += tx_bi->bytecount;
+
 		xdp_ring->next_to_use++;
 		if (xdp_ring->next_to_use == xdp_ring->count)
 			xdp_ring->next_to_use = 0;
@@ -420,6 +424,7 @@ static bool i40e_xmit_zc(struct i40e_ring *xdp_ring, unsigned int budget)
 		i40e_xdp_ring_update_tail(xdp_ring);
 
 		xsk_umem_consume_tx_done(xdp_ring->xsk_umem);
+		i40e_update_tx_stats(xdp_ring, sent_frames, total_bytes);
 	}
 
 	return !!budget && work_done;
@@ -434,6 +439,7 @@ static void i40e_clean_xdp_tx_buffer(struct i40e_ring *tx_ring,
 				     struct i40e_tx_buffer *tx_bi)
 {
 	xdp_return_frame(tx_bi->xdpf);
+	tx_ring->xdp_tx_active--;
 	dma_unmap_single(tx_ring->dev,
 			 dma_unmap_addr(tx_bi, dma),
 			 dma_unmap_len(tx_bi, len), DMA_TO_DEVICE);
@@ -447,27 +453,25 @@ static void i40e_clean_xdp_tx_buffer(struct i40e_ring *tx_ring,
  *
  * Returns true if cleanup/tranmission is done.
  **/
-bool i40e_clean_xdp_tx_irq(struct i40e_vsi *vsi,
-			   struct i40e_ring *tx_ring, int napi_budget)
+bool i40e_clean_xdp_tx_irq(struct i40e_vsi *vsi, struct i40e_ring *tx_ring)
 {
-	unsigned int ntc, total_bytes = 0, budget = vsi->work_limit;
-	u32 i, completed_frames, frames_ready, xsk_frames = 0;
+	unsigned int ntc, budget = vsi->work_limit;
 	struct xdp_umem *umem = tx_ring->xsk_umem;
+	u32 i, completed_frames, xsk_frames = 0;
 	u32 head_idx = i40e_get_head(tx_ring);
-	bool work_done = true, xmit_done;
 	struct i40e_tx_buffer *tx_bi;
+	bool xmit_done;
 
 	if (head_idx < tx_ring->next_to_clean)
 		head_idx += tx_ring->count;
-	frames_ready = head_idx - tx_ring->next_to_clean;
+	completed_frames = head_idx - tx_ring->next_to_clean;
 
-	if (frames_ready == 0) {
+	if (completed_frames == 0)
 		goto out_xmit;
-	} else if (frames_ready > budget) {
-		completed_frames = budget;
-		work_done = false;
-	} else {
-		completed_frames = frames_ready;
+
+	if (likely(!tx_ring->xdp_tx_active)) {
+		xsk_frames = completed_frames;
+		goto skip;
 	}
 
 	ntc = tx_ring->next_to_clean;
@@ -475,18 +479,18 @@ bool i40e_clean_xdp_tx_irq(struct i40e_vsi *vsi,
 	for (i = 0; i < completed_frames; i++) {
 		tx_bi = &tx_ring->tx_bi[ntc];
 
-		if (tx_bi->xdpf)
+		if (tx_bi->xdpf) {
 			i40e_clean_xdp_tx_buffer(tx_ring, tx_bi);
-		else
+			tx_bi->xdpf = NULL;
+		} else {
 			xsk_frames++;
-
-		tx_bi->xdpf = NULL;
-		total_bytes += tx_bi->bytecount;
+		}
 
 		if (++ntc >= tx_ring->count)
 			ntc = 0;
 	}
 
+skip:
 	tx_ring->next_to_clean += completed_frames;
 	if (unlikely(tx_ring->next_to_clean >= tx_ring->count))
 		tx_ring->next_to_clean -= tx_ring->count;
@@ -494,8 +498,7 @@ bool i40e_clean_xdp_tx_irq(struct i40e_vsi *vsi,
 	if (xsk_frames)
 		xsk_umem_complete_tx(umem, xsk_frames);
 
-	i40e_arm_wb(tx_ring, vsi, budget);
-	i40e_update_tx_stats(tx_ring, completed_frames, total_bytes);
+	i40e_arm_wb(tx_ring, vsi, completed_frames);
 
 out_xmit:
 	if (xsk_umem_uses_need_wakeup(tx_ring->xsk_umem))
@@ -503,7 +506,7 @@ out_xmit:
 
 	xmit_done = i40e_xmit_zc(tx_ring, budget);
 
-	return work_done && xmit_done;
+	return xmit_done;
 }
 
 /**