DPDK patches and discussions
 help / color / mirror / Atom feed
* [PATCH 14/21] net/ena/base: remove redundant zeroing of Tx desc
@ 2025-10-15  7:14 Shai Brandes
  0 siblings, 0 replies; only message in thread
From: Shai Brandes @ 2025-10-15  7:14 UTC (permalink / raw)
  To: stephen; +Cc: dev, Shai Brandes

For LLQ, tx descriptors are stored in the bounce buffer,
which is already being zeroed out upon every call to
ena_com_get_next_bounce_buffer.

This patch utilizes the existing zero-out operation of the
bounce buffer, eliminating redundant zeroing of tx
descriptors in LLQ case.

Signed-off-by: Shai Brandes <shaibran@amazon.com>
Reviewed-by: Amit Bernstein <amitbern@amazon.com>
Reviewed-by: Yosef Raisman <yraisman@amazon.com>
---
 drivers/net/ena/base/ena_eth_com.c | 23 ++++++++++++++---------
 1 file changed, 14 insertions(+), 9 deletions(-)

diff --git a/drivers/net/ena/base/ena_eth_com.c b/drivers/net/ena/base/ena_eth_com.c
index 7d982854c0..c1e3c55f82 100644
--- a/drivers/net/ena/base/ena_eth_com.c
+++ b/drivers/net/ena/base/ena_eth_com.c
@@ -234,12 +234,21 @@ static int ena_com_close_bounce_buffer(struct ena_com_io_sq *io_sq)
 	return ENA_COM_OK;
 }
 
-static void *get_sq_desc(struct ena_com_io_sq *io_sq)
+/* There is an assumption in function get_tx_sq_desc() that both descriptors have the same size */
+_Static_assert(sizeof(struct ena_eth_io_tx_meta_desc) == (sizeof(struct ena_eth_io_tx_desc)),
+	       "ena_eth_io_tx_meta_desc should be the same size as ena_eth_io_tx_desc");
+static void *get_tx_sq_desc(struct ena_com_io_sq *io_sq)
 {
+	void *tx_desc;
+
 	if (io_sq->mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV)
 		return get_sq_desc_llq(io_sq);
 
-	return get_sq_desc_regular_queue(io_sq);
+	tx_desc = get_sq_desc_regular_queue(io_sq);
+
+	memset(tx_desc, 0x0, sizeof(struct ena_eth_io_tx_desc));
+
+	return tx_desc;
 }
 
 static int ena_com_sq_update_llq_tail(struct ena_com_io_sq *io_sq)
@@ -372,12 +381,10 @@ static int ena_com_create_meta(struct ena_com_io_sq *io_sq,
 {
 	struct ena_eth_io_tx_meta_desc *meta_desc = NULL;
 
-	meta_desc = get_sq_desc(io_sq);
+	meta_desc = get_tx_sq_desc(io_sq);
 	if (unlikely(!meta_desc))
 		return ENA_COM_FAULT;
 
-	memset(meta_desc, 0x0, sizeof(struct ena_eth_io_tx_meta_desc));
-
 	meta_desc->len_ctrl |= ENA_ETH_IO_TX_META_DESC_META_DESC_MASK;
 
 	meta_desc->len_ctrl |= ENA_ETH_IO_TX_META_DESC_EXT_VALID_MASK;
@@ -535,7 +542,7 @@ int ena_com_prepare_tx(struct ena_com_io_sq *io_sq,
 		return rc;
 	}
 
-	desc = get_sq_desc(io_sq);
+	desc = get_tx_sq_desc(io_sq);
 	if (unlikely(!desc))
 		return ENA_COM_FAULT;
 
@@ -597,12 +604,10 @@ int ena_com_prepare_tx(struct ena_com_io_sq *io_sq,
 				return rc;
 			}
 
-			desc = get_sq_desc(io_sq);
+			desc = get_tx_sq_desc(io_sq);
 			if (unlikely(!desc))
 				return ENA_COM_FAULT;
 
-			memset(desc, 0x0, sizeof(struct ena_eth_io_tx_desc));
-
 			desc->len_ctrl |= ENA_FIELD_PREP((u32)io_sq->phase,
 							 ENA_ETH_IO_TX_DESC_PHASE_MASK,
 							 ENA_ETH_IO_TX_DESC_PHASE_SHIFT);
-- 
2.17.1


^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2025-10-15  7:15 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2025-10-15  7:14 [PATCH 14/21] net/ena/base: remove redundant zeroing of Tx desc Shai Brandes

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).