octeontx2-pf: AF_XDP: code clean up

The current API, otx2_xdp_sq_append_pkt, verifies the number of available
descriptors before sending packets to the hardware.

However, for AF_XDP, this check is unnecessary because the batch value
is already determined based on the free descriptors.

This patch introduces a new API, "otx2_xsk_sq_append_pkt" to address this.

Remove the logic for releasing the TX buffers, as it is implicitly handled
by xsk_tx_peek_release_desc_batch

Signed-off-by: Hariprasad Kelam <hkelam@marvell.com>
Link: https://patch.msgid.link/20250420032350.4047706-1-hkelam@marvell.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Author: Hariprasad Kelam

drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h

@@ -1107,6 +1107,8 @@ int otx2_enable_rxvlan(struct otx2_nic *pf, bool enable);
 int otx2_install_rxvlan_offload_flow(struct otx2_nic *pfvf);
 bool otx2_xdp_sq_append_pkt(struct otx2_nic *pfvf, struct xdp_frame *xdpf,
 			    u64 iova, int len, u16 qidx, u16 flags);
+void otx2_xdp_sqe_add_sg(struct otx2_snd_queue *sq, struct xdp_frame *xdpf,
+			 u64 dma_addr, int len, int *offset, u16 flags);
 u16 otx2_get_max_mtu(struct otx2_nic *pfvf);
 int otx2_handle_ntuple_tc_features(struct net_device *netdev,
 				   netdev_features_t features);

drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c

@@ -1410,9 +1410,8 @@ void otx2_free_pending_sqe(struct otx2_nic *pfvf)
 	}
 }
 
-static void otx2_xdp_sqe_add_sg(struct otx2_snd_queue *sq,
-				struct xdp_frame *xdpf,
-				u64 dma_addr, int len, int *offset, u16 flags)
+void otx2_xdp_sqe_add_sg(struct otx2_snd_queue *sq, struct xdp_frame *xdpf,
+			 u64 dma_addr, int len, int *offset, u16 flags)
 {
 	struct nix_sqe_sg_s *sg = NULL;
 	u64 *iova = NULL;

drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.c

@@ -11,6 +11,7 @@
 #include <net/xdp.h>
 
 #include "otx2_common.h"
+#include "otx2_struct.h"
 #include "otx2_xsk.h"
 
 int otx2_xsk_pool_alloc_buf(struct otx2_nic *pfvf, struct otx2_pool *pool,
@@ -196,11 +197,39 @@ void otx2_attach_xsk_buff(struct otx2_nic *pfvf, struct otx2_snd_queue *sq, int
 		sq->xsk_pool = xsk_get_pool_from_qid(pfvf->netdev, qidx);
 }
 
+static void otx2_xsk_sq_append_pkt(struct otx2_nic *pfvf, u64 iova, int len,
+				   u16 qidx)
+{
+	struct nix_sqe_hdr_s *sqe_hdr;
+	struct otx2_snd_queue *sq;
+	int offset;
+
+	sq = &pfvf->qset.sq[qidx];
+	memset(sq->sqe_base + 8, 0, sq->sqe_size - 8);
+
+	sqe_hdr = (struct nix_sqe_hdr_s *)(sq->sqe_base);
+
+	if (!sqe_hdr->total) {
+		sqe_hdr->aura = sq->aura_id;
+		sqe_hdr->df = 1;
+		sqe_hdr->sq = qidx;
+		sqe_hdr->pnc = 1;
+	}
+	sqe_hdr->total = len;
+	sqe_hdr->sqe_id = sq->head;
+
+	offset = sizeof(*sqe_hdr);
+
+	otx2_xdp_sqe_add_sg(sq, NULL, iova, len, &offset, OTX2_AF_XDP_FRAME);
+	sqe_hdr->sizem1 = (offset / 16) - 1;
+	pfvf->hw_ops->sqe_flush(pfvf, sq, offset, qidx);
+}
+
 void otx2_zc_napi_handler(struct otx2_nic *pfvf, struct xsk_buff_pool *pool,
 			  int queue, int budget)
 {
 	struct xdp_desc *xdp_desc = pool->tx_descs;
-	int err, i, work_done = 0, batch;
+	int  i, batch;
 
 	budget = min(budget, otx2_read_free_sqe(pfvf, queue));
 	batch = xsk_tx_peek_release_desc_batch(pool, budget);
@@ -211,15 +240,6 @@ void otx2_zc_napi_handler(struct otx2_nic *pfvf, struct xsk_buff_pool *pool,
 		dma_addr_t dma_addr;
 
 		dma_addr = xsk_buff_raw_get_dma(pool, xdp_desc[i].addr);
-		err = otx2_xdp_sq_append_pkt(pfvf, NULL, dma_addr, xdp_desc[i].len,
-					     queue, OTX2_AF_XDP_FRAME);
-		if (!err) {
-			netdev_err(pfvf->netdev, "AF_XDP: Unable to transfer packet err%d\n", err);
-			break;
-		}
-		work_done++;
+		otx2_xsk_sq_append_pkt(pfvf, dma_addr, xdp_desc[i].len, queue);
 	}
-
-	if (work_done)
-		xsk_tx_release(pool);
 }