[PATCH net] net: airoha: Update tx cpu dma ring idx at the end of xmit loop

[Lorenzo Bianconi]

Move the tx cpu dma ring index update out of transmit loop of
airoha_dev_xmit routine in order to not start transmitting the packet
before it is fully DMA mapped (e.g. fragmented skbs).

Fixes: 23020f049327 ("net: airoha: Introduce ethernet support for EN7581 SoC")
Reported-by: Felix Fietkau <nbd at nbd.name>
Signed-off-by: Lorenzo Bianconi <lorenzo at kernel.org>
---
 drivers/net/ethernet/mediatek/airoha_eth.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/mediatek/airoha_eth.c b/drivers/net/ethernet/mediatek/airoha_eth.c
index 930f180688e5caf1175481a43d243812bc07be07..2c26eb185283725ff50803437dac5143d7884b5a 100644
--- a/drivers/net/ethernet/mediatek/airoha_eth.c
+++ b/drivers/net/ethernet/mediatek/airoha_eth.c
@@ -2471,10 +2471,6 @@ static netdev_tx_t airoha_dev_xmit(struct sk_buff *skb,
 		e->dma_addr = addr;
 		e->dma_len = len;
 
-		airoha_qdma_rmw(qdma, REG_TX_CPU_IDX(qid),
-				TX_RING_CPU_IDX_MASK,
-				FIELD_PREP(TX_RING_CPU_IDX_MASK, index));
-
 		data = skb_frag_address(frag);
 		len = skb_frag_size(frag);
 	}
@@ -2483,6 +2479,11 @@ static netdev_tx_t airoha_dev_xmit(struct sk_buff *skb,
 	q->queued += i;
 
 	skb_tx_timestamp(skb);
+	if (!netdev_xmit_more())
+		airoha_qdma_rmw(qdma, REG_TX_CPU_IDX(qid),
+				TX_RING_CPU_IDX_MASK,
+				FIELD_PREP(TX_RING_CPU_IDX_MASK, q->head));
+
 	if (q->ndesc - q->queued < q->free_thr)
 		netif_tx_stop_queue(txq);
 

---
base-commit: 096c0fa42afa92b6ffa4e441c4c72a2f805c5a88
change-id: 20241004-airoha-eth-7581-mapping-fix-919481cbb340

Best regards,
-- 
Lorenzo Bianconi <lorenzo at kernel.org>