netdev - [PATCH net-next 2/2] net: airoha: Simplify Tx napi logic

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20241029-airoha-en7581-tx-napi-work-v1-2-96ad1686b946@kernel.org>
Date: Tue, 29 Oct 2024 13:17:10 +0100
From: Lorenzo Bianconi <lorenzo@...nel.org>
To: Felix Fietkau <nbd@....name>, Sean Wang <sean.wang@...iatek.com>, 
 Mark Lee <Mark-MC.Lee@...iatek.com>, Andrew Lunn <andrew+netdev@...n.ch>, 
 "David S. Miller" <davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>, 
 Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>, 
 Matthias Brugger <matthias.bgg@...il.com>, 
 AngeloGioacchino Del Regno <angelogioacchino.delregno@...labora.com>
Cc: linux-arm-kernel@...ts.infradead.org, 
 linux-mediatek@...ts.infradead.org, netdev@...r.kernel.org, 
 Lorenzo Bianconi <lorenzo@...nel.org>
Subject: [PATCH net-next 2/2] net: airoha: Simplify Tx napi logic

Simplify Tx napi logic relying just on the packet index provided by
completion queue indicating the completed packet that can be removed
from the Tx DMA ring.
This is a preliminary patch to add Qdisc offload for airoha_eth driver.

Signed-off-by: Lorenzo Bianconi <lorenzo@...nel.org>
---
 drivers/net/ethernet/mediatek/airoha_eth.c | 73 +++++++++++++++++-------------
 1 file changed, 41 insertions(+), 32 deletions(-)

diff --git a/drivers/net/ethernet/mediatek/airoha_eth.c b/drivers/net/ethernet/mediatek/airoha_eth.c
index 6cd8901ed38f0640a8a8f72174c120668b364045..6c683a12d5aa52dd9d966df123509075a989c0b3 100644
--- a/drivers/net/ethernet/mediatek/airoha_eth.c
+++ b/drivers/net/ethernet/mediatek/airoha_eth.c
@@ -1670,8 +1670,12 @@ static int airoha_qdma_tx_napi_poll(struct napi_struct *napi, int budget)
 	irq_queued = FIELD_GET(IRQ_ENTRY_LEN_MASK, status);
 
 	while (irq_queued > 0 && done < budget) {
-		u32 qid, last, val = irq_q->q[head];
+		u32 qid, val = irq_q->q[head];
+		struct airoha_qdma_desc *desc;
+		struct airoha_queue_entry *e;
 		struct airoha_queue *q;
+		u32 index, desc_ctrl;
+		struct sk_buff *skb;
 
 		if (val == 0xff)
 			break;
@@ -1681,9 +1685,7 @@ static int airoha_qdma_tx_napi_poll(struct napi_struct *napi, int budget)
 		irq_queued--;
 		done++;
 
-		last = FIELD_GET(IRQ_DESC_IDX_MASK, val);
 		qid = FIELD_GET(IRQ_RING_IDX_MASK, val);
-
 		if (qid >= ARRAY_SIZE(qdma->q_tx))
 			continue;
 
@@ -1691,46 +1693,53 @@ static int airoha_qdma_tx_napi_poll(struct napi_struct *napi, int budget)
 		if (!q->ndesc)
 			continue;
 
+		index = FIELD_GET(IRQ_DESC_IDX_MASK, val);
+		if (index >= q->ndesc)
+			continue;
+
 		spin_lock_bh(&q->lock);
 
-		while (q->queued > 0) {
-			struct airoha_qdma_desc *desc = &q->desc[q->tail];
-			struct airoha_queue_entry *e = &q->entry[q->tail];
-			u32 desc_ctrl = le32_to_cpu(desc->ctrl);
-			struct sk_buff *skb = e->skb;
-			u16 index = q->tail;
+		if (!q->queued)
+			goto unlock;
 
-			if (!(desc_ctrl & QDMA_DESC_DONE_MASK) &&
-			    !(desc_ctrl & QDMA_DESC_DROP_MASK))
-				break;
+		desc = &q->desc[index];
+		desc_ctrl = le32_to_cpu(desc->ctrl);
 
-			q->tail = (q->tail + 1) % q->ndesc;
-			q->queued--;
+		if (!(desc_ctrl & QDMA_DESC_DONE_MASK) &&
+		    !(desc_ctrl & QDMA_DESC_DROP_MASK))
+			goto unlock;
 
-			dma_unmap_single(eth->dev, e->dma_addr, e->dma_len,
-					 DMA_TO_DEVICE);
+		e = &q->entry[index];
+		skb = e->skb;
 
-			WRITE_ONCE(desc->msg0, 0);
-			WRITE_ONCE(desc->msg1, 0);
+		dma_unmap_single(eth->dev, e->dma_addr, e->dma_len,
+				 DMA_TO_DEVICE);
+		memset(e, 0, sizeof(*e));
+		WRITE_ONCE(desc->msg0, 0);
+		WRITE_ONCE(desc->msg1, 0);
+		q->queued--;
 
-			if (skb) {
-				u16 queue = skb_get_queue_mapping(skb);
-				struct netdev_queue *txq;
+		/* completion ring can report out-of-order indexes if hw QoS
+		 * is enabled and packets with different priority are queued
+		 * to same DMA ring. Take into account possible out-of-order
+		 * reports incrementing DMA ring tail pointer
+		 */
+		while (q->tail != q->head && !q->entry[q->tail].dma_addr)
+			q->tail = (q->tail + 1) % q->ndesc;
 
-				txq = netdev_get_tx_queue(skb->dev, queue);
-				netdev_tx_completed_queue(txq, 1, skb->len);
-				if (netif_tx_queue_stopped(txq) &&
-				    q->ndesc - q->queued >= q->free_thr)
-					netif_tx_wake_queue(txq);
+		if (skb) {
+			u16 queue = skb_get_queue_mapping(skb);
+			struct netdev_queue *txq;
 
-				dev_kfree_skb_any(skb);
-				e->skb = NULL;
-			}
+			txq = netdev_get_tx_queue(skb->dev, queue);
+			netdev_tx_completed_queue(txq, 1, skb->len);
+			if (netif_tx_queue_stopped(txq) &&
+			    q->ndesc - q->queued >= q->free_thr)
+				netif_tx_wake_queue(txq);
 
-			if (index == last)
-				break;
+			dev_kfree_skb_any(skb);
 		}
-
+unlock:
 		spin_unlock_bh(&q->lock);
 	}
 

-- 
2.47.0