[PATCH 5/5] net: hns3: streaming dma buffer sync between cpu and device

From: Barry Song <song.bao.hua@hisilicon.com>
To: <davem@davemloft.net>, <kuba@kernel.org>
Cc: <yisen.zhuang@huawei.com>, <salil.mehta@huawei.com>,
	<netdev@vger.kernel.org>, <linyunsheng@huawei.com>,
	<linux-kernel@vger.kernel.org>, <linuxarm@huawei.com>,
	Barry Song <song.bao.hua@hisilicon.com>
Subject: [PATCH 5/5] net: hns3: streaming dma buffer sync between cpu and device
Date: Thu, 18 Jun 2020 13:02:11 +1200	[thread overview]
Message-ID: <20200618010211.75840-6-song.bao.hua@hisilicon.com> (raw)
In-Reply-To: <20200618010211.75840-1-song.bao.hua@hisilicon.com>

Right now they are empty functions for our SoC since hardware can keep
cache coherent, but it is still good to align with streaming DMA APIs
as device drivers should not make an assumption of SoC.

Reviewed-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Barry Song <song.bao.hua@hisilicon.com>
---
 .../net/ethernet/hisilicon/hns3/hns3_enet.c    | 18 +++++++++++++++++-
 1 file changed, 17 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
index 1330820152fa..b319a766889f 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
@@ -2473,6 +2473,11 @@ static void hns3_reuse_buffer(struct hns3_enet_ring *ring, int i)
 	ring->desc[i].addr = cpu_to_le64(ring->desc_cb[i].dma +
 					 ring->desc_cb[i].page_offset);
 	ring->desc[i].rx.bd_base_info = 0;
+
+	dma_sync_single_for_device(ring_to_dev(ring),
+			ring->desc_cb[i].dma + ring->desc_cb[i].page_offset,
+			hns3_buf_size(ring),
+			DMA_FROM_DEVICE);
 }
 
 static void hns3_nic_reclaim_desc(struct hns3_enet_ring *ring, int head,
@@ -2918,6 +2923,11 @@ static int hns3_add_frag(struct hns3_enet_ring *ring)
 			skb = ring->tail_skb;
 		}
 
+		dma_sync_single_for_cpu(ring_to_dev(ring),
+				desc_cb->dma + desc_cb->page_offset,
+				hns3_buf_size(ring),
+				DMA_FROM_DEVICE);
+
 		hns3_nic_reuse_page(skb, ring->frag_num++, ring, 0, desc_cb);
 		trace_hns3_rx_desc(ring);
 		ring_ptr_move_fw(ring, next_to_clean);
@@ -3069,9 +3079,15 @@ static int hns3_handle_rx_bd(struct hns3_enet_ring *ring)
 	if (unlikely(!(bd_base_info & BIT(HNS3_RXD_VLD_B))))
 		return -ENXIO;
 
-	if (!skb)
+	if (!skb) {
 		ring->va = desc_cb->buf + desc_cb->page_offset;
 
+		dma_sync_single_for_cpu(ring_to_dev(ring),
+				desc_cb->dma + desc_cb->page_offset,
+				hns3_buf_size(ring),
+				DMA_FROM_DEVICE);
+	}
+
 	/* Prefetch first cache line of first page
 	 * Idea is to cache few bytes of the header of the packet. Our L1 Cache
 	 * line size is 64B so need to prefetch twice to make it 128B. But in
-- 
2.23.0


