net: ena: Fix redundant device NUMA node override

[ Upstream commit 2dc8b1e ]

The driver overrides the NUMA node id of the device regardless of
whether it knows its correct value (often setting it to -1 even though
the node id is advertised in 'struct device'). This can lead to
suboptimal configurations.

This patch fixes this behavior and makes the shared memory allocation
functions use the NUMA node id advertised by the underlying device.

Fixes: 1738cd3 ("net: ena: Add a driver for Amazon Elastic Network Adapters (ENA)")
Signed-off-by: Shay Agroskin <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>
Signed-off-by: Sasha Levin <[email protected]>

drivers/net/ethernet/amazon/ena/ena_com.c

@@ -312,7 +312,6 @@ static int ena_com_init_io_sq(struct ena_com_dev *ena_dev,
 			      struct ena_com_io_sq *io_sq)
 {
 	size_t size;
-	int dev_node = 0;
 
 	memset(&io_sq->desc_addr, 0x0, sizeof(io_sq->desc_addr));
 
@@ -325,12 +324,9 @@ static int ena_com_init_io_sq(struct ena_com_dev *ena_dev,
 	size = io_sq->desc_entry_size * io_sq->q_depth;
 
 	if (io_sq->mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_HOST) {
-		dev_node = dev_to_node(ena_dev->dmadev);
-		set_dev_node(ena_dev->dmadev, ctx->numa_node);
 		io_sq->desc_addr.virt_addr =
 			dma_alloc_coherent(ena_dev->dmadev, size, &io_sq->desc_addr.phys_addr,
 					   GFP_KERNEL);
-		set_dev_node(ena_dev->dmadev, dev_node);
 		if (!io_sq->desc_addr.virt_addr) {
 			io_sq->desc_addr.virt_addr =
 				dma_alloc_coherent(ena_dev->dmadev, size,
@@ -354,10 +350,7 @@ static int ena_com_init_io_sq(struct ena_com_dev *ena_dev,
 		size = (size_t)io_sq->bounce_buf_ctrl.buffer_size *
 			io_sq->bounce_buf_ctrl.buffers_num;
 
-		dev_node = dev_to_node(ena_dev->dmadev);
-		set_dev_node(ena_dev->dmadev, ctx->numa_node);
 		io_sq->bounce_buf_ctrl.base_buffer = devm_kzalloc(ena_dev->dmadev, size, GFP_KERNEL);
-		set_dev_node(ena_dev->dmadev, dev_node);
 		if (!io_sq->bounce_buf_ctrl.base_buffer)
 			io_sq->bounce_buf_ctrl.base_buffer =
 				devm_kzalloc(ena_dev->dmadev, size, GFP_KERNEL);
@@ -397,7 +390,6 @@ static int ena_com_init_io_cq(struct ena_com_dev *ena_dev,
 			      struct ena_com_io_cq *io_cq)
 {
 	size_t size;
-	int prev_node = 0;
 
 	memset(&io_cq->cdesc_addr, 0x0, sizeof(io_cq->cdesc_addr));
 
@@ -409,11 +401,8 @@ static int ena_com_init_io_cq(struct ena_com_dev *ena_dev,
 
 	size = io_cq->cdesc_entry_size_in_bytes * io_cq->q_depth;
 
-	prev_node = dev_to_node(ena_dev->dmadev);
-	set_dev_node(ena_dev->dmadev, ctx->numa_node);
 	io_cq->cdesc_addr.virt_addr =
 		dma_alloc_coherent(ena_dev->dmadev, size, &io_cq->cdesc_addr.phys_addr, GFP_KERNEL);
-	set_dev_node(ena_dev->dmadev, prev_node);
 	if (!io_cq->cdesc_addr.virt_addr) {
 		io_cq->cdesc_addr.virt_addr =
 			dma_alloc_coherent(ena_dev->dmadev, size, &io_cq->cdesc_addr.phys_addr,