[PATCH v1 net] net: ena: Fix redundant device NUMA node override

[PATCH v1 net] net: ena: Fix redundant device NUMA node override

[Shay Agroskin]

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: 1738cd3ed342 ("net: ena: Add a driver for Amazon Elastic Network Adapters (ENA)")
Signed-off-by: Shay Agroskin <shayagr@amazon.com>
---
 drivers/net/ethernet/amazon/ena/ena_com.c | 11 -----------
 1 file changed, 11 deletions(-)

diff --git a/drivers/net/ethernet/amazon/ena/ena_com.c b/drivers/net/ethernet/amazon/ena/ena_com.c
index 2d8a66ea82fa..713a595370bf 100644
--- a/drivers/net/ethernet/amazon/ena/ena_com.c
+++ b/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,
-- 
2.34.1

Re: [PATCH v1 net] net: ena: Fix redundant device NUMA node override

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Tue, 28 May 2024 20:09:12 +0300 you wrote:
> 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.
> 
> [...]

Here is the summary with links:
  - [v1,net] net: ena: Fix redundant device NUMA node override
    https://git.kernel.org/netdev/net/c/2dc8b1e7177d

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html