diff mbox series

net: ethernet: mtk_eth_soc: fix misuse of mem alloc interface netdev_alloc_frag

Message ID 1654229435-2934-1-git-send-email-chen45464546@163.com (mailing list archive)
State New, archived
Headers show
Series net: ethernet: mtk_eth_soc: fix misuse of mem alloc interface netdev_alloc_frag | expand

Commit Message

Chen Lin June 3, 2022, 4:10 a.m. UTC
When rx_flag == MTK_RX_FLAGS_HWLRO, 
rx_data_len = MTK_MAX_LRO_RX_LENGTH(4096 * 3) > PAGE_SIZE.
netdev_alloc_frag is for alloction of page fragment only.
Reference to other drivers and Documentation/vm/page_frags.rst

Branch to use kmalloc when rx_data_len > PAGE_SIZE.

Signed-off-by: Chen Lin <chen45464546@163.com>
---
 drivers/net/ethernet/mediatek/mtk_eth_soc.c |    5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

Comments

Jakub Kicinski June 3, 2022, 4:24 a.m. UTC | #1
On Fri,  3 Jun 2022 12:10:35 +0800 Chen Lin wrote:
> -		ring->data[i] = netdev_alloc_frag(ring->frag_size);
> +		if (ring->frag_size <= PAGE_SIZE)
> +			ring->data[i] = netdev_alloc_frag(ring->frag_size);
> +		else
> +			ring->data[i] = kmalloc(ring->frag_size, GFP_KERNEL);

Is it legal to allocate pages with kmalloc()? I mean they will end up
getting freed by page_frag_free(), not kfree().

Also there's more frag allocations here, search for napi_alloc_frag().
Felix Fietkau June 3, 2022, 4:30 a.m. UTC | #2
On 03.06.22 06:10, Chen Lin wrote:
> When rx_flag == MTK_RX_FLAGS_HWLRO,
> rx_data_len = MTK_MAX_LRO_RX_LENGTH(4096 * 3) > PAGE_SIZE.
> netdev_alloc_frag is for alloction of page fragment only.
> Reference to other drivers and Documentation/vm/page_frags.rst
> 
> Branch to use kmalloc when rx_data_len > PAGE_SIZE.
> 
> Signed-off-by: Chen Lin <chen45464546@163.com>
> ---
>   drivers/net/ethernet/mediatek/mtk_eth_soc.c |    5 ++++-
>   1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ethernet/mediatek/mtk_eth_soc.c b/drivers/net/ethernet/mediatek/mtk_eth_soc.c
> index b3b3c07..d0eebca 100644
> --- a/drivers/net/ethernet/mediatek/mtk_eth_soc.c
> +++ b/drivers/net/ethernet/mediatek/mtk_eth_soc.c
> @@ -1914,7 +1914,10 @@ static int mtk_rx_alloc(struct mtk_eth *eth, int ring_no, int rx_flag)
>   		return -ENOMEM;
>   
>   	for (i = 0; i < rx_dma_size; i++) {
> -		ring->data[i] = netdev_alloc_frag(ring->frag_size);
> +		if (ring->frag_size <= PAGE_SIZE)
> +			ring->data[i] = netdev_alloc_frag(ring->frag_size);
> +		else
> +			ring->data[i] = kmalloc(ring->frag_size, GFP_KERNEL);
I'm pretty sure you also need to update all the other places in the code 
that currently assume that the buffer is allocated using the page frag 
allocator.

- Felix
diff mbox series

Patch

diff --git a/drivers/net/ethernet/mediatek/mtk_eth_soc.c b/drivers/net/ethernet/mediatek/mtk_eth_soc.c
index b3b3c07..d0eebca 100644
--- a/drivers/net/ethernet/mediatek/mtk_eth_soc.c
+++ b/drivers/net/ethernet/mediatek/mtk_eth_soc.c
@@ -1914,7 +1914,10 @@  static int mtk_rx_alloc(struct mtk_eth *eth, int ring_no, int rx_flag)
 		return -ENOMEM;
 
 	for (i = 0; i < rx_dma_size; i++) {
-		ring->data[i] = netdev_alloc_frag(ring->frag_size);
+		if (ring->frag_size <= PAGE_SIZE)
+			ring->data[i] = netdev_alloc_frag(ring->frag_size);
+		else
+			ring->data[i] = kmalloc(ring->frag_size, GFP_KERNEL);
 		if (!ring->data[i])
 			return -ENOMEM;
 	}