Re: [PATCH net v1] net: stmmac: TSO: Fix unaligned DMA unmap for non-paged SKB data

[Jon Hunter <jonathanh@nvidia.com>]

On 05/12/2024 09:18, Furong Xu wrote:
> Commit 66600fac7a98 ("net: stmmac: TSO: Fix unbalanced DMA map/unmap for
> non-paged SKB data") assigns a wrong DMA buffer address that is added an
> offset of proto_hdr_len to tx_q->tx_skbuff_dma[entry].buf on a certain
> platform that the DMA AXI address width is configured to 40-bit/48-bit,

In our case it is 40/40 ...

  dwc-eth-dwmac 2490000.ethernet: Using 40/40 bits DMA host/device width

> stmmac_tx_clean() will try to unmap this illegal DMA buffer address
> and many crashes are reported: [1] [2].
> 
> This patch guarantees that DMA address is passed to stmmac_tx_clean()
> unmodified and without offset.
> 
> [1] https://lore.kernel.org/all/d8112193-0386-4e14-b516-37c2d838171a@nvidia.com/
> [2] https://lore.kernel.org/all/klkzp5yn5kq5efgtrow6wbvnc46bcqfxs65nz3qy77ujr5turc@bwwhelz2l4dw/
> 
> Reported-by: Jon Hunter <jonathanh@nvidia.com>
> Reported-by: Thierry Reding <thierry.reding@gmail.com>
> Suggested-by: Russell King (Oracle) <linux@armlinux.org.uk>
> Fixes: 66600fac7a98 ("net: stmmac: TSO: Fix unbalanced DMA map/unmap for non-paged SKB data")
> Signed-off-by: Furong Xu <0x1207@gmail.com>
> ---
>   drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 5 +++--
>   1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> index 9b262cdad60b..7227f8428b5e 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> @@ -4192,8 +4192,8 @@ static netdev_tx_t stmmac_tso_xmit(struct sk_buff *skb, struct net_device *dev)
>   	struct stmmac_txq_stats *txq_stats;
>   	struct stmmac_tx_queue *tx_q;
>   	u32 pay_len, mss, queue;
> +	dma_addr_t tso_hdr, des;
>   	u8 proto_hdr_len, hdr;
> -	dma_addr_t des;
>   	bool set_ic;
>   	int i;
>   
> @@ -4279,6 +4279,7 @@ static netdev_tx_t stmmac_tso_xmit(struct sk_buff *skb, struct net_device *dev)
>   			     DMA_TO_DEVICE);
>   	if (dma_mapping_error(priv->device, des))
>   		goto dma_map_err;
> +	tso_hdr = des;
>   
>   	if (priv->dma_cap.addr64 <= 32) {
>   		first->des0 = cpu_to_le32(des);
> @@ -4310,7 +4311,7 @@ static netdev_tx_t stmmac_tso_xmit(struct sk_buff *skb, struct net_device *dev)
>   	 * this DMA buffer right after the DMA engine completely finishes the
>   	 * full buffer transmission.
>   	 */
> -	tx_q->tx_skbuff_dma[tx_q->cur_tx].buf = des;
> +	tx_q->tx_skbuff_dma[tx_q->cur_tx].buf = tso_hdr;
>   	tx_q->tx_skbuff_dma[tx_q->cur_tx].len = skb_headlen(skb);
>   	tx_q->tx_skbuff_dma[tx_q->cur_tx].map_as_page = false;
>   	tx_q->tx_skbuff_dma[tx_q->cur_tx].buf_type = STMMAC_TXBUF_T_SKB;

Otherwise ...

Tested-by: Jon Hunter <jonathanh@nvidia.com>

Jon

-- 
nvpublic