Hi Niklas,
Thanks for the detailed explanation!
On 30-11-2017 03:51, Niklas Cassel wrote:
>
> Could you try to see if the following patch
> solves your problem:
> (still don't see why it should be needed,
> considering stmmac_tx_clean() should set all non-NULL
> entries to NULL)
>
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> index 1763e48c84e2..1d30b20b3740 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> @@ -2830,6 +2830,7 @@ static netdev_tx_t stmmac_tso_xmit(struct sk_buff *skb,
> struct net_device *dev)
>
> tx_q->tx_skbuff_dma[first_entry].buf = des;
> tx_q->tx_skbuff_dma[first_entry].len = skb_headlen(skb);
> + tx_q->tx_skbuff[first_entry] = NULL;
>
> first->des0 = cpu_to_le32(des);
>
> @@ -3003,6 +3004,8 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb,
> struct net_device *dev)
>
> first = desc;
>
> + tx_q->tx_skbuff[first_entry] = NULL;
> +
> enh_desc = priv->plat->enh_desc;
> /* To program the descriptors according to the size of the frame */
> if (enh_desc)
I confirm that applying 05cf0d1bf4 ("net: stmmac: free an skb
first when there are no longer any descriptors using it") and
this patch makes my setup work perfectly in multi-queue
configuration. Can you send an official patch to -net?
You can add my:
Tested-by: Jose Abreu <joab...@synopsys.com>
Also, maybe we should try to understand why stmmac_tx_clean() is
not setting the entry to NULL?
Thanks and Best Regards,
Jose Miguel Abreu
