Re: [PATCH 6/7] macb: fix tx ring size

[Alexander Aring <alex.aring@gmail.com>]

Hi,

On Fri, Feb 08, 2013 at 10:18:49AM +0100, Jean-Christophe PLAGNIOL-VILLARD wrote:
> the mininal tx ring size is 2 as if one we wrap on the same descriptor
> and can cause IP lock on GEM (gigabit version) this is always the case
> 
> Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
> Signed-off-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
> ---
>  drivers/net/macb.c |   50 ++++++++++++++++++++++++++++++++++----------------
>  1 file changed, 34 insertions(+), 16 deletions(-)
> 
> diff --git a/drivers/net/macb.c b/drivers/net/macb.c
> index d6b60aa..3a65af7 100644
> --- a/drivers/net/macb.c
> +++ b/drivers/net/macb.c
> @@ -53,15 +53,17 @@
>  
>  #define MACB_RX_BUFFER_SIZE	128
>  #define RX_BUFFER_MULTIPLE	64  /* bytes */
> -#define RX_RING_SIZE		32 /* must be power of 2 */
> -#define RX_RING_BYTES		(sizeof(struct macb_dma_desc) * RX_RING_SIZE)
> +#define RX_NB_PACKET		10
> +#define TX_RING_SIZE		2 /* must be power of 2 */
>  
> -#define TX_RING_BYTES		(sizeof(struct macb_dma_desc))
> +#define RX_RING_BYTES(bp)	(sizeof(struct macb_dma_desc) * bp->rx_ring_size)
> +#define TX_RING_BYTES		(sizeof(struct macb_dma_desc) * TX_RING_SIZE)
>  
>  struct macb_device {
>  	void			__iomem *regs;
>  
>  	unsigned int		rx_tail;
> +	unsigned int		tx_head;
>  	unsigned int		tx_tail;
>  
>  	void			*rx_buffer;
> @@ -86,23 +88,36 @@ static int macb_send(struct eth_device *edev, void *packet,
>  {
>  	struct macb_device *macb = edev->priv;
>  	unsigned long ctrl;
> -	int ret;
> -
> -	dev_dbg(macb->dev, "%s\n", __func__);
> +	int ret = 0;
> +	uint64_t start;
> +	unsigned int tx_head = macb->tx_head;
>  
>  	ctrl = MACB_BF(TX_FRMLEN, length);
> -	ctrl |= MACB_BIT(TX_LAST) | MACB_BIT(TX_WRAP);
> +	ctrl |= MACB_BIT(TX_LAST);
>  
> -	macb->tx_ring[0].ctrl = ctrl;
> -	macb->tx_ring[0].addr = (ulong)packet;
> +	if (tx_head == (TX_RING_SIZE - 1)) {
> +		ctrl |= MACB_BIT(TX_WRAP);
> +		macb->tx_head = 0;
> +	} else {
> +		macb->tx_head++;
> +	}
> +
> +	macb->tx_ring[tx_head].ctrl = ctrl;
> +	macb->tx_ring[tx_head].addr = (ulong)packet;
>  	barrier();
>  	dma_flush_range((ulong) packet, (ulong)packet + length);
>  	macb_writel(macb, NCR, MACB_BIT(TE) | MACB_BIT(RE) | MACB_BIT(TSTART));
>  
> -	ret = wait_on_timeout(100 * MSECOND,
> -		!(macb->tx_ring[0].ctrl & MACB_BIT(TX_USED)));
> -
> -	ctrl = macb->tx_ring[0].ctrl;
> +	start = get_time_ns();
> +	ret = -ETIMEDOUT;

Is this ret ever used?

> +	do {
> +		barrier();
> +		ctrl = macb->tx_ring[0].ctrl;
> +		if (ctrl & MACB_BIT(TX_USED)) {
> +			ret = 0;
> +			break;
> +		}
> +	} while (!is_timeout(start, 100 * MSECOND));
>  
>  	if (ctrl & MACB_BIT(TX_UNDERRUN))
>  		dev_err(macb->dev, "TX underrun\n");
> @@ -244,10 +259,13 @@ static void macb_init(struct macb_device *macb)
>  	}
>  	macb->rx_ring[RX_RING_SIZE - 1].addr |= MACB_BIT(RX_WRAP);
>  
> -	macb->tx_ring[0].addr = 0;
> -	macb->tx_ring[0].ctrl = MACB_BIT(TX_USED) | MACB_BIT(TX_WRAP);
> +	for (i = 0; i < TX_RING_SIZE; i++) {
> +		macb->tx_ring[i].addr = 0;
> +		macb->tx_ring[i].ctrl = MACB_BIT(TX_USED);
> +	}
> +	macb->tx_ring[TX_RING_SIZE - 1].addr |= MACB_BIT(TX_WRAP);
>  
> -	macb->rx_tail = macb->tx_tail = 0;
> +	macb->rx_tail = macb->tx_head = macb->tx_tail = 0;
>  
>  	macb_writel(macb, RBQP, (ulong)macb->rx_ring);
>  	macb_writel(macb, TBQP, (ulong)macb->tx_ring);
> -- 
> 1.7.10.4
> 
> 
> _______________________________________________
> barebox mailing list
> barebox@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/barebox

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox