Re: [PATCH 1/2] pch_can: Stop netif queue if last message buffer is filled

[Wolfgang Grandegger <wg@grandegger.com>]

Hi Alexander,

On 02/21/2012 05:00 PM, Alexander Stein wrote:
> If all Tx buffers are empty and we start queuing frames (more than Tx
> buffer size) beginning at the last message buffer we will override the
> first inserted entry if it doesn't get sent meanwhile, e.g. low priority
> or ACK error. There is no stop condition when all buffers are used.
> We will get get awoken by pch_can_tx_complete when the last entry is sent.
> 
> Signed-off-by: Alexander Stein <alexander.stein@systec-electronic.com>
> ---
>  drivers/net/can/pch_can.c |    4 +---
>  1 files changed, 1 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/net/can/pch_can.c b/drivers/net/can/pch_can.c
> index 6edc25e..3af5341 100644
> --- a/drivers/net/can/pch_can.c
> +++ b/drivers/net/can/pch_can.c
> @@ -903,9 +903,7 @@ static netdev_tx_t pch_xmit(struct sk_buff *skb, struct net_device *ndev)
>  
>  	tx_obj_no = priv->tx_obj;
>  	if (priv->tx_obj == PCH_TX_OBJ_END) {
> -		if (ioread32(&priv->regs->treq2) & PCH_TREQ2_TX_MASK)
> -			netif_stop_queue(ndev);
> -
> +		netif_stop_queue(ndev);
>  		priv->tx_obj = PCH_TX_OBJ_START;
>  	} else {
>  		priv->tx_obj++;

The pch_can uses a C_CAN core (and for maintenance reasons we would like
to get ride of it sooner than later). In that driver the queue is
stopped here:

http://lxr.linux.no/#linux+v3.2.7/drivers/net/can/c_can/c_can.c#L508

Therefore I'm not sure if your fix is enough (without looking into the
manual). I could send you a "pch_pci" driver patch for C_CAN, which
would allow you to use it. What kernel version are you using?

Thanks,

Wolfgang.