[net-next,10/12] bna: TX Queue Depth Fix

sk_buff unmap_array grows greater than 65536 (x2) with Tx ring of 65536.
Reducing TXQ depth and safe(max) acking of Tx events to 32768 (same as Rx).
Add defines for TX and RX queue depths.

Signed-off-by: Gurunatha Karaje <gkaraje@brocade.com>
Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bnad.h         |    4 ++++
 drivers/net/ethernet/brocade/bna/bnad_ethtool.c |    8 ++++----
 2 files changed, 8 insertions(+), 4 deletions(-)

Le mardi 30 août 2011 à 18:27 -0700, Rasesh Mody a écrit :
> sk_buff unmap_array grows greater than 65536 (x2) with Tx ring of 65536.
> Reducing TXQ depth and safe(max) acking of Tx events to 32768 (same as Rx).
> Add defines for TX and RX queue depths.
> 
> Signed-off-by: Gurunatha Karaje <gkaraje@brocade.com>
> Signed-off-by: Rasesh Mody <rmody@brocade.com>
> ---
>  drivers/net/ethernet/brocade/bna/bnad.h         |    4 ++++
>  drivers/net/ethernet/brocade/bna/bnad_ethtool.c |    8 ++++----
>  2 files changed, 8 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/net/ethernet/brocade/bna/bnad.h b/drivers/net/ethernet/brocade/bna/bnad.h
> index 0c9d736..6c42c14 100644
> --- a/drivers/net/ethernet/brocade/bna/bnad.h
> +++ b/drivers/net/ethernet/brocade/bna/bnad.h
> @@ -86,6 +86,10 @@ struct bnad_rx_ctrl {
>  #define BNAD_MAX_Q_DEPTH		0x10000
>  #define BNAD_MIN_Q_DEPTH		0x200
>  
> +#define BNAD_MAX_RXQ_DEPTH		(BNAD_MAX_Q_DEPTH / bnad_rxqs_per_cq)
> +/* keeping MAX TX and RX Q depth equal */
> +#define BNAD_MAX_TXQ_DEPTH		BNAD_MAX_RXQ_DEPTH
> +
>  #define BNAD_JUMBO_MTU			9000
>  
>  #define BNAD_NETIF_WAKE_THRESHOLD	8
> diff --git a/drivers/net/ethernet/brocade/bna/bnad_ethtool.c b/drivers/net/ethernet/brocade/bna/bnad_ethtool.c
> index 96ff700..4842224 100644
> --- a/drivers/net/ethernet/brocade/bna/bnad_ethtool.c
> +++ b/drivers/net/ethernet/brocade/bna/bnad_ethtool.c
> @@ -418,10 +418,10 @@ bnad_get_ringparam(struct net_device *netdev,
>  {
>  	struct bnad *bnad = netdev_priv(netdev);
>  
> -	ringparam->rx_max_pending = BNAD_MAX_Q_DEPTH / bnad_rxqs_per_cq;
> +	ringparam->rx_max_pending = BNAD_MAX_RXQ_DEPTH;
>  	ringparam->rx_mini_max_pending = 0;
>  	ringparam->rx_jumbo_max_pending = 0;
> -	ringparam->tx_max_pending = BNAD_MAX_Q_DEPTH;
> +	ringparam->tx_max_pending = BNAD_MAX_TXQ_DEPTH;
>  
>  	ringparam->rx_pending = bnad->rxq_depth;
>  	ringparam->rx_mini_max_pending = 0;
> @@ -445,13 +445,13 @@ bnad_set_ringparam(struct net_device *netdev,
>  	}
>  
>  	if (ringparam->rx_pending < BNAD_MIN_Q_DEPTH ||
> -	    ringparam->rx_pending > BNAD_MAX_Q_DEPTH / bnad_rxqs_per_cq ||
> +	    ringparam->rx_pending > BNAD_MAX_RXQ_DEPTH ||
>  	    !BNA_POWER_OF_2(ringparam->rx_pending)) {
>  		mutex_unlock(&bnad->conf_mutex);
>  		return -EINVAL;
>  	}
>  	if (ringparam->tx_pending < BNAD_MIN_Q_DEPTH ||
> -	    ringparam->tx_pending > BNAD_MAX_Q_DEPTH ||
> +	    ringparam->tx_pending > BNAD_MAX_TXQ_DEPTH ||
>  	    !BNA_POWER_OF_2(ringparam->tx_pending)) {
>  		mutex_unlock(&bnad->conf_mutex);
>  		return -EINVAL;

BNAD_MAX_Q_DEPTH is now defined but not used.

BTW, 32768 slots in a RX queue is insane.


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

On Tue, Aug 30, 2011 at 9:24 PM, Eric Dumazet <eric.dumazet@gmail.com> wrote:
> Le mardi 30 août 2011 à 18:27 -0700, Rasesh Mody a écrit :
>> sk_buff unmap_array grows greater than 65536 (x2) with Tx ring of 65536.
>> Reducing TXQ depth and safe(max) acking of Tx events to 32768 (same as Rx).
>> Add defines for TX and RX queue depths.

I was happy at the 'reducing txq depth' portion of the email, then I
saw the value it was reduced to.

thx for commenting thusly...

> BNAD_MAX_Q_DEPTH is now defined but not used.
>
> BTW, 32768 slots in a RX queue is insane.

As that is many lunar distances worth of buffering... Are they
expecting transmissions from mars?

Le mardi 30 août 2011 à 21:34 -0700, Dave Taht a écrit :
> On Tue, Aug 30, 2011 at 9:24 PM, Eric Dumazet <eric.dumazet@gmail.com> wrote:

> > BNAD_MAX_Q_DEPTH is now defined but not used.
> >
> > BTW, 32768 slots in a RX queue is insane.
> 
> As that is many lunar distances worth of buffering... Are they
> expecting transmissions from mars?
> 

At 10GB rate, this can be filled in ~2msec with small frames.

Apparently some people business depend on not losing frames.

But they should be warned of extra memory needs, and latency increases
(because of a larger working set)



--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html