Re: [PATCH] net: pktgen: support injecting packets for qdisc testing

[John Fastabend <john.fastabend@gmail.com>]

On 16-01-07 06:09 PM, John Fastabend wrote:
> Add another xmit_mode to pktgen to allow testing xmit functionality
> of qdiscs. The new mode "queue_xmit" injects packets at
> __dev_queue_xmit() so that qdisc is called.
> 
> Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
> ---


[...]

>  		} while (--burst > 0);
>  		goto out; /* Skips xmit_mode M_START_XMIT */
> +	} else if (pkt_dev->xmit_mode == M_QUEUE_XMIT) {
> +		atomic_add(burst, &pkt_dev->skb->users);
> +		local_bh_disable();
> +
> +		do {
> +			ret = dev_queue_xmit(pkt_dev->skb);
> +			switch (ret) {
> +			case NET_XMIT_SUCCESS:
> +			case NET_XMIT_DROP:
> +			case NET_XMIT_CN:
> +			case NET_XMIT_POLICED:
> +			/* These are all valid return codes for a qdisc so
> +			 * unlike start_xmit report this as successful
> +			 * transactions. I expect testers will be also be
> +			 * reviewing qdisc stats for more details.
> +			 */
> +				pkt_dev->last_ok = 1;
> +				pkt_dev->sofar++;
> +				pkt_dev->seq_num++;
> +				pkt_dev->tx_bytes += pkt_dev->last_pkt_size;
> +				break;
> +			case NETDEV_TX_LOCKED:
> +			case NETDEV_TX_BUSY:
> +			/* qdisc may call dev_hard_start_xmit directly in cases
> +			 * where no queues exist e.g. loopback device, virtual
> +			 * devices, etc. In this case we need to handle
> +			 * NETDEV_TX_ codes.
> +			 */
> +				pkt_dev->errors++;
> +				break;
> +			default:
> +				net_info_ratelimited("%s xmit error: %d\n",
> +						     pkt_dev->odevname, ret);
> +				pkt_dev->errors++;
> +				break;
> +			}
> +		} while (--burst > 0);


oops missed the case where burst != 0 here and we need to do an
atomic_sub on skb->users.

.John