netdev - Re: [PATCH net-next v2 2/3] dpaa2-eth: Support multiple traffic classes on Tx

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 06 Jun 2019 11:02:33 -0700 (PDT)
From:   David Miller <davem@...emloft.net>
To:     ruxandra.radulescu@....com
Cc:     netdev@...r.kernel.org, ioana.ciornei@....com
Subject: Re: [PATCH net-next v2 2/3] dpaa2-eth: Support multiple traffic
 classes on Tx

From: Ioana Radulescu <ruxandra.radulescu@....com>
Date: Thu,  6 Jun 2019 11:50:28 +0300

> DPNI objects can have multiple traffic classes, as reflected by
> the num_tc attribute. Until now we ignored its value and only
> used traffic class 0.
> 
> This patch adds support for multiple Tx traffic classes; the skb
> priority information received from the stack is used to select the
> hardware Tx queue on which to enqueue the frame.
> 
> Signed-off-by: Ioana Radulescu <ruxandra.radulescu@....com>
> ---
> v2: Extra processing on the fast path happens only when TC is used
> 
>  drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c | 47 ++++++++++++++++--------
>  drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.h |  9 ++++-
>  2 files changed, 40 insertions(+), 16 deletions(-)
> 
> diff --git a/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c b/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c
> index a12fc45..98de092 100644
> --- a/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c
> +++ b/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c
> @@ -757,6 +757,7 @@ static netdev_tx_t dpaa2_eth_tx(struct sk_buff *skb, struct net_device *net_dev)
>  	u16 queue_mapping;
>  	unsigned int needed_headroom;
>  	u32 fd_len;
> +	u8 prio = 0;
>  	int err, i;
>  
>  	percpu_stats = this_cpu_ptr(priv->percpu_stats);
> @@ -814,6 +815,18 @@ static netdev_tx_t dpaa2_eth_tx(struct sk_buff *skb, struct net_device *net_dev)
>  	 * a queue affined to the same core that processed the Rx frame
>  	 */
>  	queue_mapping = skb_get_queue_mapping(skb);
> +
> +	if (net_dev->num_tc) {
> +		prio = netdev_txq_to_tc(net_dev, queue_mapping);
> +		/* Hardware interprets priority level 0 as being the highest,
> +		 * so we need to do a reverse mapping to the netdev tc index
> +		 */
> +		prio = net_dev->num_tc - prio - 1;
> +		/* We have only one FQ array entry for all Tx hardware queues
> +		 * with the same flow id (but different priority levels)
> +		 */
> +		queue_mapping %= dpaa2_eth_queue_count(priv);

This doesn't make any sense.

queue_mapping came from skb_get_queue_mapping().

The core limits the queue mapping value to whatever you told the
generic networking layer was the maximum number of queues.

And you set that to dpaa2_eth_queue_count():

	/* Set actual number of queues in the net device */
	num_queues = dpaa2_eth_queue_count(priv);
	err = netif_set_real_num_tx_queues(net_dev, num_queues);

Therfore the modulus cannot be needed.