netdev - Re: [PATCH net-next 1/2] net/mlx4_en: Add Low Latency Socket (LLS) support

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:	Mon, 17 Jun 2013 04:37:04 -0700
From:	Eric Dumazet <eric.dumazet@...il.com>
To:	Amir Vadai <amirv@...lanox.com>
Cc:	"David S. Miller" <davem@...emloft.net>, netdev@...r.kernel.org,
	eliezer.tamir@...ux.intel.com, Or Gerlitz <ogerlitz@...lanox.com>
Subject: Re: [PATCH net-next 1/2] net/mlx4_en: Add Low Latency Socket (LLS)
 support

On Mon, 2013-06-17 at 13:58 +0300, Amir Vadai wrote:
> Add basic support for LLS.
> 
> Signed-off-by: Amir Vadai <amirv@...lanox.com>
> ---
>  drivers/net/ethernet/mellanox/mlx4/en_cq.c     |   2 +
>  drivers/net/ethernet/mellanox/mlx4/en_netdev.c |  49 +++++++++-
>  drivers/net/ethernet/mellanox/mlx4/en_rx.c     |   8 ++
>  drivers/net/ethernet/mellanox/mlx4/mlx4_en.h   | 124 +++++++++++++++++++++++++
>  4 files changed, 181 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ethernet/mellanox/mlx4/en_cq.c b/drivers/net/ethernet/mellanox/mlx4/en_cq.c
> index 1e6c594..5e47efd 100644
> --- a/drivers/net/ethernet/mellanox/mlx4/en_cq.c
> +++ b/drivers/net/ethernet/mellanox/mlx4/en_cq.c
> @@ -139,6 +139,7 @@ int mlx4_en_activate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq,
>  
>  	if (!cq->is_tx) {
>  		netif_napi_add(cq->dev, &cq->napi, mlx4_en_poll_rx_cq, 64);
> +		napi_hash_add(&cq->napi);
>  		napi_enable(&cq->napi);
>  	}
>  
> @@ -162,6 +163,7 @@ void mlx4_en_deactivate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq)
>  {
>  	if (!cq->is_tx) {
>  		napi_disable(&cq->napi);
> +		napi_hash_del(&cq->napi);

Are we sure an RCU grace period is respected before deletion of memory
holding cq->napi ?

>  		netif_napi_del(&cq->napi);
>  	}
>  

...

>  
>  static const struct net_device_ops mlx4_netdev_ops_master = {
> diff --git a/drivers/net/ethernet/mellanox/mlx4/en_rx.c b/drivers/net/ethernet/mellanox/mlx4/en_rx.c
> index 02aee1e..15b5980 100644
> --- a/drivers/net/ethernet/mellanox/mlx4/en_rx.c
> +++ b/drivers/net/ethernet/mellanox/mlx4/en_rx.c
> @@ -31,6 +31,7 @@
>   *
>   */
>  
> +#include <net/ll_poll.h>
>  #include <linux/mlx4/cq.h>
>  #include <linux/slab.h>
>  #include <linux/mlx4/qp.h>
> @@ -737,6 +738,8 @@ int mlx4_en_process_rx_cq(struct net_device *dev, struct mlx4_en_cq *cq, int bud
>  					       timestamp);
>  		}
>  
> +		skb_mark_ll(skb, &cq->napi);
> +

It seems there is a second point in the driver calling napi_gro_frags(),
around line 696

It looks like LLS wont be supported for this case ?

>  		/* Push it up the stack */
>  		netif_receive_skb(skb);
>  
> @@ -781,8 +784,13 @@ int mlx4_en_poll_rx_cq(struct napi_struct *napi, int budget)
>  	struct mlx4_en_priv *priv = netdev_priv(dev);
>  	int done;
>  
> +	if (!mlx4_en_cq_lock_napi(cq))
> +		return budget;
> +
>  	done = mlx4_en_process_rx_cq(dev, cq, budget);
>  
> +	mlx4_en_cq_unlock_napi(cq);
> +
>  	/* If we used up all the quota - we're probably not done yet... */
>  	if (done == budget)
>  		INC_PERF_COUNTER(priv->pstats.napi_quota);


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