From mboxrd@z Thu Jan  1 00:00:00 1970
From: Stephen Hemminger <stephen@networkplumber.org>
Subject: Re: [PATCH v2 1/2] ethdev: stop overriding rx_nombuf by
 rte_eth_stats_get
Date: Wed, 23 Aug 2017 14:56:31 -0700
Message-ID: <20170823145631.5fce68dc@xeon-e3>
References: <20170823011937.37579-1-dharton@cisco.com>
 <20170823025555.19022-1-dharton@cisco.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Cc: thomas@monjalon.net, dev@dpdk.org
To: David Harton <dharton@cisco.com>
Return-path: <dev-bounces@dpdk.org>
Received: from mail-pg0-f51.google.com (mail-pg0-f51.google.com [74.125.83.51])
 by dpdk.org (Postfix) with ESMTP id A05A87D5E
 for <dev@dpdk.org>; Wed, 23 Aug 2017 23:56:39 +0200 (CEST)
Received: by mail-pg0-f51.google.com with SMTP id u191so5972043pgc.2
 for <dev@dpdk.org>; Wed, 23 Aug 2017 14:56:39 -0700 (PDT)
In-Reply-To: <20170823025555.19022-1-dharton@cisco.com>
List-Id: DPDK patches and discussions <dev.dpdk.org>
List-Unsubscribe: <http://dpdk.org/ml/options/dev>,
 <mailto:dev-request@dpdk.org?subject=unsubscribe>
List-Archive: <http://dpdk.org/ml/archives/dev/>
List-Post: <mailto:dev@dpdk.org>
List-Help: <mailto:dev-request@dpdk.org?subject=help>
List-Subscribe: <http://dpdk.org/ml/listinfo/dev>,
 <mailto:dev-request@dpdk.org?subject=subscribe>
Errors-To: dev-bounces@dpdk.org
Sender: "dev" <dev-bounces@dpdk.org>

On Tue, 22 Aug 2017 22:55:55 -0400
David Harton <dharton@cisco.com> wrote:

> rte_eth_stats_get() unconditonally would set rx_nombuf
> even if the device was setting the value.  A check has
> been added in rte_eth_stats_get() to leave the device
> value in-tact when non-zero.
> 
> Signed-off-by: David Harton <dharton@cisco.com>
> ---
> 
> v2: Fixed braces complaint required by other coding standards.
> 
>  lib/librte_ether/rte_ethdev.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c
> index 0597641..0a1d3b8 100644
> --- a/lib/librte_ether/rte_ethdev.c
> +++ b/lib/librte_ether/rte_ethdev.c
> @@ -1336,8 +1336,11 @@ struct rte_eth_dev *
>  	memset(stats, 0, sizeof(*stats));
>  
>  	RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->stats_get, -ENOTSUP);
> -	stats->rx_nombuf = dev->data->rx_mbuf_alloc_failed;
>  	(*dev->dev_ops->stats_get)(dev, stats);
> +	/* only set rx_nombuf if not set by the device */
> +	if (!stats->rx_nombuf)
> +		stats->rx_nombuf = dev->data->rx_mbuf_alloc_failed;
> +
>  	return 0;
>  }
>  

This seems backwards. It seems like the original way worked fine.
If device specific code wanted to override rx_nombuf it could do so either
by adding it's additional value or just setting rx_nombuf.

Adding special cases seems like it would start a bad precedent and the
could would end up quite complex as some values had one semantic and others
were only from driver.