From mboxrd@z Thu Jan  1 00:00:00 1970
From: Bruce Richardson <bruce.richardson@intel.com>
Subject: Re: [PATCH 2/3] eal: add functions parsing EAL arguments
Date: Wed, 31 May 2017 16:46:00 +0100
Message-ID: <20170531154559.GA46408@bricha3-MOBL3.ger.corp.intel.com>
References: <1496133037-3014-1-git-send-email-jacekx.piasecki@intel.com>
 <1496133037-3014-3-git-send-email-jacekx.piasecki@intel.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: dev@dpdk.org, deepak.k.jain@intel.com, michalx.k.jastrzebski@intel.com,
 Kuba Kozak <kubax.kozak@intel.com>
To: Jacek Piasecki <jacekx.piasecki@intel.com>
Return-path: <dev-bounces@dpdk.org>
Received: from mga11.intel.com (mga11.intel.com [192.55.52.93])
 by dpdk.org (Postfix) with ESMTP id 1E6DB7CC3
 for <dev@dpdk.org>; Wed, 31 May 2017 17:46:07 +0200 (CEST)
Content-Disposition: inline
In-Reply-To: <1496133037-3014-3-git-send-email-jacekx.piasecki@intel.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, May 30, 2017 at 10:30:36AM +0200, Jacek Piasecki wrote:
> From: Kuba Kozak <kubax.kozak@intel.com>
> 
> added function rte_eal_configure which translate
> options from config file into argc, **argv form.
> 
> changed function rte_eal_init to meld
> argc, argv** options from config file with
> these from command line and then parse as
> before
> 

Hi,

while this approach to merging the values from two sources (config
struct and cmdline) works, I'm not sure it's the best way to implement
this. I would have thought it more logical to separate out cmdline args
into name-value pairs to add to the cfgfile struct internally instead.
However, this has the disadvantage of forcing a dependency on cfgfile,
or adding a third name-value pair struct to DPDK [after kvargs, which
supports flat pairs, and cfgfile, which supports sections and pairs].
Using the cfgfile struct also saves us from "flattening out"
hierarchical information we may wish to add in the config file. For
example, consider the following cfgfile structure:

[DPDK]
lcores = 20-25
socket-mem = 2048, 1024
vdevs = 2

[DPDK.vdev0]
name = eth_pcap0
rx_pcap = /path/to/file
tx_pcap = /path/to/other_file

[DPDK.vdev1]
name = eth_tap0
iface = iface_arg

Longer term, it would be good to keep those vdev entries separated as passed
as such to the driver, rather than merging them back into a single
string to be separated out again by kvargs lib.

/Bruce