From mboxrd@z Thu Jan  1 00:00:00 1970
From: Thomas Monjalon <thomas@monjalon.net>
Subject: Re: [PATCH v7 12/17] devargs: parse bus policies
Date: Sun, 09 Jul 2017 16:50:28 +0200
Message-ID: <8124922.CSY8T9Vcb2@xps>
References: <cover.1499211959.git.gaetan.rivet@6wind.com>
 <cover.1499384589.git.gaetan.rivet@6wind.com>
 <8816dcfaf7d7554cf1a105cf941b43dc8a9e7423.1499384590.git.gaetan.rivet@6wind.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7Bit
Cc: dev@dpdk.org, jblunck@infradead.org, matan@mellanox.com
To: Gaetan Rivet <gaetan.rivet@6wind.com>
Return-path: <dev-bounces@dpdk.org>
Received: from out3-smtp.messagingengine.com (out3-smtp.messagingengine.com
 [66.111.4.27]) by dpdk.org (Postfix) with ESMTP id C33AF58CE
 for <dev@dpdk.org>; Sun,  9 Jul 2017 16:50:29 +0200 (CEST)
In-Reply-To: <8816dcfaf7d7554cf1a105cf941b43dc8a9e7423.1499384590.git.gaetan.rivet@6wind.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>

Hi,

I think there is a real bug to fix in 17.08-rc2.
More details below. Proposals are welcome.

07/07/2017 02:04, Gaetan Rivet:
> --- a/lib/librte_eal/common/eal_common_devargs.c
> +++ b/lib/librte_eal/common/eal_common_devargs.c
> @@ -143,6 +143,21 @@ rte_eal_devargs_add(enum rte_devtype devtype, const char *devargs_str)
>  
>  		break;
>  	}
> +	if (devargs->type == RTE_DEVTYPE_WHITELISTED_PCI) {
> +		if (bus->conf.scan_mode == RTE_BUS_SCAN_UNDEFINED) {
> +			bus->conf.scan_mode = RTE_BUS_SCAN_WHITELIST;
> +		} else if (bus->conf.scan_mode == RTE_BUS_SCAN_BLACKLIST) {
> +			fprintf(stderr, "ERROR: incompatible device type and bus scan mode\n");
> +			goto fail;
> +		}
> +	} else if (devargs->type == RTE_DEVTYPE_BLACKLISTED_PCI) {
> +		if (bus->conf.scan_mode == RTE_BUS_SCAN_UNDEFINED) {
> +			bus->conf.scan_mode = RTE_BUS_SCAN_BLACKLIST;
> +		} else if (bus->conf.scan_mode == RTE_BUS_SCAN_WHITELIST) {
> +			fprintf(stderr, "ERROR: incompatible device type and bus scan mode\n");
> +			goto fail;
> +		}
> +	}

After another look, there is something wrong here.
You are checking a probe policy (wrongly named scan_mode), in a function
which can be not related at all with bus probing.

Example with failsafe:
1/ We blacklist a device for the global EAL probe.
So the probing mode is set from "undefined" to "blacklist".
2/ We add the device as a failsafe slave.
3/ The device must be plugged, no matter of the probe policy.
But it is seen as a whitelist and rejected because the bus probing
is in blacklist mode.

I think it is a serious bug.

PS: thanks Matan for having checked it