From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from dpdk.org (dpdk.org [92.243.14.124]) by inbox.dpdk.org (Postfix) with ESMTP id DEEB3A04C0; Thu, 8 Oct 2020 18:25:07 +0200 (CEST) Received: from [92.243.14.124] (localhost [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id BB89E1B878; Thu, 8 Oct 2020 18:25:06 +0200 (CEST) Received: from mga12.intel.com (mga12.intel.com [192.55.52.136]) by dpdk.org (Postfix) with ESMTP id 35B04A69 for ; Thu, 8 Oct 2020 18:25:04 +0200 (CEST) IronPort-SDR: 1ohvSctyGRof0L6bTGavE5ok3w/P2SwkUmlE6INUG4/7BOZz/NxU+dzIn9fEXJDhxgYDPeo+tl R1lGiGrxBKKA== X-IronPort-AV: E=McAfee;i="6000,8403,9768"; a="144688729" X-IronPort-AV: E=Sophos;i="5.77,351,1596524400"; d="scan'208";a="144688729" X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga004.jf.intel.com ([10.7.209.38]) by fmsmga106.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 08 Oct 2020 09:25:02 -0700 IronPort-SDR: pTRlIDei7MdW8YzF2DXd2TE8hSXHS/5HRpNstFA4bN1vij7XE5ss5Uh9adh8XWBEc0OIZrHlMV 37TCfUVqJaSw== X-IronPort-AV: E=Sophos;i="5.77,351,1596524400"; d="scan'208";a="461865788" Received: from fyigit-mobl1.ger.corp.intel.com (HELO [10.213.226.103]) ([10.213.226.103]) by orsmga004-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 08 Oct 2020 09:25:01 -0700 From: Ferruh Yigit To: rohit.raj@nxp.com, Ray Kinsella , Neil Horman Cc: dev@dpdk.org References: <20200824082414.30535-1-rohit.raj@nxp.com> <20200826055233.26075-1-rohit.raj@nxp.com> <23bd31be-06a2-af9a-2b54-8e44cf9b2c79@intel.com> Message-ID: <9561d2ed-8392-602b-5c98-d189e1bccbd9@intel.com> Date: Thu, 8 Oct 2020 17:24:58 +0100 MIME-Version: 1.0 In-Reply-To: <23bd31be-06a2-af9a-2b54-8e44cf9b2c79@intel.com> Content-Type: text/plain; charset=utf-8; format=flowed Content-Language: en-US Content-Transfer-Encoding: 8bit Subject: Re: [dpdk-dev] [PATCH v3 1/3] eal: add API for bus close X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" On 9/30/2020 12:50 PM, Ferruh Yigit wrote: > On 8/26/2020 6:52 AM, rohit.raj@nxp.com wrote: >> From: Rohit Raj >> >> As per the current code we have API for bus probe, but the >> bus close API is missing. This breaks the multi process >> scenarios as objects are not cleaned while terminating the >> secondary processes. >> >> This patch adds a new API rte_bus_close() for cleanup of >> bus objects which were acquired during probe. >> >> Signed-off-by: Rohit Raj >> --- >> >> v3: >> * nit: combined nested if statements >> >> v2: >> * Moved rte_bus_close call to rte_eal_cleanup path. >> >>   lib/librte_eal/common/eal_common_bus.c | 32 +++++++++++++++++++++++++- >>   lib/librte_eal/include/rte_bus.h       | 25 +++++++++++++++++++- >>   lib/librte_eal/linux/eal.c             |  1 + >>   lib/librte_eal/rte_eal_version.map     |  3 +++ >>   4 files changed, 59 insertions(+), 2 deletions(-) >> >> diff --git a/lib/librte_eal/common/eal_common_bus.c >> b/lib/librte_eal/common/eal_common_bus.c >> index baa5b532a..5fd7cf6c5 100644 >> --- a/lib/librte_eal/common/eal_common_bus.c >> +++ b/lib/librte_eal/common/eal_common_bus.c >> @@ -1,5 +1,5 @@ >>   /* SPDX-License-Identifier: BSD-3-Clause >> - * Copyright 2016 NXP >> + * Copyright 2016,2020 NXP >>    */ >>   #include >> @@ -56,6 +56,36 @@ rte_bus_scan(void) >>       return 0; >>   } >> +int >> +rte_bus_close(void) >> +{ >> +    int ret; >> +    struct rte_bus *bus, *vbus = NULL; >> + >> +    TAILQ_FOREACH(bus, &rte_bus_list, next) { >> +        if (!strcmp(bus->name, "vdev")) { >> +            vbus = bus; >> +            continue; >> +        } > > This special treatment for 'vdev' bus is done in probe to be sure physically > device port ids start from '0',  I guess we don't need to do this for 'close'. > >> + >> +        if (bus->close) { >> +            ret = bus->close(); >> +            if (ret) >> +                RTE_LOG(ERR, EAL, "Bus (%s) close failed.\n", >> +                    bus->name); >> +        } >> +    } >> + >> +    if (vbus && vbus->close) { >> +        ret = vbus->close(); >> +        if (ret) >> +            RTE_LOG(ERR, EAL, "Bus (%s) close failed.\n", >> +                vbus->name); >> +    } >> + >> +    return 0; >> +} >> + >>   /* Probe all devices of all buses */ >>   int >>   rte_bus_probe(void) >> diff --git a/lib/librte_eal/include/rte_bus.h b/lib/librte_eal/include/rte_bus.h >> index d3034d0ed..af4787b18 100644 >> --- a/lib/librte_eal/include/rte_bus.h >> +++ b/lib/librte_eal/include/rte_bus.h >> @@ -1,5 +1,5 @@ >>   /* SPDX-License-Identifier: BSD-3-Clause >> - * Copyright 2016 NXP >> + * Copyright 2016,2020 NXP >>    */ >>   #ifndef _RTE_BUS_H_ >> @@ -67,6 +67,18 @@ typedef int (*rte_bus_scan_t)(void); >>    */ >>   typedef int (*rte_bus_probe_t)(void); >> +/** >> + * Implementation specific close function which is responsible for closing >> + * devices on that bus. >> + * >> + * This is called while iterating over each registered bus. >> + * >> + * @return >> + *    0 for successful close >> + *    !0 for any error while closing >> + */ >> +typedef int (*rte_bus_close_t)(void); >> + > > As I checked the 'rte_fslmc_bus->close()' ops, it iterates on all devices in the > bus instead of doing a bus level close, in that case > instead of adding a new 'close' bus operations, will it work if existing > 'bus->unplug(dev)' used? > Whatever done in the 'rte_fslmc_bus->close()' per device, can it be done under > the 'fslmc_bus_unplug()'? > > And in that case a 'rte_bus_remove()' API can be added which can call > 'bus->unplug(dev)' for all buses and it will be beneficial for all buses, and it > can fit well into the 'rte_eal_cleanup()'. > > What do you think? Hi Rohit, I have seen new version has been sent today, I want to remind above question, can you please check it?