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 D8E65A04A3; Tue, 16 Jun 2020 12:14:18 +0200 (CEST) Received: from [92.243.14.124] (localhost [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id 16BB31BEE7; Tue, 16 Jun 2020 12:14:18 +0200 (CEST) Received: from mail-il1-f170.google.com (mail-il1-f170.google.com [209.85.166.170]) by dpdk.org (Postfix) with ESMTP id 313251BEE3 for ; Tue, 16 Jun 2020 12:14:16 +0200 (CEST) Received: by mail-il1-f170.google.com with SMTP id z2so584562ilq.0 for ; Tue, 16 Jun 2020 03:14:16 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=emumba-com.20150623.gappssmtp.com; s=20150623; h=mime-version:references:in-reply-to:from:date:message-id:subject:to :cc; bh=4+eWsqTORT16aIhceRw7Zeab8lBIGeeTsQT6lBqVYFE=; b=xEm/l6R7WE3RpovW7/PNP8qjzw7KmSe8AiNO/MuzoCtxpUgg2EBX60zo84eLFgu1W2 FA3PTcet82nlQbUjoBo6YAIoEgvEdTjN9stlZs1AF0Xj6Yno+npXYMRTRPbXcqWuRrfe BFekho8yp+wE6UcHVO3mY4tPewOW+4YjiA7JfzCA7bO0zfymiFxO5vitDx59N2lTXskr WQDnIghqWPuj4hhuc67v4zUDVQGF9GqGCYypG2ogFxId70I4wC/Np6a/jsllPVay4UqY lRwUsmJ8woKIi1HWMacVucLjoD9GpRzmSfoAiAW/ZCCw8q65u0jKI/TKTU3s21aXRSxR Qlxg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to:cc; bh=4+eWsqTORT16aIhceRw7Zeab8lBIGeeTsQT6lBqVYFE=; b=cLLVWoiEAqeRNXjwRfrU+IFZ5Il1JNzT0Z5B/MQr2WgmQcpPMGQnlpsuP8RJJbipF8 juq78DRgxZD5kLsor1brD9l9Ga4KeaYIVTyL998JqX1Jk+HV29kFCS/dw12mz3Bkb40r gl36hDTmpJbr9eayYblCg+qnktMJtvpLZXxYrzkJ9XpRwvTF4F40++DCIGzYn0Djl+jz CyfUROVn71JWE2C4EvQEhbedgoPQOtyleGm2yGwfez/9hRDQkqElgI0eigJFR0SBSQKk ylUjX82WIXwvxj4+xmYyACcimcCkqErNxBRSXiAw98zRVP1y2Oq2WRqLyDiY+oPY1Pe3 nhaQ== X-Gm-Message-State: AOAM531PZ11TZDDH3+guoKsrRtBl5kY70imPsqj0pD5WW+zeQjGtGDoZ Ch6qCn12qwwdnoI9py/Muw1eD5WiPRpkOkhwGiqkEA== X-Google-Smtp-Source: ABdhPJz+YsQuTWGDv5FhM9jNvgRh0y7cnnRusA68v3c/NPpKcrfsWu7vneR/qJj+419K2ioLiHPQZQ9RL2x8qiXw0OY= X-Received: by 2002:a05:6e02:13f4:: with SMTP id w20mr2383899ilj.294.1592302455382; Tue, 16 Jun 2020 03:14:15 -0700 (PDT) MIME-Version: 1.0 References: <20200519085444.4562-1-m.bilal@emumba.com> In-Reply-To: From: Muhammad Bilal Date: Tue, 16 Jun 2020 15:14:03 +0500 Message-ID: To: Sunil Kumar Kori Cc: "declan.doherty@intel.com" , "tomasz.kantecki@intel.com" , Pavan Nikhilesh Bhagavatula , "dev@dpdk.org" , "jgrajcia@cisco.com" , "vipin.varghese@intel.com" Content-Type: text/plain; charset="UTF-8" Subject: Re: [dpdk-dev] [EXT] [PATCH 1/5] examples/l2fwd-event: free resources in case of error 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 Tue, Jun 16, 2020 at 10:53 AM Sunil Kumar Kori wrote: > > >-----Original Message----- > >From: Muhammad Bilal > >Sent: Monday, June 15, 2020 5:30 PM > >To: Sunil Kumar Kori > >Cc: declan.doherty@intel.com; tomasz.kantecki@intel.com; Pavan Nikhilesh > >Bhagavatula ; dev@dpdk.org; > >jgrajcia@cisco.com; vipin.varghese@intel.com > >Subject: Re: [EXT] [PATCH 1/5] examples/l2fwd-event: free resources in case > >of error > > > >On Wed, Jun 10, 2020 at 3:01 PM Sunil Kumar Kori > >wrote: > >> > >> >-----Original Message----- > >> >From: Muhammad Bilal > >> >Sent: Tuesday, June 2, 2020 5:57 PM > >> >To: Sunil Kumar Kori > >> >Cc: declan.doherty@intel.com; tomasz.kantecki@intel.com; Pavan > >> >Nikhilesh Bhagavatula ; dev@dpdk.org; > >> >jgrajcia@cisco.com; vipin.varghese@intel.com > >> >Subject: Re: [EXT] [PATCH 1/5] examples/l2fwd-event: free resources > >> >in case of error > >> > > >> >On Tue, May 19, 2020 at 2:35 PM Sunil Kumar Kori > >> >wrote: > >> >> > >> >> >-----Original Message----- > >> >> >From: Muhammad Bilal > >> >> >Sent: Tuesday, May 19, 2020 2:25 PM > >> >> >To: declan.doherty@intel.com; tomasz.kantecki@intel.com; Pavan > >> >> >Nikhilesh Bhagavatula ; Sunil Kumar Kori > >> >> > > >> >> >Cc: dev@dpdk.org; Muhammad Bilal > >> >> >Subject: [EXT] [PATCH 1/5] examples/l2fwd-event: free resources in > >> >> >case of error > >> >> > > >> >> >External Email > >> >> > > >> >> >------------------------------------------------------------------ > >> >> >--- > >> >> >- Freeing the resources and call rte_eal_cleanup in case of error > >> >> >exit. > >> >> >Signed-off-by: Muhammad Bilal > >> >> >--- > >> >> > examples/l2fwd-event/main.c | 43 > >> >> >++++++++++++++++++++++++++++++------- > >> >> > 1 file changed, 35 insertions(+), 8 deletions(-) > >> >> > > >> >> >diff --git a/examples/l2fwd-event/main.c > >> >> >b/examples/l2fwd-event/main.c index 9593ef11e..442a664e9 100644 > >> >> >--- a/examples/l2fwd-event/main.c > >> >> >+++ b/examples/l2fwd-event/main.c > >> >> >@@ -556,13 +556,26 @@ signal_handler(int signum) > >> >> > } > >> >> > } > >> >> > > >> >> >+static void > >> >> >+stop_and_close_eth_dev(uint16_t portid) { > >> >> >+ RTE_ETH_FOREACH_DEV(portid) { > >> >> >+ printf("Closing port %d...", portid); > >> >> >+ rte_eth_dev_stop(portid); > >> >> >+ rte_eth_dev_close(portid); > >> >> >+ printf(" Done\n"); > >> >> >+ } > >> >> >+ rte_eal_cleanup(); > >> >> >+} > >> >> >+ > >> >> > int > >> >> > main(int argc, char **argv) > >> >> > { > >> >> > struct l2fwd_resources *rsrc; > >> >> > uint16_t nb_ports_available = 0; > >> >> > uint32_t nb_ports_in_mask = 0; > >> >> >- uint16_t port_id, last_port; > >> >> >+ uint16_t port_id = 0; > >> >> >+ uint16_t last_port; > >> >> > uint32_t nb_mbufs; > >> >> > uint16_t nb_ports; > >> >> > int i, ret; > >> >> >@@ -581,20 +594,26 @@ main(int argc, char **argv) > >> >> > > >> >> > /* parse application arguments (after the EAL ones) */ > >> >> > ret = l2fwd_event_parse_args(argc, argv, rsrc); > >> >> >- if (ret < 0) > >> >> >+ if (ret < 0) { > >> >> >+ stop_and_close_eth_dev(port_id); > >> >> > rte_panic("Invalid L2FWD arguments\n"); > >> >> >+ } > >> >> > > >> >Yes you are right we should use rte_exit instead of rte_panic, as > >> >rte_exit internally calls rte_eal_cleanup function. > >> >> IMO, up to this point only eal_init is done so rte_eal_cleanup will > >> >> be > >> >sufficient for this. > >> >> Also another way to handle this, use rte_exit instead rte_panic. > >> >> rte_exit internally calls rte_eal_cleanup. Refer l2fwd. > >> >> > >> >> Also I think, it is better to release the relevant resources on error. > >> >Here I'm solving the problem reported in bugzilla id 437. The bug was > >> >that if we use --vdev=net_memif with l2fwd application (and with its > >> >other variants) then a socket is created by memif PMD, after > >> >rte_eal_init function has run successfully. And if an error occurs > >> >then the application exits without freeing the resources (socket). On > >> >running the application 2nd time we get an error of "socket already exists". > >> >As in the previous version of patch " > >> >https://urldefense.proofpoint.com/v2/url?u=http- > >> > >>3A__patches.dpdk.org_patch_70081_&d=DwIBaQ&c=nKjWec2b6R0mOyPaz7 > >x > >> > >>tfQ&r=dXeXaAMkP5COgn1zxHMyaF1_d9IIuq6vHQO6NrIPjaE&m=XKcRI2e7sM > >v > >> > >>Y0nGabnBQl_Q8meL03FXFAjeNGdCV91A&s=TKq1J0W3QbnkeuG4c63payDW > >c > >> >Pc4zTg4DumA95RVzwg&e= " it was recommended to clean the resources > >> >when an error occurs. > >> > > >> >Here only using rte_eal_cleanup() is not solving the problem as using > >> >memif PMD is creating a socket and it's only cleaned when > >> >rte_eth_dev_close(portid) function is called. so that's why using it > >> >along with rte_exit or rte_panic. > >> > >> > >> Understood but I am only thinking from user's perspective that user > >> didn't do rte_eth_dev_configure and related APIs so closing the device > >> using rte_eth_dev_close does not look good. > >> May be other's (eal and memif PMD owners) can suggest something better. > >> Please redirect this query to them for suggestions also. > >Maintainer also recommended to call rte_eth_dev_close() for this problem in > >bug No. 437 Here : https://urldefense.proofpoint.com/v2/url?u=https- > >3A__bugs.dpdk.org_show-5Fbug.cgi-3Fid-3D437- > >23c1&d=DwIBaQ&c=nKjWec2b6R0mOyPaz7xtfQ&r=dXeXaAMkP5COgn1zxHMy > >aF1_d9IIuq6vHQO6NrIPjaE&m=C6PYaZqbOP-IfP6yK- > >nkxpSLonovGalGzl9QJ03qBmU&s=SEiTU3- > >EhOV5r16Kk9ZbW67E_dzuHvls4Fn41WyfCN0&e= > > In mentioned link, I see that Vipin has also suggested to clean this resource in memif PMD > But did not found any reason to drop that idea. I am still in favor of this. You are right, Vipin initially recommended change in Memif PMD but later agreed upon cleaning the resources in applications. > Also as a query, I see that changes are only l2fwd and its variants. > Is this problem not applicable to other applications where memif is used ? This problem is applicable to other applications where memif is used. We are working on other applications as time permits, Some patches already submitted e.g: http://patches.dpdk.org/project/dpdk/list/?series=10458 > > >> > >> >> > >> >> > printf("MAC updating %s\n", rsrc->mac_updating ? "enabled" : > >> >> > "disabled"); > >> >> > > >> >> > nb_ports = rte_eth_dev_count_avail(); > >> >> >- if (nb_ports == 0) > >> >> >+ if (nb_ports == 0) { > >> >> >+ stop_and_close_eth_dev(port_id); > >> >> > rte_panic("No Ethernet ports - bye\n"); > >> >> >+ } > >> >> > > >> >> Same as above. > >> >> > >> >> > /* check port mask to possible port mask */ > >> >> >- if (rsrc->enabled_port_mask & ~((1 << nb_ports) - 1)) > >> >> >+ if (rsrc->enabled_port_mask & ~((1 << nb_ports) - 1)) { > >> >> >+ stop_and_close_eth_dev(port_id); > >> >> > rte_panic("Invalid portmask; possible (0x%x)\n", > >> >> > (1 << nb_ports) - 1); > >> >> >+ } > >> >> > > >> >> Same as above. > >> >> > >> >> > if (!rsrc->port_pairs) { > >> >> > last_port = 0; > >> >> >@@ -621,8 +640,10 @@ main(int argc, char **argv) > >> >> > rsrc->dst_ports[last_port] = last_port; > >> >> > } > >> >> > } else { > >> >> >- if (check_port_pair_config(rsrc) < 0) > >> >> >+ if (check_port_pair_config(rsrc) < 0) { > >> >> >+ stop_and_close_eth_dev(port_id); > >> >> > rte_panic("Invalid port pair config\n"); > >> >> >+ } > >> >> > } > >> >> > > >> >> > nb_mbufs = RTE_MAX(nb_ports * (RTE_TEST_RX_DESC_DEFAULT + > >> >@@ > >> >> >-634,12 +655,16 @@ main(int argc, char **argv) > >> >> > rsrc->pktmbuf_pool = rte_pktmbuf_pool_create("mbuf_pool", > >> >> > nb_mbufs, MEMPOOL_CACHE_SIZE, 0, > >> >> > RTE_MBUF_DEFAULT_BUF_SIZE, rte_socket_id()); > >> >> >- if (rsrc->pktmbuf_pool == NULL) > >> >> >+ if (rsrc->pktmbuf_pool == NULL) { > >> >> >+ stop_and_close_eth_dev(port_id); > >> >> > rte_panic("Cannot init mbuf pool\n"); > >> >> >+ } > >> >> > > >> >> > nb_ports_available = l2fwd_event_init_ports(rsrc); > >> >> >- if (!nb_ports_available) > >> >> >+ if (!nb_ports_available) { > >> >> >+ stop_and_close_eth_dev(port_id); > >> >> > rte_panic("All available ports are disabled. Please > >> >> >set portmask.\n"); > >> >> >+ } > >> >> > > >> >> > /* Configure eventdev parameters if required */ > >> >> > if (rsrc->event_mode) > >> >> >@@ -659,9 +684,11 @@ main(int argc, char **argv) > >> >> > continue; > >> >> > > >> >> > ret = rte_eth_dev_start(port_id); > >> >> >- if (ret < 0) > >> >> >+ if (ret < 0) { > >> >> >+ stop_and_close_eth_dev(port_id); > >> >> > rte_panic("rte_eth_dev_start:err=%d, > >> >> >port=%u\n", ret, > >> >> > port_id); > >> >> >+ } > >> >> > } > >> >> > > >> >> > if (rsrc->event_mode) > >> >> >-- > >> >> >2.17.1 > >> >>