From mboxrd@z Thu Jan 1 00:00:00 1970
Return-Path: <konstantin.ananyev@intel.com>
Received: from mga01.intel.com (mga01.intel.com [192.55.52.88])
by dpdk.org (Postfix) with ESMTP id 3E2EB71
for <dev@dpdk.org>; Thu, 2 Oct 2014 10:57:50 +0200 (CEST)
Received: from fmsmga001.fm.intel.com ([10.253.24.23])
by fmsmga101.fm.intel.com with ESMTP; 02 Oct 2014 02:04:38 -0700
X-ExtLoop1: 1
X-IronPort-AV: E=Sophos;i="5.04,638,1406617200"; d="scan'208";a="599715179"
Received: from irsmsx103.ger.corp.intel.com ([163.33.3.157])
by fmsmga001.fm.intel.com with ESMTP; 02 Oct 2014 02:04:33 -0700
Received: from irsmsx107.ger.corp.intel.com (163.33.3.99) by
IRSMSX103.ger.corp.intel.com (163.33.3.157) with Microsoft SMTP Server (TLS)
id 14.3.195.1; Thu, 2 Oct 2014 10:04:28 +0100
Received: from irsmsx105.ger.corp.intel.com ([169.254.7.174]) by
IRSMSX107.ger.corp.intel.com ([169.254.10.68]) with mapi id 14.03.0195.001;
Thu, 2 Oct 2014 10:04:28 +0100
From: "Ananyev, Konstantin" <konstantin.ananyev@intel.com>
To: "Richardson, Bruce" <bruce.richardson@intel.com>, Neil Horman
<nhorman@tuxdriver.com>
Thread-Topic: [dpdk-dev] [PATCH v3] distributor_app: new sample app
Thread-Index: AQHP3Jr3nDHv7Xjdv0yMyvh7Wjw+5JwZexeAgAAMNwCAABbFAIABpRAAgAACmwCAAAuBAIABLxxg
Date: Thu, 2 Oct 2014 09:04:27 +0000
Message-ID: <2601191342CEEE43887BDE71AB977258213904F6@IRSMSX105.ger.corp.intel.com>
References: <1411568210-2555-1-git-send-email-reshma.pattan@intel.com>
<1412073577-12248-1-git-send-email-reshma.pattan@intel.com>
<20140930113445.GB2193@hmsreliant.think-freely.org>
<20140930121828.GA9312@BRICHA3-MOBL>
<20140930133958.GG2193@hmsreliant.think-freely.org>
<3AEA2BF9852C6F48A459DA490692831FE20FB6@IRSMSX109.ger.corp.intel.com>
<20141001145620.GB24028@localhost.localdomain>
<20141001153730.GA9292@BRICHA3-MOBL>
In-Reply-To: <20141001153730.GA9292@BRICHA3-MOBL>
Accept-Language: en-IE, en-US
Content-Language: en-US
X-MS-Has-Attach:
X-MS-TNEF-Correlator:
x-originating-ip: [163.33.239.180]
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
Cc: "dev@dpdk.org" <dev@dpdk.org>
Subject: Re: [dpdk-dev] [PATCH v3] distributor_app: new sample app
X-BeenThere: dev@dpdk.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: patches and discussions about DPDK <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>
X-List-Received-Date: Thu, 02 Oct 2014 08:57:51 -0000
> -----Original Message-----
> From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Bruce Richardson
> Sent: Wednesday, October 01, 2014 4:38 PM
> To: Neil Horman
> Cc: dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH v3] distributor_app: new sample app
>=20
> On Wed, Oct 01, 2014 at 10:56:20AM -0400, Neil Horman wrote:
> > On Wed, Oct 01, 2014 at 02:47:00PM +0000, Pattan, Reshma wrote:
> > >
> > >
> > > > -----Original Message-----
> > > > From: Neil Horman [mailto:nhorman@tuxdriver.com]
> > > > Sent: Tuesday, September 30, 2014 2:40 PM
> > > > To: Richardson, Bruce
> > > > Cc: Pattan, Reshma; dev@dpdk.org
> > > > Subject: Re: [dpdk-dev] [PATCH v3] distributor_app: new sample app
> > > >
> > > > On Tue, Sep 30, 2014 at 01:18:28PM +0100, Bruce Richardson wrote:
> > > > > On Tue, Sep 30, 2014 at 07:34:45AM -0400, Neil Horman wrote:
> > > > > > On Tue, Sep 30, 2014 at 11:39:37AM +0100, reshmapa wrote:
> > > > > > > From: Reshma Pattan <reshma.pattan@intel.com>
> > > > > > >
> > > > > > > A new sample app that shows the usage of the distributor libr=
ary.
> > > > > > > This app works as follows:
> > > > > > >
> > > > > > > * An RX thread runs which pulls packets from each ethernet po=
rt in turn
> > > > > > > and passes those packets to worker using a distributor comp=
onent.
> > > > > > > * The workers take the packets in turn, and determine the out=
put port
> > > > > > > for those packets using basic l2forwarding doing an xor on =
the source
> > > > > > > port id.
> > > > > > > * The RX thread takes the returned packets from the workers a=
nd enqueue
> > > > > > > those packets into an rte_ring structure.
> > > > > > > * A TX thread pulls the packets off the rte_ring structure an=
d then
> > > > > > > sends each packet out the output port specified previously =
by
> > > > > > > the worker
> > > > > > > * Command-line option support provided only for portmask.
> > > > > > >
> > > > > > > Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
> > > > > > > Signed-off-by: Reshma Pattan <reshma.pattan@intel.com>
> > > > > > > ---
> > > > > > > examples/Makefile | 1 +
> > > > > > > examples/distributor_app/Makefile | 57 ++++
> > > > > > > examples/distributor_app/main.c | 600
> > > > +++++++++++++++++++++++++++++++++++++
> > > > > > > examples/distributor_app/main.h | 46 +++
> > > > > > > 4 files changed, 704 insertions(+), 0 deletions(-) create m=
ode
> > > > > > > 100644 examples/distributor_app/Makefile create mode 100644
> > > > > > > examples/distributor_app/main.c create mode 100644
> > > > > > > examples/distributor_app/main.h
> > > > > > >
> > > > > > > diff --git a/examples/Makefile b/examples/Makefile index
> > > > > > > 6245f83..2ba82b0 100644
> > > > > > > --- a/examples/Makefile
> > > > > > > +++ b/examples/Makefile
> > > > > > > @@ -66,5 +66,6 @@ DIRS-y +=3D vhost
> > > > > > > DIRS-$(CONFIG_RTE_LIBRTE_XEN_DOM0) +=3D vhost_xen DIRS-y +=
=3D vmdq
> > > > > > > DIRS-y +=3D vmdq_dcb
> > > > > > > +DIRS-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR) +=3D distributor_app
> > > > > > >
> > > > > > > include $(RTE_SDK)/mk/rte.extsubdir.mk diff --git
> > > > > > > a/examples/distributor_app/Makefile
> > > > > > > b/examples/distributor_app/Makefile
> > > > > > > new file mode 100644
> > > > > > > index 0000000..6a5bada
> > > > > > > --- /dev/null
> > > > > > > +++ b/examples/distributor_app/Makefile
> > > > > > > @@ -0,0 +1,57 @@
> > > > > > > +# BSD LICENSE
> > > > > > > +#
> > > > > > > +# Copyright(c) 2010-2014 Intel Corporation. All rights res=
erved.
> > > > > > > +# All rights reserved.
> > > > > > > +#
> > > > > > > +# Redistribution and use in source and binary forms, with =
or without
> > > > > > > +# modification, are permitted provided that the following =
conditions
> > > > > > > +# are met:
> > > > > > > +#
> > > > > > > +# * Redistributions of source code must retain the above=
copyright
> > > > > > > +# notice, this list of conditions and the following di=
sclaimer.
> > > > > > > +# * Redistributions in binary form must reproduce the ab=
ove copyright
> > > > > > > +# notice, this list of conditions and the following di=
sclaimer in
> > > > > > > +# the documentation and/or other materials provided wi=
th the
> > > > > > > +# distribution.
> > > > > > > +# * Neither the name of Intel Corporation nor the names =
of its
> > > > > > > +# contributors may be used to endorse or promote produ=
cts derived
> > > > > > > +# from this software without specific prior written pe=
rmission.
> > > > > > > +#
> > > > > > > +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND
> > > > CONTRIBUTORS
> > > > > > > +# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING=
, BUT
> > > > NOT
> > > > > > > +# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AN=
D
> > > > FITNESS FOR
> > > > > > > +# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL T=
HE
> > > > COPYRIGHT
> > > > > > > +# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT=
,
> > > > INCIDENTAL,
> > > > > > > +# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
> > > > BUT NOT
> > > > > > > +# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
> > > > LOSS OF USE,
> > > > > > > +# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUS=
ED
> > > > AND ON ANY
> > > > > > > +# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILI=
TY, OR
> > > > TORT
> > > > > > > +# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY O=
UT
> > > > OF THE USE
> > > > > > > +# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF =
SUCH
> > > > DAMAGE.
> > > > > > > +
> > > > > > > +ifeq ($(RTE_SDK),)
> > > > > > > +$(error "Please define RTE_SDK environment variable") endif
> > > > > > > +
> > > > > > > +# Default target, can be overriden by command line or enviro=
nment
> > > > > > > +RTE_TARGET ?=3D x86_64-native-linuxapp-gcc
> > > > > > > +
> > > > > > > +include $(RTE_SDK)/mk/rte.vars.mk
> > > > > > > +
> > > > > > > +# binary name
> > > > > > > +APP =3D distributor_app
> > > > > > > +
> > > > > > > +# all source are stored in SRCS-y SRCS-y :=3D main.c
> > > > > > > +
> > > > > > > +CFLAGS +=3D $(WERROR_FLAGS)
> > > > > > > +
> > > > > > > +# workaround for a gcc bug with noreturn attribute #
> > > > > > > +http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D12603
> > > > > > > +ifeq ($(CONFIG_RTE_TOOLCHAIN_GCC),y) CFLAGS_main.o +=3D
> > > > > > > +-Wno-return-type endif
> > > > > > > +
> > > > > > > +EXTRA_CFLAGS +=3D -O3 -Wfatal-errors
> > > > > > > +
> > > > > > > +include $(RTE_SDK)/mk/rte.extapp.mk
> > > > > > > diff --git a/examples/distributor_app/main.c
> > > > > > > b/examples/distributor_app/main.c new file mode 100644 index
> > > > > > > 0000000..f555d93
> > > > > > > --- /dev/null
> > > > > > > +++ b/examples/distributor_app/main.c
> > > > > > > @@ -0,0 +1,600 @@
> > > > > > > +/*-
> > > > > > > + * BSD LICENSE
> > > > > > > + *
> > > > > > > + * Copyright(c) 2010-2014 Intel Corporation. All rights re=
served.
> > > > > > > + * All rights reserved.
> > > > > > > + *
> > > > > > > + * Redistribution and use in source and binary forms, with=
or without
> > > > > > > + * modification, are permitted provided that the following=
conditions
> > > > > > > + * are met:
> > > > > > > + *
> > > > > > > + * * Redistributions of source code must retain the abov=
e copyright
> > > > > > > + * notice, this list of conditions and the following d=
isclaimer.
> > > > > > > + * * Redistributions in binary form must reproduce the a=
bove copyright
> > > > > > > + * notice, this list of conditions and the following d=
isclaimer in
> > > > > > > + * the documentation and/or other materials provided w=
ith the
> > > > > > > + * distribution.
> > > > > > > + * * Neither the name of Intel Corporation nor the names=
of its
> > > > > > > + * contributors may be used to endorse or promote prod=
ucts derived
> > > > > > > + * from this software without specific prior written p=
ermission.
> > > > > > > + *
> > > > > > > + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND
> > > > CONTRIBUTORS
> > > > > > > + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDIN=
G,
> > > > BUT NOT
> > > > > > > + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY A=
ND
> > > > FITNESS FOR
> > > > > > > + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL =
THE
> > > > COPYRIGHT
> > > > > > > + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIREC=
T,
> > > > INCIDENTAL,
> > > > > > > + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING=
,
> > > > BUT NOT
> > > > > > > + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES=
;
> > > > LOSS OF USE,
> > > > > > > + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAU=
SED
> > > > AND ON ANY
> > > > > > > + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABIL=
ITY, OR
> > > > TORT
> > > > > > > + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY =
OUT
> > > > OF THE USE
> > > > > > > + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF=
SUCH
> > > > DAMAGE.
> > > > > > > + */
> > > > > > > +
> > > > > > > +#include <stdint.h>
> > > > > > > +#include <inttypes.h>
> > > > > > > +#include <unistd.h>
> > > > > > > +#include <signal.h>
> > > > > > > +#include <getopt.h>
> > > > > > > +
> > > > > > > +#include <rte_eal.h>
> > > > > > > +#include <rte_ethdev.h>
> > > > > > > +#include <rte_cycles.h>
> > > > > > > +#include <rte_malloc.h>
> > > > > > > +#include <rte_debug.h>
> > > > > > > +#include <rte_distributor.h>
> > > > > > > +
> > > > > > > +#include "main.h"
> > > > > > > +
> > > > > > > +#define RX_RING_SIZE 256
> > > > > > > +#define RX_FREE_THRESH 32
> > > > > > > +#define RX_PTHRESH 8
> > > > > > > +#define RX_HTHRESH 8
> > > > > > > +#define RX_WTHRESH 0
> > > > > > > +
> > > > > > > +#define TX_RING_SIZE 512
> > > > > > > +#define TX_FREE_THRESH 32
> > > > > > > +#define TX_PTHRESH 32
> > > > > > > +#define TX_HTHRESH 0
> > > > > > > +#define TX_WTHRESH 0
> > > > > > > +#define TX_RSBIT_THRESH 32
> > > > > > > +#define TX_Q_FLAGS (ETH_TXQ_FLAGS_NOMULTSEGS |
> > > > ETH_TXQ_FLAGS_NOVLANOFFL |\
> > > > > > > + ETH_TXQ_FLAGS_NOXSUMSCTP | ETH_TXQ_FLAGS_NOXSUMUDP | \
> > > > > > > + ETH_TXQ_FLAGS_NOXSUMTCP)
> > > > > > > +
> > > > > > > +#define NUM_MBUFS ((64*1024)-1)
> > > > > > > +#define MBUF_SIZE (2048 + sizeof(struct rte_mbuf) +
> > > > > > > +RTE_PKTMBUF_HEADROOM) #define MBUF_CACHE_SIZE 250 #define
> > > > > > > +BURST_SIZE 32 #define RTE_RING_SZ 1024
> > > > > > > +
> > > > > > > +/* uncommnet below line to enable debug logs */
> > > > > > > +/* #define DEBUG */
> > > > > > > +
> > > > > > > +#ifdef DEBUG
> > > > > > > +#define LOG_LEVEL RTE_LOG_DEBUG
> > > > > > > +#define LOG_DEBUG(log_type, fmt, args...) do { \
> > > > > > > + RTE_LOG(DEBUG, log_type, fmt, ##args) \
> > > > > > > +} while (0)
> > > > > > > +#else
> > > > > > > +#define LOG_LEVEL RTE_LOG_INFO
> > > > > > > +#define LOG_DEBUG(log_type, fmt, args...) do {} while (0) #e=
ndif
> > > > > > > +
> > > > > > > +#define RTE_LOGTYPE_DISTRAPP RTE_LOGTYPE_USER1
> > > > > > > +
> > > > > > > +/* mask of enabled ports */
> > > > > > > +static uint32_t enabled_port_mask =3D 0;
> > > > > > > +
> > > > > > > +static volatile struct app_stats {
> > > > > > > + struct {
> > > > > > > + uint64_t rx_pkts;
> > > > > > > + uint64_t returned_pkts;
> > > > > > > + uint64_t enqueued_pkts;
> > > > > > > + } rx __rte_cache_aligned;
> > > > > > > +
> > > > > > > + struct {
> > > > > > > + uint64_t dequeue_pkts;
> > > > > > > + uint64_t tx_pkts;
> > > > > > > + } tx __rte_cache_aligned;
> > > > > > > +} app_stats;
> > > > > > > +
> > > > > > > +static const struct rte_eth_conf port_conf_default =3D {
> > > > > > > + .rxmode =3D {
> > > > > > > + .mq_mode =3D ETH_MQ_RX_RSS,
> > > > > > > + .max_rx_pkt_len =3D ETHER_MAX_LEN,
> > > > > > > + .split_hdr_size =3D 0,
> > > > > > > + .header_split =3D 0, /**< Header Split disabled */
> > > > > > > + .hw_ip_checksum =3D 0, /**< IP checksum offload enabled */
> > > > > > > + .hw_vlan_filter =3D 0, /**< VLAN filtering disabled */
> > > > > > > + .jumbo_frame =3D 0, /**< Jumbo Frame Support disabled *=
/
> > > > > > > + .hw_strip_crc =3D 0, /**< CRC stripped by hardware */
> > > > > > > + },
> > > > > > > + .txmode =3D {
> > > > > > > + .mq_mode =3D ETH_MQ_TX_NONE,
> > > > > > > + },
> > > > > > > + .lpbk_mode =3D 0,
> > > > > > > + .rx_adv_conf =3D {
> > > > > > > + .rss_conf =3D {
> > > > > > > + .rss_hf =3D ETH_RSS_IPV4 | ETH_RSS_IPV6 |
> > > > > > > + ETH_RSS_IPV4_TCP |
> > > > ETH_RSS_IPV4_UDP |
> > > > > > > + ETH_RSS_IPV6_TCP |
> > > > ETH_RSS_IPV6_UDP,
> > > > > > > + }
> > > > > > > + },
> > > > > > > +};
> > > > > > > +
> > > > > > > +static const struct rte_eth_rxconf rx_conf_default =3D {
> > > > > > > + .rx_thresh =3D {
> > > > > > > + .pthresh =3D RX_PTHRESH,
> > > > > > > + .hthresh =3D RX_HTHRESH,
> > > > > > > + .wthresh =3D RX_WTHRESH,
> > > > > > > + },
> > > > > > > + .rx_free_thresh =3D RX_FREE_THRESH,
> > > > > > > + .rx_drop_en =3D 0,
> > > > > > > +};
> > > > > > > +
> > > > > > > +static const struct rte_eth_txconf tx_conf_default =3D {
> > > > > > > + .tx_thresh =3D {
> > > > > > > + .pthresh =3D TX_PTHRESH,
> > > > > > > + .hthresh =3D TX_HTHRESH,
> > > > > > > + .wthresh =3D TX_WTHRESH,
> > > > > > > + },
> > > > > > > + .tx_free_thresh =3D TX_FREE_THRESH,
> > > > > > > + .tx_rs_thresh =3D TX_RSBIT_THRESH,
> > > > > > > + .txq_flags =3D TX_Q_FLAGS
> > > > > > > +
> > > > > > > +};
> > > > > > > +
> > > > > > > +struct output_buffer {
> > > > > > > + unsigned count;
> > > > > > > + struct rte_mbuf *mbufs[BURST_SIZE]; };
> > > > > > > +
> > > > > > > +/*
> > > > > > > + * Initialises a given port using global settings and with t=
he rx
> > > > > > > +buffers
> > > > > > > + * coming from the mbuf_pool passed as parameter */ static
> > > > > > > +inline int port_init(uint8_t port, struct rte_mempool *mbuf_=
pool)
> > > > > > > +{
> > > > > > > + struct rte_eth_conf port_conf =3D port_conf_default;
> > > > > > > + const uint16_t rxRings =3D 1, txRings =3D rte_lcore_count()=
- 1;
> > > > > > > + int retval;
> > > > > > > + uint16_t q;
> > > > > > > +
> > > > > > > + if (port >=3D rte_eth_dev_count())
> > > > > > > + return -1;
> > > > > > > +
> > > > > > > + retval =3D rte_eth_dev_configure(port, rxRings, txRings, &p=
ort_conf);
> > > > > > > + if (retval !=3D 0)
> > > > > > > + return retval;
> > > > > > > +
> > > > > > > + for (q =3D 0; q < rxRings; q++) {
> > > > > > > + retval =3D rte_eth_rx_queue_setup(port, q, RX_RING_SIZE,
> > > > > > > + rte_eth_dev_socket_id(port),
> > > > > > > + &rx_conf_default, mbuf_pool);
> > > > > > > + if (retval < 0)
> > > > > > > + return retval;
> > > > > > > + }
> > > > > > > +
> > > > > > > + for (q =3D 0; q < txRings; q++) {
> > > > > > > + retval =3D rte_eth_tx_queue_setup(port, q, TX_RING_SIZE,
> > > > > > > + rte_eth_dev_socket_id(port),
> > > > > > > + &tx_conf_default);
> > > > > > > + if (retval < 0)
> > > > > > > + return retval;
> > > > > > > + }
> > > > > > > +
> > > > > > > + retval =3D rte_eth_dev_start(port);
> > > > > > > + if (retval < 0)
> > > > > > > + return retval;
> > > > > > > +
> > > > > > > + struct rte_eth_link link;
> > > > > > > + rte_eth_link_get_nowait(port, &link);
> > > > > > > + if (!link.link_status) {
> > > > > > > + sleep(1);
> > > > > > > + rte_eth_link_get_nowait(port, &link);
> > > > > > > + }
> > > > > > > +
> > > > > > > + if (!link.link_status) {
> > > > > > > + printf("Link down on port %"PRIu8"\n", port);
> > > > > > > + return 0;
> > > > > > > + }
> > > > > > > +
> > > > > > > + struct ether_addr addr;
> > > > > > > + rte_eth_macaddr_get(port, &addr);
> > > > > > > + printf("Port %u MAC: %02"PRIx8" %02"PRIx8" %02"PRIx8
> > > > > > > + " %02"PRIx8" %02"PRIx8" %02"PRIx8"\n",
> > > > > > > + (unsigned)port,
> > > > > > > + addr.addr_bytes[0], addr.addr_bytes[1],
> > > > > > > + addr.addr_bytes[2], addr.addr_bytes[3],
> > > > > > > + addr.addr_bytes[4], addr.addr_bytes[5]);
> > > > > > > +
> > > > > > > + rte_eth_promiscuous_enable(port);
> > > > > > > +
> > > > > > > + return 0;
> > > > > > > +}
> > > > > > > +
> > > > > > > +struct lcore_params {
> > > > > > > + unsigned worker_id;
> > > > > > > + struct rte_distributor *d;
> > > > > > > + struct rte_ring *r;
> > > > > > > +};
> > > > > > > +
> > > > > > > +static __attribute__((noreturn)) void lcore_rx(struct
> > > > > > > +lcore_params *p) {
> > > > > > > + struct rte_distributor *d =3D p->d;
> > > > > > > + struct rte_ring *r =3D p->r;
> > > > > > > + const uint8_t nb_ports =3D rte_eth_dev_count();
> > > > > > > + const int socket_id =3D rte_socket_id();
> > > > > > > + uint8_t port;
> > > > > > > +
> > > > > > > + for (port =3D 0; port < nb_ports; port++) {
> > > > > > > + /* skip ports that are not enabled */
> > > > > > > + if ((enabled_port_mask & (1 << port)) =3D=3D 0)
> > > > > > > + continue;
> > > > > > > +
> > > > > > > + if (rte_eth_dev_socket_id(port) > 0 &&
> > > > > > > + rte_eth_dev_socket_id(port) !=3D socket_id)
> > > > > > > + printf("WARNING, port %u is on remote NUMA node to
> > > > "
> > > > > > > + "RX thread.\n\tPerformance will not "
> > > > > > > + "be optimal.\n", port);
> > > > > > > + }
> > > > > > > +
> > > > > > > + printf("\nCore %u doing packet RX.\n", rte_lcore_id());
> > > > > > > + port =3D 0;
> > > > > > > + for (;;) {
> > > > > > > + /* skip ports that are not enabled */
> > > > > > > + if ((enabled_port_mask & (1 << port)) =3D=3D 0) {
> > > > > > > + if (++port =3D=3D nb_ports)
> > > > > > > + port =3D 0;
> > > > > > > + continue;
> > > > > > > + }
> > > > > > > + struct rte_mbuf *bufs[BURST_SIZE*2];
> > > > > > > + const uint16_t nb_rx =3D rte_eth_rx_burst(port, 0, bufs,
> > > > > > > + BURST_SIZE);
> > > > > > > + app_stats.rx.rx_pkts +=3D nb_rx;
> > > > > > > +
> > > > > > > + rte_distributor_process(d, bufs, nb_rx);
> > > > > > > + const uint16_t nb_ret =3D rte_distributor_returned_pkts(d,
> > > > > > > + bufs, BURST_SIZE*2);
> > > > > > > + app_stats.rx.returned_pkts +=3D nb_ret;
> > > > > > > + if (unlikely(nb_ret =3D=3D 0))
> > > > > > > + continue;
> > > > > > > +
> > > > > > > + uint16_t sent =3D rte_ring_enqueue_burst(r, (void *)bufs, =
nb_ret);
> > > > > > > + app_stats.rx.enqueued_pkts +=3D sent;
> > > > > > > + if (unlikely(sent < nb_ret)) {
> > > > > > > + LOG_DEBUG(DISTRAPP, "%s:Packet loss due to full
> > > > ring\n", __func__);
> > > > > > > + while (sent < nb_ret)
> > > > > > > + rte_pktmbuf_free(bufs[sent++]);
> > > > > > > + }
> > > > > > > + if (++port =3D=3D nb_ports)
> > > > > > > + port =3D 0;
> > > > > > > + }
> > > > > > > +}
> > > > > > > +
> > > > > > > +static inline void
> > > > > > > +flush_one_port(struct output_buffer *outbuf, uint8_t outp) {
> > > > > > > + unsigned nb_tx =3D rte_eth_tx_burst(outp, 0, outbuf->mbufs,
> > > > > > > + outbuf->count);
> > > > > > > + app_stats.tx.tx_pkts +=3D nb_tx;
> > > > > > > +
> > > > > > > + if (unlikely(nb_tx < outbuf->count)) {
> > > > > > > + LOG_DEBUG(DISTRAPP, "%s:Packet loss with tx_burst\n",
> > > > __func__);
> > > > > > > + do {
> > > > > > > + rte_pktmbuf_free(outbuf->mbufs[nb_tx]);
> > > > > > > + } while (++nb_tx < outbuf->count);
> > > > > > > + }
> > > > > > > + outbuf->count =3D 0;
> > > > > > > +}
> > > > > > > +
> > > > > > > +static inline void
> > > > > > > +flush_all_ports(struct output_buffer *tx_buffers, uint8_t
> > > > > > > +nb_ports) {
> > > > > > > + uint8_t outp;
> > > > > > > + for (outp =3D 0; outp < nb_ports; outp++) {
> > > > > > > + /* skip ports that are not enabled */
> > > > > > > + if ((enabled_port_mask & (1 << outp)) =3D=3D 0)
> > > > > > > + continue;
> > > > > > > +
> > > > > > > + if (tx_buffers[outp].count =3D=3D 0)
> > > > > > > + continue;
> > > > > > > +
> > > > > > > + flush_one_port(&tx_buffers[outp], outp);
> > > > > > > + }
> > > > > > > +}
> > > > > > > +
> > > > > > > +static __attribute__((noreturn)) void lcore_tx(struct rte_ri=
ng
> > > > > > > +*in_r) {
> > > > > > > + static struct output_buffer tx_buffers[RTE_MAX_ETHPORTS];
> > > > > > > + const uint8_t nb_ports =3D rte_eth_dev_count();
> > > > > > > + const int socket_id =3D rte_socket_id();
> > > > > > > + uint8_t port;
> > > > > > > +
> > > > > > > + for (port =3D 0; port < nb_ports; port++) {
> > > > > > > + /* skip ports that are not enabled */
> > > > > > > + if ((enabled_port_mask & (1 << port)) =3D=3D 0)
> > > > > > > + continue;
> > > > > > > +
> > > > > > > + if (rte_eth_dev_socket_id(port) > 0 &&
> > > > > > > + rte_eth_dev_socket_id(port) !=3D socket_id)
> > > > > > > + printf("WARNING, port %u is on remote NUMA node to
> > > > "
> > > > > > > + "TX thread.\n\tPerformance will not "
> > > > > > > + "be optimal.\n", port);
> > > > > > > + }
> > > > > > > +
> > > > > > > + printf("\nCore %u doing packet TX.\n", rte_lcore_id());
> > > > > > > + for (;;) {
> > > > > > > + for (port =3D 0; port < nb_ports; port++) {
> > > > > > > + /* skip ports that are not enabled */
> > > > > > > + if ((enabled_port_mask & (1 << port)) =3D=3D 0)
> > > > > > > + continue;
> > > > > > > +
> > > > > > > + struct rte_mbuf *bufs[BURST_SIZE];
> > > > > > > + const uint16_t nb_rx =3D rte_ring_dequeue_burst(in_r,
> > > > > > > + (void *)bufs, BURST_SIZE);
> > > > > > > + app_stats.tx.dequeue_pkts +=3D nb_rx;
> > > > > > > +
> > > > > > > + /* if we get no traffic, flush anything we have */
> > > > > > > + if (unlikely(nb_rx =3D=3D 0)) {
> > > > > > > + flush_all_ports(tx_buffers, nb_ports);
> > > > > > > + continue;
> > > > > > > + }
> > > > > > > +
> > > > > > > + /* for traffic we receive, queue it up for transmit */
> > > > > > > + uint16_t i;
> > > > > > > + _mm_prefetch(bufs[0], 0);
> > > > > > > + _mm_prefetch(bufs[1], 0);
> > > > > > > + _mm_prefetch(bufs[2], 0);
> > > > > > > + for (i =3D 0; i < nb_rx; i++) {
> > > > > > > + struct output_buffer *outbuf;
> > > > > > > + uint8_t outp;
> > > > > > > + _mm_prefetch(bufs[i + 3], 0);
> > > > > > > + /* workers should update in_port to hold the
> > > > > > > + * output port value */
> > > > > > > + outp =3D bufs[i]->port;
> > > > > > > + /* skip ports that are not enabled */
> > > > > > > + if ((enabled_port_mask & (1 << outp)) =3D=3D 0)
> > > > > > > + continue;
> > > > > > > +
> > > > > > > + outbuf =3D &tx_buffers[outp];
> > > > > > > + outbuf->mbufs[outbuf->count++] =3D bufs[i];
> > > > > > > + if (outbuf->count =3D=3D BURST_SIZE)
> > > > > > > + flush_one_port(outbuf, outp);
> > > > > > > + }
> > > > > > > + }
> > > > > > > + }
> > > > > > > +}
> > > > > > > +
> > > > > > > +
> > > > > > > +static __attribute__((noreturn)) void lcore_worker(struct
> > > > > > > +lcore_params *p) {
> > > > > > > + struct rte_distributor *d =3D p->d;
> > > > > > > + const unsigned id =3D p->worker_id;
> > > > > > > + /* for single port, xor_val will be zero so we won't modify=
the output
> > > > > > > + * port, otherwise we send traffic from 0 to 1, 2 to 3, and=
vice versa
> > > > > > > + */
> > > > > > > + const unsigned xor_val =3D (rte_eth_dev_count() > 1);
> > > > > > > + struct rte_mbuf *buf =3D NULL;
> > > > > > > +
> > > > > > > + printf("\nCore %u acting as worker core.\n", rte_lcore_id()=
);
> > > > > > > + for (;;) {
> > > > > > > + buf =3D rte_distributor_get_pkt(d, id, buf);
> > > > > > > + buf->port ^=3D xor_val;
> > > > > > > + }
> > > > > > > +}
> > > > > > > +
> > > > > > > +static void
> > > > > > > +int_handler(int sig_num)
> > > > > > > +{
> > > > > > > + struct rte_eth_stats eth_stats;
> > > > > > > + unsigned i;
> > > > > > > +
> > > > > > > + printf("Exiting on signal %d\n", sig_num);
> > > > > > > +
> > > > > > > + printf("\nRX thread stats:\n");
> > > > > > > + printf(" - Received: %"PRIu64"\n", app_stats.rx.rx_pkts)=
;
> > > > > > > + printf(" - Processed: %"PRIu64"\n", app_stats.rx.returned=
_pkts);
> > > > > > > + printf(" - Enqueued: %"PRIu64"\n", app_stats.rx.enqueued=
_pkts);
> > > > > > > +
> > > > > > > + printf("\nTX thread stats:\n");
> > > > > > > + printf(" - Dequeued: %"PRIu64"\n", app_stats.tx.dequeue_=
pkts);
> > > > > > > + printf(" - Transmitted: %"PRIu64"\n", app_stats.tx.tx_pkts)=
;
> > > > > > > +
> > > > > > > + for (i =3D 0; i < rte_eth_dev_count(); i++) {
> > > > > > > + rte_eth_stats_get(i, ð_stats);
> > > > > > > + printf("\nPort %u stats:\n", i);
> > > > > > > + printf(" - Pkts in: %"PRIu64"\n", eth_stats.ipackets);
> > > > > > > + printf(" - Pkts out: %"PRIu64"\n", eth_stats.opackets);
> > > > > > > + printf(" - In Errs: %"PRIu64"\n", eth_stats.ierrors);
> > > > > > > + printf(" - Out Errs: %"PRIu64"\n", eth_stats.oerrors);
> > > > > > > + printf(" - Mbuf Errs: %"PRIu64"\n", eth_stats.rx_nombuf);
> > > > > > > + }
> > > > > > > + exit(0);
> > > > > > rte_exit here? Also, this is a pretty ungraceful exit strategy=
as
> > > > > > all the threads you've created and memory you've allocated are =
just
> > > > forgotten here.
> > > > > > Given that dpdk mempools are shared, this has the potential to =
leak
> > > > > > lots of memory if other apps are using the dpdk at the same tim=
e
> > > > > > that you run this. You probably want to use the sigint handler=
to
> > > > > > raise a flag to the tx/rx threads to shutdown gracefully, and t=
hen free your
> > > > allocated memory and mempool.
> > > > > >
> > > > > > Neil
> > > > > >
> > > > >
> > > > > Unless the different processes are explicitly cooperating as
> > > > > primary/secondary, the mempools are not shared. I just don't see =
the
> > > > > need for this app to do more cleanup on ctrl-c signal, as it's no=
t
> > > > > intended to be a multiprocess app, and there is little that any
> > > > > secondary process could do to work with this app, except possibly=
some
> > > > > resource monitoring, which would be completely unaffected by it e=
xiting the
> > > > way it does.
> > > > >
> > > > Ah, ok, so we don't use a common shared pool between isolated proce=
sses
> > > > then, thats good. Still though, this is a sample application, I th=
ink its lazy
> > > > programming practice to illustrate to application developers that i=
ts generally ok
> > > > to exit programs without freeing your resources. Its about 20 line=
s of additional
> > > > code to change the sigint handler to flag an exit condition, and ha=
ve all the
> > > > other threads join on it.
> > > >
> > > > Neil
> > >
> > > 1)I had sent v5 patch which handles graceful shutdown of rx and tx th=
reads upon SIGINT
> > I see it and will take a look shortly, thanks.
> >
> > > 2)Worker thread graceful shutdown was not handled as of now as it nee=
ds some change in lcore_worker logic , which will be done
> in future enhancements.
> > Not sure I understand what you mean here. Can you elaborate?
> >
> > > 3)Freeing of mempool is also not handled , as the framework support i=
s not available.
> > Ew, I hadn't noticed that, freeing of mempools seems like something we =
should
> > implement.
> >
> > > 4)Cleaning of rx/tx queues not done, as it needs some extensive logic=
which we haven't planned as of now. Will check the
> possibility of doing it in future enhancements i.e in next version of=
sample application.
> > We can't just flush the queues after we shutdown the workers? I presum=
e a queue
> > flush operation exists, yes?
> > Neil
>=20
> Other than code hygiene, which does have some value in itself, I can't
> really see what the practical point of such cleanup would be.
>=20
> If traffic is going through the system, and the process is killed packets
> will be dropped, whatever we do, as packet reception will stop. If traffi=
c
> is not going through the system, then there are no packets in flight and
> therefore no relevant cleanup to be done. [And if the traffic is stopped
> just before shutting down the app, at a throughput rate of a couple of
> million packets per second, the app should be flushed of packets within t=
iny
> fractions of a second].
I think that in theory not resetting HW at process termination can cause a =
problem.
Something like that:
- DPDK app has HW RX/TX queues active with armed RXDs, but the link is idle=
(no packets are flying).=20
- DPDK app terminates abnormally.
- user deletes DPDK hugepages files.
- Hugepage memory that was used by DPDK for RXDs/data buffers are given to =
other app (or kernel).
- Packet arrives -a s HW is still active - it will do a write to the RXD an=
d data-buffer.=20
- Silent memory corruption.
Saying that, I don't think we can completely eliminate that problem from us=
er-space code -
as not process signals can be handled.=20
>>From other side - our whole concept is to move away from custom kernel modu=
les...
Though we probably can make it less possible to happen - create a terminati=
on handler,
that would try to reset all active HW.
And make sure it is called atexit and all catchable signals that cause proc=
ess termination.
But I don' t think it is a good idea to duplicate such code in each and eve=
ry sample app.
I think it should be in the librte_eal, and yes - it is a subject of a sepa=
rate patch/discussion :)
Konstantin
=20
> So overall, I just don't see complicated jumping through hoops for flushi=
ng
> and cleaning up things being worth the effort. These apps are designed to
> run in a forever loop, and, in the exception case, when killed the cleanu=
p
> done by kernel is sufficient.
>=20
> regards,
> /Bruce