From: Luca Boccassi <bluca@debian.org>
To: alangordondewar@gmail.com, cristian.dumitrescu@intel.com
Cc: dev@dpdk.org, Alan Dewar <alan.dewar@att.com>
Subject: Re: [dpdk-dev] [PATCH v2] test: new sched WRR unit-test
Date: Thu, 30 Nov 2017 10:48:03 +0000 [thread overview]
Message-ID: <1512038883.10466.1.camel@debian.org> (raw)
In-Reply-To: <1512032726-30807-1-git-send-email-alan.dewar@att.com>
On Thu, 2017-11-30 at 09:05 +0000, alangordondewar@gmail.com wrote:
> From: Alan Dewar <alan.dewar@att.com>
>
> New unit-test for the librte_sched WRR weighting code.
>
> With the standard 17.11 code, the first three sub-tests pass, but
> the last three fail due to bugs in the WRR weighting code.
>
> With v1 of the "sched: fix overflow errors in WRR weighting code"
> patch the first five sub-tests pass, and the last sub-test fails
> badly.
>
> With v2 of the "sched: fix overflow errors in WRR weighting code"
> patch the first five sub-tests pass, and the last sub-test is a very
> near miss (i.e. measured packets counts are one away from the
> expected
> counts).
>
> Signed-off-by: Alan Dewar <alan.dewar@att.com>
> ---
> v2 - add new 255-254-253-1 weightings sub-test
>
> test/test/Makefile | 1 +
> test/test/test_sched_wrr.c | 491
> +++++++++++++++++++++++++++++++++++++++++++++
> 2 files changed, 492 insertions(+)
> create mode 100644 test/test/test_sched_wrr.c
>
> diff --git a/test/test/Makefile b/test/test/Makefile
> index bb54c98..0ab0ed3 100644
> --- a/test/test/Makefile
> +++ b/test/test/Makefile
> @@ -173,6 +173,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_NET) += test_crc.c
> ifeq ($(CONFIG_RTE_LIBRTE_SCHED),y)
> SRCS-y += test_red.c
> SRCS-y += test_sched.c
> +SRCS-y += test_sched_wrr.c
> endif
>
> SRCS-$(CONFIG_RTE_LIBRTE_METER) += test_meter.c
> diff --git a/test/test/test_sched_wrr.c b/test/test/test_sched_wrr.c
> new file mode 100644
> index 0000000..df5a231
> --- /dev/null
> +++ b/test/test/test_sched_wrr.c
> @@ -0,0 +1,491 @@
> +/*-
> + * BSD LICENSE
> + *
> + * Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
> + * Copyright(c) 2017 ATT Intellectual Property. All rights
> reserved.
> + * 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
> disclaimer.
> + * * Redistributions in binary form must reproduce the above
> copyright
> + * notice, this list of conditions and the following
> disclaimer in
> + * the documentation and/or other materials provided with the
> + * distribution.
> + * * Neither the name of Intel Corporation nor the names of its
> + * contributors may be used to endorse or promote products
> derived
> + * from this software without specific prior written
> permission.
> + *
> + * 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 AND
> FITNESS FOR
> + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
> 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 CAUSED AND
> ON ANY
> + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 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 <stdlib.h>
> +#include <stdio.h>
> +#include <string.h>
> +#include <stdint.h>
> +#include <unistd.h>
> +
> +#include "test.h"
> +
> +#include <rte_cycles.h>
> +#include <rte_ether.h>
> +#include <rte_ip.h>
> +#include <rte_byteorder.h>
> +#include <rte_sched.h>
> +
> +
> +#define SUBPORT 0
> +#define PIPE 1
> +#define TC 0
> +#define QUEUE 3
> +
> +static struct rte_sched_subport_params subport_param[] = {
> + {
> + .tb_rate = 1250000000,
> + .tb_size = 1000000,
> +
> + .tc_rate = {1250000000, 1250000000, 1250000000,
> 1250000000},
> + .tc_period = 10,
> + },
> +};
> +
> +static struct rte_sched_pipe_params pipe_profile[] = {
> + { /* Profile #0 */
> + .tb_rate = 3051750,
> + .tb_size = 1000000,
> +
> + .tc_rate = {3051750, 3051750, 3051750, 3051750},
> + .tc_period = 160,
> +
> + .wrr_weights = {1, 1, 1, 1,
> + 1, 1, 1, 1,
> + 1, 1, 1, 1,
> + 1, 1, 1, 1},
> + },
> +};
> +
> +static struct rte_sched_port_params port_param = {
> + .socket = 0, /* computed */
> + .rate = 0, /* computed */
> + .mtu = 1522,
> + .frame_overhead = RTE_SCHED_FRAME_OVERHEAD_DEFAULT,
> + .n_subports_per_port = 1,
> + .n_pipes_per_subport = 1024,
> + .qsize = {32, 32, 32, 32},
> + .pipe_profiles = pipe_profile,
> + .n_pipe_profiles = 1,
> +};
> +
> +#define NB_MBUF (4 * 32)
> +#define MBUF_DATA_SZ (2048 + RTE_PKTMBUF_HEADROOM)
> +#define MEMPOOL_CACHE_SZ 0
> +#define SOCKET 0
> +
> +
> +static struct rte_mempool *
> +create_mempool(uint32_t total_packets)
> +{
> + struct rte_mempool *mp;
> +
> + mp = rte_mempool_lookup("test_sched wrr");
> + if (!mp)
> + mp = rte_pktmbuf_pool_create("test_sched wrr",
> total_packets,
> + MEMPOOL_CACHE_SZ, 0, MBUF_DATA_SZ, SOCKET);
> +
> + return mp;
> +}
> +
> +static void
> +delete_mempool(struct rte_mempool *mp)
> +{
> + rte_mempool_free(mp);
> +}
> +
> +static void
> +prepare_pkt(struct rte_mbuf *mbuf, uint32_t tc, uint32_t queue)
> +{
> + struct ether_hdr *eth_hdr;
> + struct vlan_hdr *vlan1, *vlan2;
> + struct ipv4_hdr *ip_hdr;
> +
> + /* Simulate a classifier */
> + eth_hdr = rte_pktmbuf_mtod(mbuf, struct ether_hdr *);
> + vlan1 = (struct vlan_hdr *)(ð_hdr->ether_type);
> + vlan2 = (struct vlan_hdr *)((uintptr_t)ð_hdr->ether_type
> +
> + sizeof(struct vlan_hdr));
> + eth_hdr = (struct ether_hdr *)((uintptr_t)ð_hdr-
> >ether_type +
> + 2 * sizeof(struct vlan_hdr));
> + ip_hdr = (struct ipv4_hdr *)((uintptr_t)eth_hdr +
> + sizeof(eth_hdr->ether_type));
> +
> + vlan1->vlan_tci = rte_cpu_to_be_16(SUBPORT);
> + vlan2->vlan_tci = rte_cpu_to_be_16(PIPE);
> + eth_hdr->ether_type = rte_cpu_to_be_16(ETHER_TYPE_IPv4);
> + ip_hdr->dst_addr = IPv4(0, 0, TC, QUEUE);
> +
> +
> + rte_sched_port_pkt_write(mbuf, SUBPORT, PIPE, tc, queue,
> + e_RTE_METER_YELLOW);
> +
> + /* 64 byte packet */
> + mbuf->pkt_len = 60;
> + mbuf->data_len = 60;
> +}
> +
> +/*
> + * This function carries out the core of the enqueue/dequeue
> testing.
> + * This is where we should detect failures if the WRR code is
> broken.
> + */
> +static int
> +test_sched_wrr_enqueue_dequeue(const char *subtest_name,
> + struct rte_mempool *mp,
> + struct rte_sched_port *port,
> + struct rte_mbuf **in_mbufs,
> + struct rte_mbuf **out_mbufs,
> + int32_t enqueue_packets,
> + int32_t dequeue_packets,
> + uint32_t *expected_counts)
> +{
> + uint32_t wrr_counts[RTE_SCHED_QUEUES_PER_TRAFFIC_CLASS] = {
> 0 };
> + uint32_t pipe;
> + int32_t i;
> + int err;
> +
> + /*
> + * Create the packets to be enqueued, spread them evenly
> across
> + * each of the four WRR queues of the test-TC
> + */
> + for (i = 0; i < enqueue_packets; i++) {
> + in_mbufs[i] = rte_pktmbuf_alloc(mp);
> + TEST_ASSERT_NOT_NULL(in_mbufs[i],
> + "%s: Packet allocation failed
> on packet "
> + "%d\n", subtest_name, i);
> + prepare_pkt(in_mbufs[i], TC, (i & 0x3));
> + }
> +
> + /*
> + * Queue all the enqueue packets, none should be dropped as
> the
> + * four queues should be long enough.
> + */
> + err = rte_sched_port_enqueue(port, in_mbufs,
> enqueue_packets);
> + TEST_ASSERT_EQUAL(err, enqueue_packets,
> + "%s: Enqueue failed, err: %d != %d\n",
> + subtest_name, err, enqueue_packets);
> +
> + /*
> + * Dequeue the required number of packets.
> + */
> + err = rte_sched_port_dequeue(port, out_mbufs,
> dequeue_packets);
> + TEST_ASSERT_EQUAL(err, dequeue_packets,
> + "%s: Dequeue failed, err: %d != %d\n",
> + subtest_name, err, dequeue_packets);
> +
> + /*
> + * Check each packet and count which WRR queue it came from.
> + */
> + for (i = 0; i < dequeue_packets; i++) {
> + enum rte_meter_color color;
> + uint32_t subport, traffic_class, queue;
> +
> + color = rte_sched_port_pkt_read_color(out_mbufs[i]);
> + TEST_ASSERT_EQUAL(color, e_RTE_METER_YELLOW,
> + "%s: Wrong color\n",
> subtest_name);
> +
> + rte_sched_port_pkt_read_tree_path(out_mbufs[i],
> + &subport, &pipe, &traffic_class,
> &queue);
> +
> + TEST_ASSERT_EQUAL(subport, SUBPORT, "%s: Wrong
> subport\n",
> + subtest_name);
> + TEST_ASSERT_EQUAL(pipe, PIPE, "%s: Wrong pipe\n",
> subtest_name);
> + TEST_ASSERT_EQUAL(traffic_class, TC,
> + "%s: Wrong traffic-class\n",
> subtest_name);
> + wrr_counts[queue]++;
> + rte_pktmbuf_free(out_mbufs[i]);
> + }
> +
> + /*
> + * Check the number of packets dequeued from each WRR queue
> + * against the expected counts.
> + */
> + err = 0;
> + for (i = 0; i < RTE_SCHED_QUEUES_PER_TRAFFIC_CLASS; i++) {
> + printf("%s - WRR queue %d, dequeued: %u, expected:
> %u\n",
> + subtest_name, i, wrr_counts[i],
> expected_counts[i]);
> + if (wrr_counts[i] != expected_counts[i])
> + err = -1;
> + }
> + return err;
> +}
> +
> +/*
> + * This function does the test config set-up and tear-down.
> + * If we see failures in here it is probably due to the test
> configuration.
> + */
> +static int
> +test_sched_wrr_test(const char *subtest_name, uint16_t *tc_qlengths,
> + uint8_t *wrr_weights, int32_t enqueue_packets,
> + int32_t dequeue_packets, uint32_t
> *expected_counts)
> +{
> + struct rte_mbuf **in_mbufs;
> + struct rte_mbuf **out_mbufs;
> + struct rte_mempool *mp = NULL;
> + struct rte_sched_port *port = NULL;
> + int32_t total_expected = 0;
> + uint32_t pipe;
> + int32_t i;
> + int err;
> +
> + /*
> + * Some inbound argument checking
> + */
> + TEST_ASSERT_EQUAL((tc_qlengths[TC] * 4), enqueue_packets,
> + "%s: Queue length/Enqueue packet mismatch
> - "
> + "%u vs %d\n",
> + subtest_name, (tc_qlengths[TC] * 4),
> enqueue_packets);
> + TEST_ASSERT((dequeue_packets <= enqueue_packets),
> + "%s: Dequeue packets %d > Enqueue packets %d\n",
> + subtest_name, dequeue_packets, enqueue_packets);
> +
> + for (i = 0; i < RTE_SCHED_QUEUES_PER_TRAFFIC_CLASS; i++)
> + total_expected += expected_counts[i];
> +
> + TEST_ASSERT((dequeue_packets == total_expected),
> + "%s: Dequeue packets %d != Total expected %u\n",
> + subtest_name, dequeue_packets, total_expected);
> +
> + /*
> + * Create the mempool and allocate arrays to hold the
> rte_mbuf pointers
> + */
> + mp = create_mempool(enqueue_packets);
> + TEST_ASSERT_NOT_NULL(mp, "%s: Error creating mempool\n",
> subtest_name);
> +
> + in_mbufs = malloc(sizeof(struct rte_mbuf *) *
> enqueue_packets);
> + TEST_ASSERT_NOT_NULL(in_mbufs, "%s: Error creating in_mbuf
> array\n",
> + subtest_name);
> +
> + out_mbufs = malloc(sizeof(struct rte_mbuf *) *
> dequeue_packets);
> + TEST_ASSERT_NOT_NULL(out_mbufs, "%s: Error creating out_mbuf
> array\n",
> + subtest_name);
> +
> + /*
> + * Set up the port and pipe profiles with the TC's queue
> lengths
> + * and the WRR queue weightings
> + */
> + port_param.socket = 0;
> + port_param.rate = (uint64_t) 10000 * 1000 * 1000 / 8;
> + for (i = 0; i < RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE; i++)
> + port_param.qsize[i] = tc_qlengths[i];
> +
> + for (i = 0; i < RTE_SCHED_QUEUES_PER_PIPE; i++)
> + pipe_profile[0].wrr_weights[i] = wrr_weights[i];
> +
> + port = rte_sched_port_config(&port_param);
> + TEST_ASSERT_NOT_NULL(port, "%s: Error config sched port\n",
> + subtest_name);
> +
> + err = rte_sched_subport_config(port, SUBPORT,
> subport_param);
> + TEST_ASSERT_SUCCESS(err, "%s: Error config sched, err=%d\n",
> + subtest_name, err);
> +
> + for (pipe = 0; pipe < port_param.n_pipes_per_subport;
> pipe++) {
> + err = rte_sched_pipe_config(port, SUBPORT, pipe, 0);
> + TEST_ASSERT_SUCCESS(err,
> + "%s: Error config sched pipe %u,
> err=%d\n",
> + subtest_name, pipe, err);
> + }
> +
> + /*
> + * Enqueue and dequeue packets checking that each WRR queue
> dequeued
> + * the correct number of packets.
> + */
> + err = test_sched_wrr_enqueue_dequeue(subtest_name, mp, port,
> in_mbufs,
> + out_mbufs,
> enqueue_packets,
> + dequeue_packets,
> + expected_counts);
> +
> + /*
> + * Free up allocated resources
> + */
> + free(in_mbufs);
> + free(out_mbufs);
> + rte_sched_port_free(port);
> + delete_mempool(mp);
> +
> + return err;
> +}
> +
> +static int
> +test_sched_wrr_even_weights(void)
> +{
> + /*
> + * Even weighting - each WRR queue should dequeue the same
> number of
> + * packets
> + */
> + uint16_t tc_qlengths[] = { 64, 64, 64, 64 };
> + uint8_t wrr_weights[] = { 1, 1, 1, 1, // TC-0
> + 1, 1, 1, 1, // TC-1
> + 1, 1, 1, 1, // TC-2
> + 1, 1, 1, 1 }; // TC-3
> + int32_t enqueue_packets = 64 * 4;
> + int32_t dequeue_packets = 32;
> + uint32_t expected_counts[] = { 8, 8, 8, 8 };
> +
> + return test_sched_wrr_test("wrr-even-weights", tc_qlengths,
> wrr_weights,
> + enqueue_packets, dequeue_packets,
> + expected_counts);
> +}
> +
> +static int
> +test_sched_wrr_8_4_2_1_weights(void)
> +{
> + /*
> + * Uneven weightings but with a low LCM of 8 * 4 * 2 * 1 =
> 64
> + */
> + uint16_t tc_qlengths[] = { 64, 64, 64, 64 };
> + uint8_t wrr_weights[] = { 8, 4, 2, 1, // TC-0
> + 1, 1, 1, 1,
> + 1, 1, 1, 1,
> + 1, 1, 1, 1 };
> + int32_t enqueue_packets = 64 * 4;
> + int32_t dequeue_packets = 15;
> + uint32_t expected_counts[] = { 8, 4, 2, 1 };
> +
> + return test_sched_wrr_test("wrr-8-4-2-1-weights",
> tc_qlengths,
> + wrr_weights, enqueue_packets,
> + dequeue_packets,
> expected_counts);
> +}
> +
> +static int
> +test_sched_wrr_1_2_3_4_weights(void)
> +{
> + /*
> + * Uneven weightings with a low LCM, but weightings in
> reverse order
> + * from previous test
> + */
> + uint16_t tc_qlengths[] = { 64, 64, 64, 64 };
> + uint8_t wrr_weights[] = { 1, 2, 3, 4,
> + 1, 1, 1, 1,
> + 1, 1, 1, 1,
> + 1, 1, 1, 1 };
> + int32_t enqueue_packets = 64 * 4;
> + int32_t dequeue_packets = 40;
> + uint32_t expected_counts[] = { 4, 8, 12, 16 };
> +
> + return test_sched_wrr_test("wrr-1-2-3-4-weights",
> tc_qlengths,
> + wrr_weights, enqueue_packets,
> + dequeue_packets,
> expected_counts);
> +}
> +
> +static int
> +test_sched_wrr_11_7_5_3_weights(void)
> +{
> + /*
> + * This test generates a LCM of 11 * 7 * 5 * 3 = 1155.
> + * 1155 / 3 = 385 which is more than can be fitted in a
> uint8_t
> + */
> + uint16_t tc_qlengths[] = { 64, 64, 64, 64 };
> + uint8_t wrr_weights[] = { 11, 7, 5, 3,
> + 1, 1, 1, 1,
> + 1, 1, 1, 1,
> + 1, 1, 1, 1 };
> + int32_t enqueue_packets = 64 * 4;
> + int32_t dequeue_packets = 26;
> + uint32_t expected_counts[] = { 11, 7, 5, 3 };
> +
> + return test_sched_wrr_test("wrr-11-7-5-3-weights",
> tc_qlengths,
> + wrr_weights, enqueue_packets,
> + dequeue_packets,
> expected_counts);
> +}
> +
> +static int
> +test_sched_wrr_100_to_97_weights(void)
> +{
> + /*
> + * The ratios between the weightings is small how well will
> the
> + * pseudo floating point wrr_cost perform.
> + */
> + uint16_t tc_qlengths[] = { 128, 1, 1, 1 };
> + uint8_t wrr_weights[] = { 100, 99, 98, 97,
> + 1, 1, 1, 1,
> + 1, 1, 1, 1,
> + 1, 1, 1, 1 };
> + int32_t enqueue_packets = 128 * 4;
> + int32_t dequeue_packets = 394;
> + uint32_t expected_counts[] = { 100, 99, 98, 97 };
> +
> + return test_sched_wrr_test("wrr-100-to-97-weights",
> tc_qlengths,
> + wrr_weights, enqueue_packets,
> + dequeue_packets,
> expected_counts);
> +}
> +
> +static int
> +test_sched_wrr_255_254_253_1_weights(void)
> +{
> + /*
> + * This test generates a LCM of 255 * 254 * 253 * 1 =
> 16386810
> + * 16386810 / 255 = 64262 = 0xFB06
> + * 16386810 / 1 = 16386810 = 0xFA0AFF
> + * 0xFA0AFF needs to be shifted 16 bits left to fit into a
> uint8_t
> + * but shifting 0xFB06 16 bits left results in a zero
> wrr_cost.
> + *
> + * This test can fail in two different ways.
> + *
> + * Very badly when the dequeued packet counts aren't even
> close to the
> + * expected counts due a bug in a recent WRR patch.
> + *
> + * A very near miss, when the dequeued packet counts are
> just one or
> + * two packets away from the expected counts. This happens
> because
> + * the weights of 255, 254, 254 and 1 get converted into
> wrr_costs
> + * of 1, 1, 1, 250 respectively. I think that this happens
> due to
> + * rounding errors in the pseudo floating point code used by
> WRR.
> + */
> + uint16_t tc_qlengths[] = { 1024, 1, 1, 1 };
> + uint8_t wrr_weights[] = { 254, 1, 253, 255,
> + 1, 1, 1, 1,
> + 1, 1, 1, 1,
> + 1, 1, 1, 1 };
> + int32_t enqueue_packets = 1024 * 4;
> + int32_t dequeue_packets = 255 + 254 + 253 + 1;
> + uint32_t expected_counts[] = { 254, 1, 253, 255 };
> +
> + return test_sched_wrr_test("wrr-255-254-253-1-weights",
> tc_qlengths,
> + wrr_weights, enqueue_packets,
> + dequeue_packets,
> expected_counts);
> +}
> +
> +/**
> + * WRR test main entrance for library sched
> + */
> +static int
> +test_sched_wrr(void)
> +{
> + TEST_ASSERT_SUCCESS(test_sched_wrr_even_weights(),
> + "even-weight test failed\n");
> + TEST_ASSERT_SUCCESS(test_sched_wrr_8_4_2_1_weights(),
> + "8-4-2-1-weight test failed\n");
> + TEST_ASSERT_SUCCESS(test_sched_wrr_1_2_3_4_weights(),
> + "1-2-3-4-weight test failed\n");
> + TEST_ASSERT_SUCCESS(test_sched_wrr_11_7_5_3_weights(),
> + "11-7-5-3-weight test failed\n");
> + TEST_ASSERT_SUCCESS(test_sched_wrr_100_to_97_weights(),
> + "100-to-97-weight test failed\n");
> + TEST_ASSERT_SUCCESS(test_sched_wrr_255_254_253_1_weights(),
> + "255-254-253-1-weight test failed\n");
> + return 0;
> +}
> +
> +REGISTER_TEST_COMMAND(sched_wrr_test, test_sched_wrr);
Reviewed-by: Luca Boccassi <bluca@debian.org>
LGTM
--
Kind regards,
Luca Boccassi
next prev parent reply other threads:[~2017-11-30 10:48 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-11-30 9:05 alangordondewar
2017-11-30 10:48 ` Luca Boccassi [this message]
2018-01-02 16:15 ` Dumitrescu, Cristian
2018-01-03 14:02 ` Dewar, Alan
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1512038883.10466.1.camel@debian.org \
--to=bluca@debian.org \
--cc=alan.dewar@att.com \
--cc=alangordondewar@gmail.com \
--cc=cristian.dumitrescu@intel.com \
--cc=dev@dpdk.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).