examples/ptpclient: add minimal PTP client
authorDaniel Mrzyglod <danielx.t.mrzyglod@intel.com>
Fri, 13 Nov 2015 16:09:13 +0000 (16:09 +0000)
committerThomas Monjalon <thomas.monjalon@6wind.com>
Fri, 13 Nov 2015 16:44:49 +0000 (17:44 +0100)
Add a sample application that acts as a PTP slave using the
DPDK ieee1588 functions.

Signed-off-by: Daniel Mrzyglod <danielx.t.mrzyglod@intel.com>
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Reviewed-by: John McNamara <john.mcnamara@intel.com>
MAINTAINERS
examples/Makefile
examples/ptpclient/Makefile [new file with mode: 0644]
examples/ptpclient/ptpclient.c [new file with mode: 0644]

index 0aa9504..d6feada 100644 (file)
@@ -522,3 +522,7 @@ F: examples/tep_termination/
 F: examples/vmdq/
 F: examples/vmdq_dcb/
 F: doc/guides/sample_app_ug/vmdq_dcb_forwarding.rst
+
+M: Pablo de Lara <pablo.de.lara.guarch@intel.com>
+M: Daniel Mrzyglod <danielx.t.mrzyglod@intel.com>
+F: examples/ptpclient/
index b4eddbd..830e31a 100644 (file)
@@ -62,6 +62,7 @@ DIRS-y += load_balancer
 DIRS-y += multi_process
 DIRS-y += netmap_compat/bridge
 DIRS-$(CONFIG_RTE_LIBRTE_REORDER) += packet_ordering
+DIRS-$(CONFIG_RTE_LIBRTE_IEEE1588) += ptpclient
 DIRS-$(CONFIG_RTE_LIBRTE_METER) += qos_meter
 DIRS-$(CONFIG_RTE_LIBRTE_SCHED) += qos_sched
 DIRS-y += quota_watermark
diff --git a/examples/ptpclient/Makefile b/examples/ptpclient/Makefile
new file mode 100644 (file)
index 0000000..b77cf71
--- /dev/null
@@ -0,0 +1,56 @@
+#   BSD LICENSE
+#
+#   Copyright(c) 2015 Intel Corporation. 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.
+
+ifeq ($(RTE_SDK),)
+$(error "Please define RTE_SDK environment variable")
+endif
+
+# Default target, can be overriddegitn by command line or environment
+RTE_TARGET ?= x86_64-native-linuxapp-gcc
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+# binary name
+APP = ptpclient
+
+# all source are stored in SRCS-y
+SRCS-y := ptpclient.c
+
+CFLAGS += -O3
+CFLAGS += $(WERROR_FLAGS)
+
+# workaround for a gcc bug with noreturn attribute
+# http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12603
+ifeq ($(CONFIG_RTE_TOOLCHAIN_GCC),y)
+CFLAGS_main.o += -Wno-return-type
+endif
+
+include $(RTE_SDK)/mk/rte.extapp.mk
diff --git a/examples/ptpclient/ptpclient.c b/examples/ptpclient/ptpclient.c
new file mode 100644 (file)
index 0000000..0af4f3b
--- /dev/null
@@ -0,0 +1,780 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright(c) 2015 Intel Corporation. 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.
+ */
+
+/*
+ * This application is a simple Layer 2 PTP v2 client. It shows delta values
+ * which are used to synchronize the PHC clock. if the "-T 1" parameter is
+ * passed to the application the Linux kernel clock is also synchronized.
+ */
+
+#include <stdint.h>
+#include <inttypes.h>
+#include <rte_eal.h>
+#include <rte_ethdev.h>
+#include <rte_cycles.h>
+#include <rte_lcore.h>
+#include <rte_mbuf.h>
+#include <rte_ip.h>
+#include <limits.h>
+#include <sys/time.h>
+#include <getopt.h>
+
+#define RX_RING_SIZE 128
+#define TX_RING_SIZE 512
+
+#define NUM_MBUFS            8191
+#define MBUF_CACHE_SIZE       250
+
+/* Values for the PTP messageType field. */
+#define SYNC                  0x0
+#define DELAY_REQ             0x1
+#define PDELAY_REQ            0x2
+#define PDELAY_RESP           0x3
+#define FOLLOW_UP             0x8
+#define DELAY_RESP            0x9
+#define PDELAY_RESP_FOLLOW_UP 0xA
+#define ANNOUNCE              0xB
+#define SIGNALING             0xC
+#define MANAGEMENT            0xD
+
+#define NSEC_PER_SEC        1000000000L
+#define KERNEL_TIME_ADJUST_LIMIT  20000
+#define PTP_PROTOCOL             0x88F7
+
+struct rte_mempool *mbuf_pool;
+uint32_t ptp_enabled_port_mask;
+uint8_t ptp_enabled_port_nb;
+static uint8_t ptp_enabled_ports[RTE_MAX_ETHPORTS];
+
+static const struct rte_eth_conf port_conf_default = {
+       .rxmode = { .max_rx_pkt_len = ETHER_MAX_LEN }
+};
+
+static const struct ether_addr ether_multicast = {
+       .addr_bytes = {0x01, 0x1b, 0x19, 0x0, 0x0, 0x0}
+};
+
+/* Structs used for PTP handling. */
+struct tstamp {
+       uint16_t   sec_msb;
+       uint32_t   sec_lsb;
+       uint32_t   ns;
+}  __attribute__((packed));
+
+struct clock_id {
+       uint8_t id[8];
+};
+
+struct port_id {
+       struct clock_id        clock_id;
+       uint16_t               port_number;
+}  __attribute__((packed));
+
+struct ptp_header {
+       uint8_t              msg_type;
+       uint8_t              ver;
+       uint16_t             message_length;
+       uint8_t              domain_number;
+       uint8_t              reserved1;
+       uint8_t              flag_field[2];
+       int64_t              correction;
+       uint32_t             reserved2;
+       struct port_id       source_port_id;
+       uint16_t             seq_id;
+       uint8_t              control;
+       int8_t               log_message_interval;
+} __attribute__((packed));
+
+struct sync_msg {
+       struct ptp_header   hdr;
+       struct tstamp       origin_tstamp;
+} __attribute__((packed));
+
+struct follow_up_msg {
+       struct ptp_header   hdr;
+       struct tstamp       precise_origin_tstamp;
+       uint8_t             suffix[0];
+} __attribute__((packed));
+
+struct delay_req_msg {
+       struct ptp_header   hdr;
+       struct tstamp       origin_tstamp;
+} __attribute__((packed));
+
+struct delay_resp_msg {
+       struct ptp_header    hdr;
+       struct tstamp        rx_tstamp;
+       struct port_id       req_port_id;
+       uint8_t              suffix[0];
+} __attribute__((packed));
+
+struct ptp_message {
+       union {
+               struct ptp_header          header;
+               struct sync_msg            sync;
+               struct delay_req_msg       delay_req;
+               struct follow_up_msg       follow_up;
+               struct delay_resp_msg      delay_resp;
+       } __attribute__((packed));
+};
+
+struct ptpv2_data_slave_ordinary {
+       struct rte_mbuf *m;
+       struct timespec tstamp1;
+       struct timespec tstamp2;
+       struct timespec tstamp3;
+       struct timespec tstamp4;
+       struct clock_id client_clock_id;
+       struct clock_id master_clock_id;
+       struct timeval new_adj;
+       int64_t delta;
+       uint8_t portid;
+       uint16_t seqID_SYNC;
+       uint16_t seqID_FOLLOWUP;
+       uint8_t ptpset;
+       uint8_t kernel_time_set;
+       uint8_t current_ptp_port;
+};
+
+static struct ptpv2_data_slave_ordinary ptp_data;
+
+static inline uint64_t timespec64_to_ns(const struct timespec *ts)
+{
+       return ((uint64_t) ts->tv_sec * NSEC_PER_SEC) + ts->tv_nsec;
+}
+
+static struct timeval
+ns_to_timeval(int64_t nsec)
+{
+       struct timespec t_spec = {0, 0};
+       struct timeval t_eval = {0, 0};
+       int32_t rem;
+
+       if (nsec == 0)
+               return t_eval;
+       rem = nsec % NSEC_PER_SEC;
+       t_spec.tv_sec = nsec / NSEC_PER_SEC;
+
+       if (rem < 0) {
+               t_spec.tv_sec--;
+               rem += NSEC_PER_SEC;
+       }
+
+       t_spec.tv_nsec = rem;
+       t_eval.tv_sec = t_spec.tv_sec;
+       t_eval.tv_usec = t_spec.tv_nsec / 1000;
+
+       return t_eval;
+}
+
+/*
+ * Initializes a given port using global settings and with the RX buffers
+ * coming from the mbuf_pool passed as a parameter.
+ */
+static inline int
+port_init(uint8_t port, struct rte_mempool *mbuf_pool)
+{
+       struct rte_eth_dev_info dev_info;
+       struct rte_eth_conf port_conf = port_conf_default;
+       const uint16_t rx_rings = 1;
+       const uint16_t tx_rings = 1;
+       int retval;
+       uint16_t q;
+
+       if (port >= rte_eth_dev_count())
+               return -1;
+
+       /* Configure the Ethernet device. */
+       retval = rte_eth_dev_configure(port, rx_rings, tx_rings, &port_conf);
+       if (retval != 0)
+               return retval;
+
+       /* Allocate and set up 1 RX queue per Ethernet port. */
+       for (q = 0; q < rx_rings; q++) {
+               retval = rte_eth_rx_queue_setup(port, q, RX_RING_SIZE,
+                               rte_eth_dev_socket_id(port), NULL, mbuf_pool);
+
+               if (retval < 0)
+                       return retval;
+       }
+
+       /* Allocate and set up 1 TX queue per Ethernet port. */
+       for (q = 0; q < tx_rings; q++) {
+               /* Setup txq_flags */
+               struct rte_eth_txconf *txconf;
+
+               rte_eth_dev_info_get(q, &dev_info);
+               txconf = &dev_info.default_txconf;
+               txconf->txq_flags = 0;
+
+               retval = rte_eth_tx_queue_setup(port, q, TX_RING_SIZE,
+                               rte_eth_dev_socket_id(port), txconf);
+               if (retval < 0)
+                       return retval;
+       }
+
+       /* Start the Ethernet port. */
+       retval = rte_eth_dev_start(port);
+       if (retval < 0)
+               return retval;
+
+       /* Enable timesync timestamping for the Ethernet device */
+       rte_eth_timesync_enable(port);
+
+       /* Enable RX in promiscuous mode for the Ethernet device. */
+       rte_eth_promiscuous_enable(port);
+
+       return 0;
+}
+
+static void
+print_clock_info(struct ptpv2_data_slave_ordinary *ptp_data)
+{
+       int64_t nsec;
+       struct timespec net_time, sys_time;
+
+       printf("Master Clock id: %02x:%02x:%02x:%02x:%02x:%02x:%02x:%02x",
+               ptp_data->master_clock_id.id[0],
+               ptp_data->master_clock_id.id[1],
+               ptp_data->master_clock_id.id[2],
+               ptp_data->master_clock_id.id[3],
+               ptp_data->master_clock_id.id[4],
+               ptp_data->master_clock_id.id[5],
+               ptp_data->master_clock_id.id[6],
+               ptp_data->master_clock_id.id[7]);
+
+       printf("\nT2 - Slave  Clock.  %lds %ldns",
+                       (ptp_data->tstamp2.tv_sec),
+                       (ptp_data->tstamp2.tv_nsec));
+
+       printf("\nT1 - Master Clock.  %lds %ldns ",
+                       ptp_data->tstamp1.tv_sec,
+                       (ptp_data->tstamp1.tv_nsec));
+
+       printf("\nT3 - Slave  Clock.  %lds %ldns",
+                       ptp_data->tstamp3.tv_sec,
+                       (ptp_data->tstamp3.tv_nsec));
+
+       printf("\nT4 - Master Clock.  %lds %ldns ",
+                       ptp_data->tstamp4.tv_sec,
+                       (ptp_data->tstamp4.tv_nsec));
+
+       printf("\nDelta between master and slave clocks:%"PRId64"ns\n",
+                       ptp_data->delta);
+
+       clock_gettime(CLOCK_REALTIME, &sys_time);
+       rte_eth_timesync_read_time(ptp_data->current_ptp_port, &net_time);
+
+       time_t ts = net_time.tv_sec;
+
+       printf("\n\nComparison between Linux kernel Time and PTP:");
+
+       printf("\nCurrent PTP Time: %.24s %.9ld ns",
+                       ctime(&ts), net_time.tv_nsec);
+
+       nsec = (int64_t)timespec64_to_ns(&net_time) -
+                       (int64_t)timespec64_to_ns(&sys_time);
+       ptp_data->new_adj = ns_to_timeval(nsec);
+
+       gettimeofday(&ptp_data->new_adj, NULL);
+
+       time_t tp = ptp_data->new_adj.tv_sec;
+
+       printf("\nCurrent SYS Time: %.24s %.6ld ns",
+                               ctime(&tp), ptp_data->new_adj.tv_usec);
+
+       printf("\nDelta between PTP and Linux Kernel time:%"PRId64"ns\n",
+                               nsec);
+
+       printf("[Ctrl+C to quit]\n");
+
+       /* Clear screen and put cursor in column 1, row 1 */
+       printf("\033[2J\033[1;1H");
+}
+
+static int64_t
+delta_eval(struct ptpv2_data_slave_ordinary *ptp_data)
+{
+       int64_t delta;
+       uint64_t t1 = 0;
+       uint64_t t2 = 0;
+       uint64_t t3 = 0;
+       uint64_t t4 = 0;
+
+       t1 = timespec64_to_ns(&ptp_data->tstamp1);
+       t2 = timespec64_to_ns(&ptp_data->tstamp2);
+       t3 = timespec64_to_ns(&ptp_data->tstamp3);
+       t4 = timespec64_to_ns(&ptp_data->tstamp4);
+
+       delta = -((int64_t)((t2 - t1) - (t4 - t3))) / 2;
+
+       return delta;
+}
+
+/*
+ * Parse the PTP SYNC message.
+ */
+static void
+parse_sync(struct ptpv2_data_slave_ordinary *ptp_data, uint16_t rx_tstamp_idx)
+{
+       struct ptp_header *ptp_hdr;
+
+       ptp_hdr = (struct ptp_header *)(rte_pktmbuf_mtod(ptp_data->m, char *)
+                       + sizeof(struct ether_hdr));
+       ptp_data->seqID_SYNC = rte_be_to_cpu_16(ptp_hdr->seq_id);
+
+       if (ptp_data->ptpset == 0) {
+               rte_memcpy(&ptp_data->master_clock_id,
+                               &ptp_hdr->source_port_id.clock_id,
+                               sizeof(struct clock_id));
+               ptp_data->ptpset = 1;
+       }
+
+       if (memcmp(&ptp_hdr->source_port_id.clock_id,
+                       &ptp_hdr->source_port_id.clock_id,
+                       sizeof(struct clock_id)) == 0) {
+
+               if (ptp_data->ptpset == 1)
+                       rte_eth_timesync_read_rx_timestamp(ptp_data->portid,
+                                       &ptp_data->tstamp2, rx_tstamp_idx);
+       }
+
+}
+
+/*
+ * Parse the PTP FOLLOWUP message and send DELAY_REQ to the master clock.
+ */
+static void
+parse_fup(struct ptpv2_data_slave_ordinary *ptp_data)
+{
+       struct ether_hdr *eth_hdr;
+       struct ptp_header *ptp_hdr;
+       struct clock_id *client_clkid;
+       struct ptp_message *ptp_msg;
+       struct rte_mbuf *created_pkt;
+       struct tstamp *origin_tstamp;
+       struct ether_addr eth_multicast = ether_multicast;
+       size_t pkt_size;
+       int wait_us;
+       struct rte_mbuf *m = ptp_data->m;
+
+       eth_hdr = rte_pktmbuf_mtod(m, struct ether_hdr *);
+       ptp_hdr = (struct ptp_header *)(rte_pktmbuf_mtod(m, char *)
+                       + sizeof(struct ether_hdr));
+       if (memcmp(&ptp_data->master_clock_id,
+                       &ptp_hdr->source_port_id.clock_id,
+                       sizeof(struct clock_id)) != 0)
+               return;
+
+       ptp_data->seqID_FOLLOWUP = rte_be_to_cpu_16(ptp_hdr->seq_id);
+       ptp_msg = (struct ptp_message *) (rte_pktmbuf_mtod(m, char *) +
+                                         sizeof(struct ether_hdr));
+
+       origin_tstamp = &ptp_msg->follow_up.precise_origin_tstamp;
+       ptp_data->tstamp1.tv_nsec = ntohl(origin_tstamp->ns);
+       ptp_data->tstamp1.tv_sec =
+               ((uint64_t)ntohl(origin_tstamp->sec_lsb)) |
+               (((uint64_t)ntohs(origin_tstamp->sec_msb)) << 32);
+
+       if (ptp_data->seqID_FOLLOWUP == ptp_data->seqID_SYNC) {
+
+               created_pkt = rte_pktmbuf_alloc(mbuf_pool);
+               pkt_size = sizeof(struct ether_hdr) +
+                       sizeof(struct ptp_message);
+               created_pkt->data_len = pkt_size;
+               created_pkt->pkt_len = pkt_size;
+               eth_hdr = rte_pktmbuf_mtod(created_pkt, struct ether_hdr *);
+               rte_eth_macaddr_get(ptp_data->portid, &eth_hdr->s_addr);
+
+               /* Set multicast address 01-1B-19-00-00-00. */
+               ether_addr_copy(&eth_multicast, &eth_hdr->d_addr);
+
+               eth_hdr->ether_type = htons(PTP_PROTOCOL);
+               ptp_msg = (struct ptp_message *)
+                       (rte_pktmbuf_mtod(created_pkt, char *) +
+                       sizeof(struct ether_hdr));
+
+               ptp_msg->delay_req.hdr.seq_id = htons(ptp_data->seqID_SYNC);
+               ptp_msg->delay_req.hdr.msg_type = DELAY_REQ;
+               ptp_msg->delay_req.hdr.ver = 2;
+               ptp_msg->delay_req.hdr.control = 1;
+               ptp_msg->delay_req.hdr.log_message_interval = 127;
+
+               /* Set up clock id. */
+               client_clkid =
+                       &ptp_msg->delay_req.hdr.source_port_id.clock_id;
+
+               client_clkid->id[0] = eth_hdr->s_addr.addr_bytes[0];
+               client_clkid->id[1] = eth_hdr->s_addr.addr_bytes[1];
+               client_clkid->id[2] = eth_hdr->s_addr.addr_bytes[2];
+               client_clkid->id[3] = 0xFF;
+               client_clkid->id[4] = 0xFE;
+               client_clkid->id[5] = eth_hdr->s_addr.addr_bytes[3];
+               client_clkid->id[6] = eth_hdr->s_addr.addr_bytes[4];
+               client_clkid->id[7] = eth_hdr->s_addr.addr_bytes[5];
+
+               rte_memcpy(&ptp_data->client_clock_id,
+                          client_clkid,
+                          sizeof(struct clock_id));
+
+               /* Enable flag for hardware timestamping. */
+               created_pkt->ol_flags |= PKT_TX_IEEE1588_TMST;
+
+               /*Read value from NIC to prevent latching with old value. */
+               rte_eth_timesync_read_tx_timestamp(ptp_data->portid,
+                               &ptp_data->tstamp3);
+
+               /* Transmit the packet. */
+               rte_eth_tx_burst(ptp_data->portid, 0, &created_pkt, 1);
+
+               wait_us = 0;
+               ptp_data->tstamp3.tv_nsec = 0;
+               ptp_data->tstamp3.tv_sec = 0;
+
+               /* Wait at least 1 us to read TX timestamp. */
+               while ((rte_eth_timesync_read_tx_timestamp(ptp_data->portid,
+                               &ptp_data->tstamp3) < 0) && (wait_us < 1000)) {
+                       rte_delay_us(1);
+                       wait_us++;
+               }
+       }
+}
+
+/*
+ * Update the kernel time with the difference between it and the current NIC
+ * time.
+ */
+static inline void
+update_kernel_time(void)
+{
+       int64_t nsec;
+       struct timespec net_time, sys_time;
+
+       clock_gettime(CLOCK_REALTIME, &sys_time);
+       rte_eth_timesync_read_time(ptp_data.current_ptp_port, &net_time);
+
+       nsec = (int64_t)timespec64_to_ns(&net_time) -
+              (int64_t)timespec64_to_ns(&sys_time);
+
+       ptp_data.new_adj = ns_to_timeval(nsec);
+
+       /*
+        * If difference between kernel time and system time in NIC is too big
+        * (more than +/- 20 microseconds), use clock_settime to set directly
+        * the kernel time, as adjtime is better for small adjustments (takes
+        * longer to adjust the time).
+        */
+
+       if (nsec > KERNEL_TIME_ADJUST_LIMIT || nsec < -KERNEL_TIME_ADJUST_LIMIT)
+               clock_settime(CLOCK_REALTIME, &net_time);
+       else
+               adjtime(&ptp_data.new_adj, 0);
+
+
+}
+
+/*
+ * Parse the DELAY_RESP message.
+ */
+static void
+parse_drsp(struct ptpv2_data_slave_ordinary *ptp_data)
+{
+       struct rte_mbuf *m = ptp_data->m;
+       struct ptp_message *ptp_msg;
+       struct tstamp *rx_tstamp;
+       uint16_t seq_id;
+
+       ptp_msg = (struct ptp_message *) (rte_pktmbuf_mtod(m, char *) +
+                                       sizeof(struct ether_hdr));
+       seq_id = rte_be_to_cpu_16(ptp_msg->delay_resp.hdr.seq_id);
+       if (memcmp(&ptp_data->client_clock_id,
+                  &ptp_msg->delay_resp.req_port_id.clock_id,
+                  sizeof(struct clock_id)) == 0) {
+               if (seq_id == ptp_data->seqID_FOLLOWUP) {
+                       rx_tstamp = &ptp_msg->delay_resp.rx_tstamp;
+                       ptp_data->tstamp4.tv_nsec = ntohl(rx_tstamp->ns);
+                       ptp_data->tstamp4.tv_sec =
+                               ((uint64_t)ntohl(rx_tstamp->sec_lsb)) |
+                               (((uint64_t)ntohs(rx_tstamp->sec_msb)) << 32);
+
+                       /* Evaluate the delta for adjustment. */
+                       ptp_data->delta = delta_eval(ptp_data);
+
+                       rte_eth_timesync_adjust_time(ptp_data->portid,
+                                                    ptp_data->delta);
+
+                       ptp_data->current_ptp_port = ptp_data->portid;
+
+                       /* Update kernel time if enabled in app parameters. */
+                       if (ptp_data->kernel_time_set == 1)
+                               update_kernel_time();
+
+
+
+               }
+       }
+}
+
+/* This function processes PTP packets, implementing slave PTP IEEE1588 L2
+ * functionality.
+ */
+static void
+parse_ptp_frames(uint8_t portid, struct rte_mbuf *m) {
+       struct ptp_header *ptp_hdr;
+       struct ether_hdr *eth_hdr;
+       uint16_t eth_type;
+
+       eth_hdr = rte_pktmbuf_mtod(m, struct ether_hdr *);
+       eth_type = rte_be_to_cpu_16(eth_hdr->ether_type);
+
+       if (eth_type == PTP_PROTOCOL) {
+               ptp_data.m = m;
+               ptp_data.portid = portid;
+               ptp_hdr = (struct ptp_header *)(rte_pktmbuf_mtod(m, char *)
+                                       + sizeof(struct ether_hdr));
+
+               switch (ptp_hdr->msg_type) {
+               case SYNC:
+                       parse_sync(&ptp_data, m->timesync);
+                       break;
+               case FOLLOW_UP:
+                       parse_fup(&ptp_data);
+                       break;
+               case DELAY_RESP:
+                       parse_drsp(&ptp_data);
+                       print_clock_info(&ptp_data);
+                       break;
+               default:
+                       break;
+               }
+       }
+}
+
+/*
+ * The lcore main. This is the main thread that does the work, reading from an
+ * input port and writing to an output port.
+ */
+static __attribute__((noreturn)) void
+lcore_main(void)
+{
+       uint8_t portid;
+       unsigned nb_rx;
+       struct rte_mbuf *m;
+
+       /*
+        * Check that the port is on the same NUMA node as the polling thread
+        * for best performance.
+        */
+       printf("\nCore %u Waiting for SYNC packets. [Ctrl+C to quit]\n",
+                       rte_lcore_id());
+
+       /* Run until the application is quit or killed. */
+
+       while (1) {
+               /* Read packet from RX queues. */
+               for (portid = 0; portid < ptp_enabled_port_nb; portid++) {
+
+                       portid = ptp_enabled_ports[portid];
+                       nb_rx = rte_eth_rx_burst(portid, 0, &m, 1);
+
+                       if (likely(nb_rx == 0))
+                               continue;
+
+                       if (m->ol_flags & PKT_RX_IEEE1588_PTP)
+                               parse_ptp_frames(portid, m);
+
+                       rte_pktmbuf_free(m);
+               }
+       }
+}
+
+static void
+print_usage(const char *prgname)
+{
+       printf("%s [EAL options] -- -p PORTMASK -T VALUE\n"
+               " -T VALUE: 0 - Disable, 1 - Enable Linux Clock"
+               " Synchronization (0 default)\n"
+               " -p PORTMASK: hexadecimal bitmask of ports to configure\n",
+               prgname);
+}
+
+static int
+ptp_parse_portmask(const char *portmask)
+{
+       char *end = NULL;
+       unsigned long pm;
+
+       /* Parse the hexadecimal string. */
+       pm = strtoul(portmask, &end, 16);
+
+       if ((portmask[0] == '\0') || (end == NULL) || (*end != '\0'))
+               return -1;
+
+       if (pm == 0)
+               return -1;
+
+       return pm;
+}
+
+static int
+parse_ptp_kernel(const char *param)
+{
+       char *end = NULL;
+       unsigned long pm;
+
+       /* Parse the hexadecimal string. */
+       pm = strtoul(param, &end, 16);
+
+       if ((param[0] == '\0') || (end == NULL) || (*end != '\0'))
+               return -1;
+       if (pm == 0)
+               return 0;
+
+       return 1;
+}
+
+/* Parse the commandline arguments. */
+static int
+ptp_parse_args(int argc, char **argv)
+{
+       int opt, ret;
+       char **argvopt;
+       int option_index;
+       char *prgname = argv[0];
+       static struct option lgopts[] = { {NULL, 0, 0, 0} };
+
+       argvopt = argv;
+
+       while ((opt = getopt_long(argc, argvopt, "p:T:",
+                                 lgopts, &option_index)) != EOF) {
+
+               switch (opt) {
+
+               /* Portmask. */
+               case 'p':
+                       ptp_enabled_port_mask = ptp_parse_portmask(optarg);
+                       if (ptp_enabled_port_mask == 0) {
+                               printf("invalid portmask\n");
+                               print_usage(prgname);
+                               return -1;
+                       }
+                       break;
+               /* Time synchronization. */
+               case 'T':
+                       ret = parse_ptp_kernel(optarg);
+                       if (ret < 0) {
+                               print_usage(prgname);
+                               return -1;
+                       }
+
+                       ptp_data.kernel_time_set = ret;
+                       break;
+
+               default:
+                       print_usage(prgname);
+                       return -1;
+               }
+       }
+
+       argv[optind-1] = prgname;
+
+       optind = 0; /* Reset getopt lib. */
+
+       return 0;
+}
+
+/*
+ * The main function, which does initialization and calls the per-lcore
+ * functions.
+ */
+int
+main(int argc, char *argv[])
+{
+       unsigned nb_ports;
+
+       uint8_t portid;
+
+       /* Initialize the Environment Abstraction Layer (EAL). */
+       int ret = rte_eal_init(argc, argv);
+
+       if (ret < 0)
+               rte_exit(EXIT_FAILURE, "Error with EAL initialization\n");
+
+       memset(&ptp_data, '\0', sizeof(struct ptpv2_data_slave_ordinary));
+
+       argc -= ret;
+       argv += ret;
+
+       ret = ptp_parse_args(argc, argv);
+       if (ret < 0)
+               rte_exit(EXIT_FAILURE, "Error with PTP initialization\n");
+
+       /* Check that there is an even number of ports to send/receive on. */
+       nb_ports = rte_eth_dev_count();
+
+       /* Creates a new mempool in memory to hold the mbufs. */
+       mbuf_pool = rte_pktmbuf_pool_create("MBUF_POOL", NUM_MBUFS * nb_ports,
+               MBUF_CACHE_SIZE, 0, RTE_MBUF_DEFAULT_BUF_SIZE, rte_socket_id());
+
+       if (mbuf_pool == NULL)
+               rte_exit(EXIT_FAILURE, "Cannot create mbuf pool\n");
+
+       /* Initialize all ports. */
+       for (portid = 0; portid < nb_ports; portid++) {
+               if ((ptp_enabled_port_mask & (1 << portid)) != 0) {
+                       if (port_init(portid, mbuf_pool) == 0) {
+                               ptp_enabled_ports[ptp_enabled_port_nb] = portid;
+                               ptp_enabled_port_nb++;
+                       } else {
+                               rte_exit(EXIT_FAILURE,
+                                        "Cannot init port %"PRIu8 "\n",
+                                        portid);
+                       }
+               } else
+                       printf("Skipping disabled port %u\n", portid);
+       }
+
+       if (ptp_enabled_port_nb == 0) {
+               rte_exit(EXIT_FAILURE,
+                       "All available ports are disabled."
+                       " Please set portmask.\n");
+       }
+
+       if (rte_lcore_count() > 1)
+               printf("\nWARNING: Too many lcores enabled. Only 1 used.\n");
+
+       /* Call lcore_main on the master core only. */
+       lcore_main();
+
+       return 0;
+}