43. PTP Client Sample Application
The PTP (Precision Time Protocol) client sample application is a simple example of using the DPDK IEEE1588 API to communicate with a PTP master clock to synchronize the time on the NIC and, optionally, on the Linux system.
Note, PTP is a time syncing protocol and cannot be used within DPDK as a time-stamping mechanism. See the following for an explanation of the protocol: Precision Time Protocol.
43.1. Limitations
The PTP sample application is intended as a simple reference implementation of a PTP client using the DPDK IEEE1588 API. In order to keep the application simple the following assumptions are made:
The first discovered master is the main for the session.
Only L2 PTP packets are supported.
Only the PTP v2 protocol is supported.
Only the slave clock is implemented.
43.2. How the Application Works
The PTP synchronization in the sample application works as follows:
Master sends Sync message - the slave saves it as T2.
Master sends Follow Up message and sends time of T1.
Slave sends Delay Request frame to PTP Master and stores T3.
Master sends Delay Response T4 time which is time of received T3.
The adjustment for slave can be represented as:
adj = -[(T2-T1)-(T4 - T3)]/2
If the command line parameter -T 1
is used the application also
synchronizes the PTP PHC clock with the Linux kernel clock.
If the command line parameter -c 1
is used,
the application will also use the servo of the local clock.
Only one type of servo is currently implemented, the PI controller.
Default 0 (not used).
43.3. Compiling the Application
To compile the sample application see Compiling the Sample Applications.
The application is located in the ptpclient
sub-directory.
43.4. Running the Application
To run the example in a linux
environment:
./<build_dir>/examples/dpdk-ptpclient -l 1 -n 4 -- -p 0x1 -T 0 -c 1
Refer to DPDK Getting Started Guide for general information on running applications and the Environment Abstraction Layer (EAL) options.
-p portmask
: Hexadecimal portmask.-T 0
: Update only the PTP slave clock.-T 1
: Update the PTP slave clock and synchronize the Linux Kernel to the PTP clock.-c 0
: Not used clock servo controller.-c 1
: The clock servo PI controller is used and the log will print informationabout “master offset”. Note that the PMD needs to support the
rte_eth_timesync_adjust_freq()
API to enable the servo controller.
Also, by adding -T 1
and -c 1
, the master offset
value printed in the log
will slowly converge and eventually stabilise at the nanosecond level.
The synchronisation accuracy is much higher compared to not using a servo controller.
43.5. Code Explanation
The following sections provide an explanation of the main components of the code.
All DPDK library functions used in the sample code are prefixed with rte_
and are explained in detail in the DPDK API Documentation.
43.5.1. The Main Function
The main()
function performs the initialization and calls the execution
threads for each lcore.
The first task is to initialize the Environment Abstraction Layer (EAL). The
argc
and argv
arguments are provided to the rte_eal_init()
function. The value returned is the number of parsed arguments:
int ret = rte_eal_init(argc, argv);
if (ret < 0)
rte_exit(EXIT_FAILURE, "Error with EAL initialization\n");
And than we parse application specific arguments
argc -= ret;
argv += ret;
force_quit = false;
signal(SIGINT, signal_handler);
signal(SIGTERM, signal_handler);
ret = ptp_parse_args(argc, argv);
if (ret < 0)
rte_exit(EXIT_FAILURE, "Error with PTP initialization\n");
The main()
also allocates a mempool to hold the mbufs (Message Buffers)
used by the application:
mbuf_pool = rte_pktmbuf_pool_create("MBUF_POOL", NUM_MBUFS * nb_ports,
MBUF_CACHE_SIZE, 0, RTE_MBUF_DEFAULT_BUF_SIZE, rte_socket_id());
Mbufs are the packet buffer structure used by DPDK. They are explained in detail in the “Mbuf Library” section of the DPDK Programmer’s Guide.
The main()
function also initializes all the ports using the user defined
port_init()
function with portmask provided by user:
RTE_ETH_FOREACH_DEV(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);
}
Once the initialization is complete, the application is ready to launch a
function on an lcore. In this example lcore_main()
is called on a single
lcore.
lcore_main();
The lcore_main()
function is explained below.
43.5.2. The Lcores Main
As we saw above the main()
function calls an application function on the
available lcores.
The main work of the application is done within the loop:
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;
/* Packet is parsed to determine which type. 8< */
if (m->ol_flags & RTE_MBUF_F_RX_IEEE1588_PTP)
parse_ptp_frames(portid, m);
/* >8 End of packet is parsed to determine which type. */
rte_pktmbuf_free(m);
}
Packets are received one by one on the RX ports and, if required, PTP response packets are transmitted on the TX ports.
If the offload flags in the mbuf indicate that the packet is a PTP packet then the packet is parsed to determine which type:
if (m->ol_flags & RTE_MBUF_F_RX_IEEE1588_PTP)
parse_ptp_frames(portid, m);
All packets are freed explicitly using rte_pktmbuf_free()
.
The forwarding loop can be interrupted and the application closed using
Ctrl-C
.
43.5.3. PTP parsing
The parse_ptp_frames()
function processes PTP packets, implementing slave
PTP IEEE1588 L2 functionality.
static void
parse_ptp_frames(uint16_t portid, struct rte_mbuf *m) {
struct ptp_header *ptp_hdr;
struct rte_ether_hdr *eth_hdr;
uint16_t eth_type;
eth_hdr = rte_pktmbuf_mtod(m, struct rte_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 = rte_pktmbuf_mtod_offset(m, struct ptp_header *,
sizeof(struct rte_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;
}
}
}
There are 3 types of packets on the RX path which we must parse to create a minimal implementation of the PTP slave client:
SYNC packet.
FOLLOW UP packet
DELAY RESPONSE packet.
When we parse the FOLLOW UP packet we also create and send a DELAY_REQUEST packet. Also when we parse the DELAY RESPONSE packet, and all conditions are met we adjust the PTP slave clock.