#include <locale.h>
#include <stdlib.h>
#include <string.h>
#include <stdint.h>
#include <ctype.h>
#include <getopt.h>
#include <rte_jobstats.h>
#define RTE_LOGTYPE_L2FWD RTE_LOGTYPE_USER1
#define NB_MBUF 8192
#define MAX_PKT_BURST 32
#define BURST_TX_DRAIN_US 100
#define RX_DESC_DEFAULT 1024
#define TX_DESC_DEFAULT 1024
static uint16_t nb_rxd = RX_DESC_DEFAULT;
static uint16_t nb_txd = TX_DESC_DEFAULT;
static struct rte_ether_addr l2fwd_ports_eth_addr[RTE_MAX_ETHPORTS];
static uint32_t l2fwd_enabled_port_mask;
static uint32_t l2fwd_dst_ports[RTE_MAX_ETHPORTS];
#define UPDATE_STEP_UP 1
#define UPDATE_STEP_DOWN 32
static unsigned int l2fwd_rx_queue_per_lcore = 1;
#define MAX_RX_QUEUE_PER_LCORE 16
#define MAX_TX_QUEUE_PER_PORT 16
unsigned n_rx_port;
unsigned rx_port_list[MAX_RX_QUEUE_PER_LCORE];
uint64_t next_flush_time[RTE_MAX_ETHPORTS];
struct rte_timer rx_timers[MAX_RX_QUEUE_PER_LCORE];
struct rte_jobstats port_fwd_jobs[MAX_RX_QUEUE_PER_LCORE];
struct rte_jobstats flush_job;
struct rte_jobstats idle_job;
struct rte_jobstats_context jobs_context;
RTE_ATOMIC(uint16_t) stats_read_pending;
};
struct lcore_queue_conf lcore_queue_conf[RTE_MAX_LCORE];
},
};
uint64_t tx;
uint64_t rx;
uint64_t dropped;
};
struct l2fwd_port_statistics port_statistics[RTE_MAX_ETHPORTS];
#define MAX_TIMER_PERIOD 86400
static int64_t timer_period = 10;
static double hz;
uint64_t drain_tsc;
static inline double
cycles_to_ns(uint64_t cycles)
{
double t = cycles;
t *= (double)NS_PER_S;
t /= hz;
return t;
}
static void
show_lcore_stats(unsigned lcore_id)
{
struct lcore_queue_conf *qconf = &lcore_queue_conf[lcore_id];
struct rte_jobstats_context *ctx = &qconf->jobs_context;
struct rte_jobstats *job;
uint8_t i;
uint64_t stats_period, loop_count;
uint64_t exec, exec_min, exec_max;
uint64_t management, management_min, management_max;
uint64_t busy, busy_min, busy_max;
const uint16_t port_cnt = qconf->n_rx_port;
uint64_t jobs_exec_cnt[port_cnt], jobs_period[port_cnt];
uint64_t jobs_exec[port_cnt], jobs_exec_min[port_cnt],
jobs_exec_max[port_cnt];
uint64_t flush_exec_cnt, flush_period;
uint64_t flush_exec, flush_exec_min, flush_exec_max;
uint64_t idle_exec_cnt;
uint64_t idle_exec, idle_exec_min, idle_exec_max;
rte_atomic_store_explicit(&qconf->stats_read_pending, 1, rte_memory_order_relaxed);
rte_atomic_store_explicit(&qconf->stats_read_pending, 0, rte_memory_order_relaxed);
stats_period = ctx->state_time - ctx->start_time;
loop_count = ctx->loop_cnt;
exec = ctx->exec_time;
exec_min = ctx->min_exec_time;
exec_max = ctx->max_exec_time;
management = ctx->management_time;
management_min = ctx->min_management_time;
management_max = ctx->max_management_time;
rte_jobstats_context_reset(ctx);
for (i = 0; i < port_cnt; i++) {
job = &qconf->port_fwd_jobs[i];
jobs_exec_cnt[i] = job->exec_cnt;
jobs_period[i] = job->period;
jobs_exec[i] = job->exec_time;
jobs_exec_min[i] = job->min_exec_time;
jobs_exec_max[i] = job->max_exec_time;
rte_jobstats_reset(job);
}
flush_exec_cnt = qconf->flush_job.exec_cnt;
flush_period = qconf->flush_job.period;
flush_exec = qconf->flush_job.exec_time;
flush_exec_min = qconf->flush_job.min_exec_time;
flush_exec_max = qconf->flush_job.max_exec_time;
rte_jobstats_reset(&qconf->flush_job);
idle_exec_cnt = qconf->idle_job.exec_cnt;
idle_exec = qconf->idle_job.exec_time;
idle_exec_min = qconf->idle_job.min_exec_time;
idle_exec_max = qconf->idle_job.max_exec_time;
rte_jobstats_reset(&qconf->idle_job);
exec -= idle_exec;
busy = exec + management;
busy_min = exec_min + management_min;
busy_max = exec_max + management_max;
#define STAT_FMT "\n%-18s %'14.0f %6.1f%% %'10.0f %'10.0f %'10.0f"
printf("\n----------------"
"\nLCore %3u: statistics (time in ns, collected in %'9.0f)"
"\n%-18s %14s %7s %10s %10s %10s "
"\n%-18s %'14.0f"
"\n%-18s %'14" PRIu64
STAT_FMT
STAT_FMT
STAT_FMT
STAT_FMT,
lcore_id, cycles_to_ns(collection_time),
"Stat type", "total", "%total", "avg", "min", "max",
"Stats duration:", cycles_to_ns(stats_period),
"Loop count:", loop_count,
"Exec time",
cycles_to_ns(exec), exec * 100.0 / stats_period,
cycles_to_ns(loop_count ? exec / loop_count : 0),
cycles_to_ns(exec_min),
cycles_to_ns(exec_max),
"Management time",
cycles_to_ns(management), management * 100.0 / stats_period,
cycles_to_ns(loop_count ? management / loop_count : 0),
cycles_to_ns(management_min),
cycles_to_ns(management_max),
"Exec + management",
cycles_to_ns(busy), busy * 100.0 / stats_period,
cycles_to_ns(loop_count ? busy / loop_count : 0),
cycles_to_ns(busy_min),
cycles_to_ns(busy_max),
"Idle (job)",
cycles_to_ns(idle_exec), idle_exec * 100.0 / stats_period,
cycles_to_ns(idle_exec_cnt ? idle_exec / idle_exec_cnt : 0),
cycles_to_ns(idle_exec_min),
cycles_to_ns(idle_exec_max));
for (i = 0; i < qconf->n_rx_port; i++) {
job = &qconf->port_fwd_jobs[i];
printf("\n\nJob %" PRIu32 ": %-20s "
"\n%-18s %'14" PRIu64
"\n%-18s %'14.0f"
STAT_FMT,
i, job->name,
"Exec count:", jobs_exec_cnt[i],
"Exec period: ", cycles_to_ns(jobs_period[i]),
"Exec time",
cycles_to_ns(jobs_exec[i]), jobs_exec[i] * 100.0 / stats_period,
cycles_to_ns(jobs_exec_cnt[i] ? jobs_exec[i] / jobs_exec_cnt[i]
: 0),
cycles_to_ns(jobs_exec_min[i]),
cycles_to_ns(jobs_exec_max[i]));
}
if (qconf->n_rx_port > 0) {
job = &qconf->flush_job;
printf("\n\nJob %" PRIu32 ": %-20s "
"\n%-18s %'14" PRIu64
"\n%-18s %'14.0f"
STAT_FMT,
i, job->name,
"Exec count:", flush_exec_cnt,
"Exec period: ", cycles_to_ns(flush_period),
"Exec time",
cycles_to_ns(flush_exec), flush_exec * 100.0 / stats_period,
cycles_to_ns(flush_exec_cnt ? flush_exec / flush_exec_cnt : 0),
cycles_to_ns(flush_exec_min),
cycles_to_ns(flush_exec_max));
}
}
static void
{
uint64_t total_packets_dropped, total_packets_tx, total_packets_rx;
unsigned portid, lcore_id;
total_packets_dropped = 0;
total_packets_tx = 0;
total_packets_rx = 0;
const char clr[] = { 27, '[', '2', 'J', '\0' };
const char topLeft[] = { 27, '[', '1', ';', '1', 'H', '\0' };
printf("%s%s"
"\nPort statistics ===================================",
clr, topLeft);
for (portid = 0; portid < RTE_MAX_ETHPORTS; portid++) {
if ((l2fwd_enabled_port_mask & (1 << portid)) == 0)
continue;
printf("\nStatistics for port %u ------------------------------"
"\nPackets sent: %24"PRIu64
"\nPackets received: %20"PRIu64
"\nPackets dropped: %21"PRIu64,
portid,
port_statistics[portid].tx,
port_statistics[portid].rx,
port_statistics[portid].dropped);
total_packets_dropped += port_statistics[portid].dropped;
total_packets_tx += port_statistics[portid].tx;
total_packets_rx += port_statistics[portid].rx;
}
printf("\nAggregate statistics ==============================="
"\nTotal packets sent: %18"PRIu64
"\nTotal packets received: %14"PRIu64
"\nTotal packets dropped: %15"PRIu64
"\n====================================================",
total_packets_tx,
total_packets_rx,
total_packets_dropped);
if (lcore_queue_conf[lcore_id].n_rx_port > 0)
show_lcore_stats(lcore_id);
}
printf("\n====================================================\n");
fflush(stdout);
}
static void
l2fwd_simple_forward(
struct rte_mbuf *m,
unsigned portid)
{
struct rte_ether_hdr *eth;
void *tmp;
int sent;
unsigned dst_port;
dst_port = l2fwd_dst_ports[portid];
tmp = ð->dst_addr.addr_bytes[0];
*((uint64_t *)tmp) = 0x000000000002 + ((uint64_t)dst_port << 40);
buffer = tx_buffer[dst_port];
if (sent)
port_statistics[dst_port].tx += sent;
}
static void
l2fwd_job_update_cb(struct rte_jobstats *job, int64_t result)
{
int64_t err = job->target - result;
int64_t histeresis = job->target / 8;
if (err < -histeresis) {
if (job->min_period + UPDATE_STEP_DOWN < job->period)
job->period -= UPDATE_STEP_DOWN;
} else if (err > histeresis) {
if (job->period + UPDATE_STEP_UP < job->max_period)
job->period += UPDATE_STEP_UP;
}
}
static void
{
struct rte_mbuf *pkts_burst[MAX_PKT_BURST];
const uint16_t port_idx = (uintptr_t) arg;
struct lcore_queue_conf *qconf = &lcore_queue_conf[lcore_id];
struct rte_jobstats *job = &qconf->port_fwd_jobs[port_idx];
const uint16_t portid = qconf->rx_port_list[port_idx];
uint8_t j;
uint16_t total_nb_rx;
rte_jobstats_start(&qconf->jobs_context, job);
MAX_PKT_BURST);
for (j = 0; j < total_nb_rx; j++) {
m = pkts_burst[j];
l2fwd_simple_forward(m, portid);
}
if (total_nb_rx == MAX_PKT_BURST) {
MAX_PKT_BURST);
total_nb_rx += nb_rx;
for (j = 0; j < nb_rx; j++) {
m = pkts_burst[j];
l2fwd_simple_forward(m, portid);
}
}
port_statistics[portid].rx += total_nb_rx;
if (rte_jobstats_finish(job, total_nb_rx) != 0) {
lcore_id, l2fwd_fwd_job, arg);
}
}
static void
{
uint64_t now;
unsigned lcore_id;
struct lcore_queue_conf *qconf;
uint16_t portid;
unsigned i;
uint32_t sent;
qconf = &lcore_queue_conf[lcore_id];
rte_jobstats_start(&qconf->jobs_context, &qconf->flush_job);
qconf = &lcore_queue_conf[lcore_id];
for (i = 0; i < qconf->n_rx_port; i++) {
portid = l2fwd_dst_ports[qconf->rx_port_list[i]];
if (qconf->next_flush_time[portid] <= now)
continue;
buffer = tx_buffer[portid];
if (sent)
port_statistics[portid].tx += sent;
}
rte_jobstats_finish(&qconf->flush_job, qconf->flush_job.target);
}
static void
l2fwd_main_loop(void)
{
unsigned lcore_id;
unsigned i, portid;
struct lcore_queue_conf *qconf;
uint8_t stats_read_pending = 0;
uint8_t need_manage;
qconf = &lcore_queue_conf[lcore_id];
if (qconf->n_rx_port == 0) {
RTE_LOG(INFO, L2FWD,
"lcore %u has nothing to do\n", lcore_id);
return;
}
RTE_LOG(INFO, L2FWD,
"entering main loop on lcore %u\n", lcore_id);
for (i = 0; i < qconf->n_rx_port; i++) {
portid = qconf->rx_port_list[i];
RTE_LOG(INFO, L2FWD,
" -- lcoreid=%u portid=%u\n", lcore_id,
portid);
}
rte_jobstats_init(&qconf->idle_job, "idle", 0, 0, 0, 0);
for (;;) {
do {
rte_jobstats_context_start(&qconf->jobs_context);
rte_jobstats_start(&qconf->jobs_context, &qconf->idle_job);
uint64_t repeats = 0;
do {
uint8_t i;
repeats++;
need_manage = qconf->flush_timer.expire < now;
stats_read_pending = rte_atomic_load_explicit(
&qconf->stats_read_pending,
rte_memory_order_relaxed);
need_manage |= stats_read_pending;
for (i = 0; i < qconf->n_rx_port && !need_manage; i++)
need_manage = qconf->rx_timers[i].expire < now;
} while (!need_manage);
rte_jobstats_finish(&qconf->idle_job, qconf->idle_job.target);
else
rte_jobstats_abort(&qconf->idle_job);
rte_jobstats_context_finish(&qconf->jobs_context);
}
while (
likely(stats_read_pending == 0));
}
}
static int
{
l2fwd_main_loop();
return 0;
}
static void
l2fwd_usage(const char *prgname)
{
printf("%s [EAL options] -- -p PORTMASK [-q NQ]\n"
" -p PORTMASK: hexadecimal bitmask of ports to configure\n"
" -q NQ: number of queue (=ports) per lcore (default is 1)\n"
" -T PERIOD: statistics will be refreshed each PERIOD seconds (0 to disable, 10 default, 86400 maximum)\n"
" -l set system default locale instead of default (\"C\" locale) for thousands separator in stats.",
prgname);
}
static int
l2fwd_parse_portmask(const char *portmask)
{
char *end = NULL;
unsigned long pm;
pm = strtoul(portmask, &end, 16);
if ((portmask[0] == '\0') || (end == NULL) || (*end != '\0'))
return 0;
return pm;
}
static unsigned int
l2fwd_parse_nqueue(const char *q_arg)
{
char *end = NULL;
unsigned long n;
n = strtoul(q_arg, &end, 10);
if ((q_arg[0] == '\0') || (end == NULL) || (*end != '\0'))
return 0;
if (n == 0)
return 0;
if (n >= MAX_RX_QUEUE_PER_LCORE)
return 0;
return n;
}
static int
l2fwd_parse_timer_period(const char *q_arg)
{
char *end = NULL;
int n;
n = strtol(q_arg, &end, 10);
if ((q_arg[0] == '\0') || (end == NULL) || (*end != '\0'))
return -1;
if (n >= MAX_TIMER_PERIOD)
return -1;
return n;
}
static int
l2fwd_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:q:T:l",
lgopts, &option_index)) != EOF) {
switch (opt) {
case 'p':
l2fwd_enabled_port_mask = l2fwd_parse_portmask(optarg);
if (l2fwd_enabled_port_mask == 0) {
printf("invalid portmask\n");
l2fwd_usage(prgname);
return -1;
}
break;
case 'q':
l2fwd_rx_queue_per_lcore = l2fwd_parse_nqueue(optarg);
if (l2fwd_rx_queue_per_lcore == 0) {
printf("invalid queue number\n");
l2fwd_usage(prgname);
return -1;
}
break;
case 'T':
timer_period = l2fwd_parse_timer_period(optarg);
if (timer_period < 0) {
printf("invalid timer period\n");
l2fwd_usage(prgname);
return -1;
}
break;
case 'l':
setlocale(LC_ALL, "");
break;
case 0:
l2fwd_usage(prgname);
return -1;
default:
l2fwd_usage(prgname);
return -1;
}
}
if (optind >= 0)
argv[optind-1] = prgname;
ret = optind-1;
optind = 1;
return ret;
}
static void
check_all_ports_link_status(uint32_t port_mask)
{
#define CHECK_INTERVAL 100
#define MAX_CHECK_TIME 90
uint16_t portid;
uint8_t count, all_ports_up, print_flag = 0;
int ret;
printf("\nChecking link status");
fflush(stdout);
for (count = 0; count <= MAX_CHECK_TIME; count++) {
all_ports_up = 1;
if ((port_mask & (1 << portid)) == 0)
continue;
memset(&link, 0, sizeof(link));
if (ret < 0) {
all_ports_up = 0;
if (print_flag == 1)
printf("Port %u link get failed: %s\n",
continue;
}
if (print_flag == 1) {
sizeof(link_status_text), &link);
printf("Port %d %s\n", portid,
link_status_text);
continue;
}
all_ports_up = 0;
break;
}
}
if (print_flag == 1)
break;
if (all_ports_up == 0) {
printf(".");
fflush(stdout);
}
if (all_ports_up == 1 || count == (MAX_CHECK_TIME - 1)) {
print_flag = 1;
printf("done\n");
}
}
}
int
main(int argc, char **argv)
{
struct lcore_queue_conf *qconf;
unsigned lcore_id, rx_lcore_id;
unsigned nb_ports_in_mask = 0;
int ret;
char name[RTE_JOBSTATS_NAMESIZE];
uint16_t nb_ports;
uint16_t nb_ports_available = 0;
uint16_t portid, last_port;
uint8_t i;
if (ret < 0)
rte_exit(EXIT_FAILURE,
"Invalid EAL arguments\n");
argc -= ret;
argv += ret;
ret = l2fwd_parse_args(argc, argv);
if (ret < 0)
rte_exit(EXIT_FAILURE,
"Invalid L2FWD arguments\n");
l2fwd_pktmbuf_pool =
if (l2fwd_pktmbuf_pool == NULL)
rte_exit(EXIT_FAILURE,
"Cannot init mbuf pool\n");
if (nb_ports == 0)
rte_exit(EXIT_FAILURE,
"No Ethernet ports - bye\n");
for (portid = 0; portid < RTE_MAX_ETHPORTS; portid++)
l2fwd_dst_ports[portid] = 0;
last_port = 0;
if ((l2fwd_enabled_port_mask & (1 << portid)) == 0)
continue;
if (nb_ports_in_mask % 2) {
l2fwd_dst_ports[portid] = last_port;
l2fwd_dst_ports[last_port] = portid;
} else
last_port = portid;
nb_ports_in_mask++;
}
if (nb_ports_in_mask % 2) {
printf("Notice: odd number of ports in portmask.\n");
l2fwd_dst_ports[last_port] = last_port;
}
rx_lcore_id = 0;
qconf = NULL;
if ((l2fwd_enabled_port_mask & (1 << portid)) == 0)
continue;
lcore_queue_conf[rx_lcore_id].n_rx_port ==
l2fwd_rx_queue_per_lcore) {
rx_lcore_id++;
if (rx_lcore_id >= RTE_MAX_LCORE)
rte_exit(EXIT_FAILURE,
"Not enough cores\n");
}
if (qconf != &lcore_queue_conf[rx_lcore_id])
qconf = &lcore_queue_conf[rx_lcore_id];
qconf->rx_port_list[qconf->n_rx_port] = portid;
qconf->n_rx_port++;
printf("Lcore %u: RX port %u\n", rx_lcore_id, portid);
}
if ((l2fwd_enabled_port_mask & (1 << portid)) == 0) {
printf("Skipping disabled port %u\n", portid);
continue;
}
nb_ports_available++;
printf("Initializing port %u... ", portid);
fflush(stdout);
if (ret != 0)
"Error during getting device (port %u) info: %s\n",
portid, strerror(-ret));
if (ret < 0)
rte_exit(EXIT_FAILURE,
"Cannot configure device: err=%d, port=%u\n",
ret, portid);
&nb_txd);
if (ret < 0)
"Cannot adjust number of descriptors: err=%d, port=%u\n",
ret, portid);
&l2fwd_ports_eth_addr[portid]);
if (ret < 0)
"Cannot get MAC address: err=%d, port=%u\n",
ret, portid);
fflush(stdout);
rxq_conf = dev_info.default_rxconf;
&rxq_conf,
l2fwd_pktmbuf_pool);
if (ret < 0)
rte_exit(EXIT_FAILURE,
"rte_eth_rx_queue_setup:err=%d, port=%u\n",
ret, portid);
txq_conf = dev_info.default_txconf;
fflush(stdout);
&txq_conf);
if (ret < 0)
"rte_eth_tx_queue_setup:err=%d, port=%u\n",
ret, portid);
if (tx_buffer[portid] == NULL)
rte_exit(EXIT_FAILURE,
"Cannot allocate buffer for tx on port %u\n",
portid);
&port_statistics[portid].dropped);
if (ret < 0)
"Cannot set error callback for tx buffer on port %u\n",
portid);
if (ret < 0)
rte_exit(EXIT_FAILURE,
"rte_eth_dev_start:err=%d, port=%u\n",
ret, portid);
printf("done:\n");
if (ret != 0) {
"rte_eth_promiscuous_enable:err=%s, port=%u\n",
return ret;
}
portid,
memset(&port_statistics, 0, sizeof(port_statistics));
}
if (!nb_ports_available) {
"All available ports are disabled. Please set portmask.\n");
}
check_all_ports_link_status(l2fwd_enabled_port_mask);
drain_tsc = (hz + US_PER_S - 1) / US_PER_S * BURST_TX_DRAIN_US;
qconf = &lcore_queue_conf[lcore_id];
if (rte_jobstats_context_init(&qconf->jobs_context) != 0)
rte_panic(
"Jobs stats context for core %u init failed\n", lcore_id);
if (qconf->n_rx_port == 0) {
"lcore %u: no ports so no jobs stats context initialization\n",
lcore_id);
continue;
}
rte_jobstats_init(&qconf->flush_job, "flush", drain_tsc, drain_tsc,
drain_tsc, 0);
lcore_id, &l2fwd_flush_job, NULL);
if (ret < 0) {
rte_exit(1,
"Failed to reset flush job timer for lcore %u: %s",
}
for (i = 0; i < qconf->n_rx_port; i++) {
struct rte_jobstats *job = &qconf->port_fwd_jobs[i];
portid = qconf->rx_port_list[i];
printf("Setting forward job for port %u\n", portid);
snprintf(name,
RTE_DIM(name),
"port %u fwd", portid);
rte_jobstats_init(job, name, 0, drain_tsc, 0, MAX_PKT_BURST);
rte_jobstats_set_update_period_function(job, l2fwd_job_update_cb);
&l2fwd_fwd_job, (void *)(uintptr_t)i);
if (ret < 0) {
rte_exit(1,
"Failed to reset lcore %u port %u job timer: %s",
}
}
}
if (timer_period)
else
RTE_LOG(INFO, L2FWD,
"Stats display disabled\n");
return -1;
}
return 0;
}