xref: /mOS-networking-stack/core/src/dpdk_module.c (revision 522d5c66)

/* for io_module_func def'ns */
#include "io_module.h"
/* for mtcp related def'ns */
#include "mtcp.h"
/* for errno */
#include <errno.h>
/* for logging */
#include "debug.h"
/* for num_devices_* */
#include "config.h"
/* for rte_max_eth_ports */
#include <rte_common.h>
/* for rte_eth_rxconf */
#include <rte_ethdev.h>
/* for delay funcs */
#include <rte_cycles.h>
/* for ip pesudo-chksum */
#include <rte_ip.h>
#define ENABLE_STATS_IOCTL		1
#ifdef ENABLE_STATS_IOCTL
/* for close */
#include <unistd.h>
/* for open */
#include <fcntl.h>
/* for ioctl */
#include <sys/ioctl.h>
#endif /* !ENABLE_STATS_IOCTL */
/*----------------------------------------------------------------------------*/
/* Essential macros */
#define MAX_RX_QUEUE_PER_LCORE		MAX_CPUS
#define MAX_TX_QUEUE_PER_PORT		MAX_CPUS

#define MBUF_SIZE 			(2048 + sizeof(struct rte_mbuf) + RTE_PKTMBUF_HEADROOM)
#define NB_MBUF				8192
#define MEMPOOL_CACHE_SIZE		256

/*
 * RX and TX Prefetch, Host, and Write-back threshold values should be
 * carefully set for optimal performance. Consult the network
 * controller's datasheet and supporting DPDK documentation for guidance
 * on how these parameters should be set.
 */
#define RX_PTHRESH 			8 /**< Default values of RX prefetch threshold reg. */
#define RX_HTHRESH 			8 /**< Default values of RX host threshold reg. */
#define RX_WTHRESH 			4 /**< Default values of RX write-back threshold reg. */

/*
 * These default values are optimized for use with the Intel(R) 82599 10 GbE
 * Controller and the DPDK ixgbe PMD. Consider using other values for other
 * network controllers and/or network drivers.
 */
#define TX_PTHRESH 			36 /**< Default values of TX prefetch threshold reg. */
#define TX_HTHRESH			0  /**< Default values of TX host threshold reg. */
#define TX_WTHRESH			0  /**< Default values of TX write-back threshold reg. */

#define MAX_PKT_BURST			/*32*/64/*128*//*32*/

/*
 * Configurable number of RX/TX ring descriptors
 */
#define RTE_TEST_RX_DESC_DEFAULT	128
#define RTE_TEST_TX_DESC_DEFAULT	512

static uint16_t nb_rxd = RTE_TEST_RX_DESC_DEFAULT;
static uint16_t nb_txd = RTE_TEST_TX_DESC_DEFAULT;
/*----------------------------------------------------------------------------*/
/* packet memory pools for storing packet bufs */
static struct rte_mempool *pktmbuf_pool[MAX_CPUS] = {NULL};
static uint8_t cpu_qid_map[RTE_MAX_ETHPORTS][MAX_CPUS] = {{0}};

//#define DEBUG				1
#ifdef DEBUG
/* ethernet addresses of ports */
static struct ether_addr ports_eth_addr[RTE_MAX_ETHPORTS];
#endif

static struct rte_eth_conf port_conf = {
	.rxmode = {
		.mq_mode	= 	ETH_MQ_RX_RSS,
		.max_rx_pkt_len = 	ETHER_MAX_LEN,
		.split_hdr_size = 	0,
		.header_split   = 	0, /**< Header Split disabled */
		.hw_ip_checksum = 	1, /**< IP checksum offload enabled */
		.hw_vlan_filter = 	0, /**< VLAN filtering disabled */
		.jumbo_frame    = 	0, /**< Jumbo Frame Support disabled */
		.hw_strip_crc   = 	1, /**< CRC stripped by hardware */
	},
	.rx_adv_conf = {
		.rss_conf = {
			.rss_key = 	NULL,
			.rss_hf = 	ETH_RSS_TCP
		},
	},
	.txmode = {
		.mq_mode = 		ETH_MQ_TX_NONE,
	},
#if 0
	.fdir_conf = {
                .mode = RTE_FDIR_MODE_PERFECT,
                .pballoc = RTE_FDIR_PBALLOC_256K,
                .status = RTE_FDIR_REPORT_STATUS_ALWAYS,
                //.flexbytes_offset = 0x6,
                .drop_queue = 127,
        },
#endif
};

static const struct rte_eth_rxconf rx_conf = {
	.rx_thresh = {
		.pthresh = 		RX_PTHRESH, /* RX prefetch threshold reg */
		.hthresh = 		RX_HTHRESH, /* RX host threshold reg */
		.wthresh = 		RX_WTHRESH, /* RX write-back threshold reg */
	},
	.rx_free_thresh = 		32,
};

static const struct rte_eth_txconf tx_conf = {
	.tx_thresh = {
		.pthresh = 		TX_PTHRESH, /* TX prefetch threshold reg */
		.hthresh = 		TX_HTHRESH, /* TX host threshold reg */
		.wthresh = 		TX_WTHRESH, /* TX write-back threshold reg */
	},
	.tx_free_thresh = 		0, /* Use PMD default values */
	.tx_rs_thresh = 		0, /* Use PMD default values */
	/*
	 * As the example won't handle mult-segments and offload cases,
	 * set the flag by default.
	 */
	.txq_flags = 			0x0,
};

struct mbuf_table {
	unsigned len; /* length of queued packets */
	struct rte_mbuf *m_table[MAX_PKT_BURST];
};

struct dpdk_private_context {
	struct mbuf_table rmbufs[RTE_MAX_ETHPORTS];
	struct mbuf_table wmbufs[RTE_MAX_ETHPORTS];
	struct rte_mempool *pktmbuf_pool;
	struct rte_mbuf *pkts_burst[MAX_PKT_BURST];
#ifdef ENABLE_STATS_IOCTL
	int fd;
#endif /* !ENABLE_STATS_IOCTL */
} __rte_cache_aligned;

#ifdef ENABLE_STATS_IOCTL
/**
 * stats struct passed on from user space to the driver
 */
struct stats_struct {
	uint64_t tx_bytes;
	uint64_t tx_pkts;
	uint64_t rx_bytes;
	uint64_t rx_pkts;
	uint8_t qid;
	uint8_t dev;
};
#endif /* !ENABLE_STATS_IOCTL */
/*----------------------------------------------------------------------------*/
void
dpdk_init_handle(struct mtcp_thread_context *ctxt)
{
	struct dpdk_private_context *dpc;
	int i, j;
	char mempool_name[20];

	/* create and initialize private I/O module context */
	ctxt->io_private_context = calloc(1, sizeof(struct dpdk_private_context));
	if (ctxt->io_private_context == NULL) {
		TRACE_ERROR("Failed to initialize ctxt->io_private_context: "
			    "Can't allocate memory\n");
		exit(EXIT_FAILURE);
	}

	sprintf(mempool_name, "mbuf_pool-%d", ctxt->cpu);
	dpc = (struct dpdk_private_context *)ctxt->io_private_context;
	dpc->pktmbuf_pool = pktmbuf_pool[ctxt->cpu];

	/* set wmbufs correctly */
	for (j = 0; j < g_config.mos->netdev_table->num; j++) {
		/* Allocate wmbufs for each registered port */
		for (i = 0; i < MAX_PKT_BURST; i++) {
			dpc->wmbufs[j].m_table[i] = rte_pktmbuf_alloc(pktmbuf_pool[ctxt->cpu]);
			if (dpc->wmbufs[j].m_table[i] == NULL) {
				TRACE_ERROR("Failed to allocate %d:wmbuf[%d] on device %d!\n",
					    ctxt->cpu, i, j);
				exit(EXIT_FAILURE);
			}
		}
		/* set mbufs queue length to 0 to begin with */
		dpc->wmbufs[j].len = 0;
	}

#ifdef ENABLE_STATS_IOCTL
	dpc->fd = open("/dev/dpdk-iface", O_RDWR);
	if (dpc->fd == -1) {
		TRACE_ERROR("Can't open /dev/dpdk-iface for context->cpu: %d!\n",
			    ctxt->cpu);
		exit(EXIT_FAILURE);
	}
#endif /* !ENABLE_STATS_IOCTL */
}
/*----------------------------------------------------------------------------*/
int
dpdk_send_pkts(struct mtcp_thread_context *ctxt, int nif)
{
	struct dpdk_private_context *dpc;
	mtcp_manager_t mtcp;
	int ret;
	int qid;

	dpc = (struct dpdk_private_context *)ctxt->io_private_context;
	mtcp = ctxt->mtcp_manager;
	ret = 0;
	qid = cpu_qid_map[nif][ctxt->cpu];

	/* if queue is unassigned, skip it.. */
	if (unlikely(qid == 0xFF))
		return 0;

	/* if there are packets in the queue... flush them out to the wire */
	if (dpc->wmbufs[nif].len >/*= MAX_PKT_BURST*/ 0) {
		struct rte_mbuf **pkts;
#ifdef ENABLE_STATS_IOCTL
		struct stats_struct ss;
#endif /* !ENABLE_STATS_IOCTL */
		int cnt = dpc->wmbufs[nif].len;
		pkts = dpc->wmbufs[nif].m_table;
#ifdef NETSTAT
		mtcp->nstat.tx_packets[nif] += cnt;
#ifdef ENABLE_STATS_IOCTL
		ss.tx_pkts = mtcp->nstat.tx_packets[nif];
		ss.tx_bytes = mtcp->nstat.tx_bytes[nif];
		ss.rx_pkts = mtcp->nstat.rx_packets[nif];
		ss.rx_bytes = mtcp->nstat.rx_bytes[nif];
		ss.qid = ctxt->cpu;
		ss.dev = nif;
		ioctl(dpc->fd, 0, &ss);
#endif /* !ENABLE_STATS_IOCTL */
#endif
		do {
			/* tx cnt # of packets */
			ret = rte_eth_tx_burst(nif, qid,
					       pkts, cnt);
			pkts += ret;
			cnt -= ret;
			/* if not all pkts were sent... then repeat the cycle */
		} while (cnt > 0);

#ifndef SHARE_IO_BUFFER
		int i;
		/* time to allocate fresh mbufs for the queue */
		for (i = 0; i < dpc->wmbufs[nif].len; i++) {
			dpc->wmbufs[nif].m_table[i] = rte_pktmbuf_alloc(pktmbuf_pool[ctxt->cpu]);
			/* error checking */
			if (unlikely(dpc->wmbufs[nif].m_table[i] == NULL)) {
				TRACE_ERROR("Failed to allocate %d:wmbuf[%d] on device %d!\n",
					    ctxt->cpu, i, nif);
				exit(EXIT_FAILURE);
			}
		}
#endif
		/* reset the len of mbufs var after flushing of packets */
		dpc->wmbufs[nif].len = 0;
	}

	return ret;
}
/*----------------------------------------------------------------------------*/
uint8_t *
dpdk_get_wptr(struct mtcp_thread_context *ctxt, int nif, uint16_t pktsize)
{
	struct dpdk_private_context *dpc;
	mtcp_manager_t mtcp;
	struct rte_mbuf *m;
	uint8_t *ptr;
	int len_of_mbuf;

	dpc = (struct dpdk_private_context *) ctxt->io_private_context;
	mtcp = ctxt->mtcp_manager;

	/* sanity check */
	if (unlikely(dpc->wmbufs[nif].len == MAX_PKT_BURST))
		return NULL;

	len_of_mbuf = dpc->wmbufs[nif].len;
	m = dpc->wmbufs[nif].m_table[len_of_mbuf];

	/* retrieve the right write offset */
	ptr = (void *)rte_pktmbuf_mtod(m, struct ether_hdr *);
	m->pkt_len = m->data_len = pktsize;
	m->nb_segs = 1;
	m->next = NULL;

#ifdef NETSTAT
	mtcp->nstat.tx_bytes[nif] += pktsize + 24;
#endif

	/* increment the len_of_mbuf var */
	dpc->wmbufs[nif].len = len_of_mbuf + 1;

	return (uint8_t *)ptr;
}
/*----------------------------------------------------------------------------*/
void
dpdk_set_wptr(struct mtcp_thread_context *ctxt, int out_nif, int in_nif, int index)
{
	struct dpdk_private_context *dpc;
	mtcp_manager_t mtcp;
	int len_of_mbuf;

	dpc = (struct dpdk_private_context *) ctxt->io_private_context;
	mtcp = ctxt->mtcp_manager;

	/* sanity check */
	if (unlikely(dpc->wmbufs[out_nif].len == MAX_PKT_BURST))
		return;

	len_of_mbuf = dpc->wmbufs[out_nif].len;
	dpc->wmbufs[out_nif].m_table[len_of_mbuf] =
		dpc->rmbufs[in_nif].m_table[index];

	dpc->wmbufs[out_nif].m_table[len_of_mbuf]->udata64 = 0;

#ifdef NETSTAT
	mtcp->nstat.tx_bytes[out_nif] += dpc->rmbufs[in_nif].m_table[index]->pkt_len + 24;
#endif

	/* increment the len_of_mbuf var */
	dpc->wmbufs[out_nif].len = len_of_mbuf + 1;

	return;
}
/*----------------------------------------------------------------------------*/
static inline void
free_pkts(struct rte_mbuf **mtable, unsigned len)
{
	int i;

	/* free the freaking packets */
	for (i = 0; i < len; i++) {
		if (mtable[i]->udata64 == 1) {
			rte_pktmbuf_free_seg(mtable[i]);
			RTE_MBUF_PREFETCH_TO_FREE(mtable[i+1]);
		}
	}
}
/*----------------------------------------------------------------------------*/
int32_t
dpdk_recv_pkts(struct mtcp_thread_context *ctxt, int ifidx)
{
	struct dpdk_private_context *dpc;
	int ret;
	uint8_t qid;

	dpc = (struct dpdk_private_context *) ctxt->io_private_context;
	qid = cpu_qid_map[ifidx][ctxt->cpu];

	/* if queue is unassigned, skip it.. */
	if (qid == 0xFF)
		return 0;

	if (dpc->rmbufs[ifidx].len != 0) {
		free_pkts(dpc->rmbufs[ifidx].m_table, dpc->rmbufs[ifidx].len);
		dpc->rmbufs[ifidx].len = 0;
	}

	ret = rte_eth_rx_burst((uint8_t)ifidx, qid,
			       dpc->pkts_burst, MAX_PKT_BURST);

	dpc->rmbufs[ifidx].len = ret;

	return ret;
}
/*----------------------------------------------------------------------------*/
uint8_t *
dpdk_get_rptr(struct mtcp_thread_context *ctxt, int ifidx, int index, uint16_t *len)
{
	struct dpdk_private_context *dpc;
	struct rte_mbuf *m;
	uint8_t *pktbuf;

	dpc = (struct dpdk_private_context *) ctxt->io_private_context;


	m = dpc->pkts_burst[index];
	/* tag to check if the packet is a local or a forwarded pkt */
	m->udata64 = 1;
	/* don't enable pre-fetching... performance goes down */
	//rte_prefetch0(rte_pktmbuf_mtod(m, void *));
	*len = m->pkt_len;
	pktbuf = rte_pktmbuf_mtod(m, uint8_t *);

	/* enqueue the pkt ptr in mbuf */
	dpc->rmbufs[ifidx].m_table[index] = m;

	return pktbuf;
}
/*----------------------------------------------------------------------------*/
int
dpdk_get_nif(struct ifreq *ifr)
{
	int i;
	static int num_dev = -1;
	static struct ether_addr ports_eth_addr[RTE_MAX_ETHPORTS];
	/* get mac addr entries of 'detected' dpdk ports */
	if (num_dev < 0) {
		num_dev = rte_eth_dev_count();
		for (i = 0; i < num_dev; i++)
			rte_eth_macaddr_get(i, &ports_eth_addr[i]);
	}

	for (i = 0; i < num_dev; i++)
		if (!memcmp(&ifr->ifr_addr.sa_data[0], &ports_eth_addr[i], ETH_ALEN))
			return i;

	return -1;
}
/*----------------------------------------------------------------------------*/
void
dpdk_destroy_handle(struct mtcp_thread_context *ctxt)
{
	struct dpdk_private_context *dpc;
	int i;

	dpc = (struct dpdk_private_context *) ctxt->io_private_context;

	/* free wmbufs */
	for (i = 0; i < g_config.mos->netdev_table->num; i++)
		free_pkts(dpc->wmbufs[i].m_table, MAX_PKT_BURST);

#ifdef ENABLE_STATS_IOCTL
	/* free fd */
	close(dpc->fd);
#endif /* !ENABLE_STATS_IOCTL */

	/* free it all up */
	free(dpc);
}
/*----------------------------------------------------------------------------*/
static void
check_all_ports_link_status(uint8_t port_num, uint32_t port_mask)
{
#define CHECK_INTERVAL 			100 /* 100ms */
#define MAX_CHECK_TIME 			90 /* 9s (90 * 100ms) in total */

	uint8_t portid, count, all_ports_up, print_flag = 0;
	struct rte_eth_link link;

	printf("\nChecking link status");
	fflush(stdout);
	for (count = 0; count <= MAX_CHECK_TIME; count++) {
		all_ports_up = 1;
		for (portid = 0; portid < port_num; portid++) {
			if ((port_mask & (1 << portid)) == 0)
				continue;
			memset(&link, 0, sizeof(link));
			rte_eth_link_get_nowait(portid, &link);
			/* print link status if flag set */
			if (print_flag == 1) {
				if (link.link_status)
					printf("Port %d Link Up - speed %u "
						"Mbps - %s\n", (uint8_t)portid,
						(unsigned)link.link_speed,
				(link.link_duplex == ETH_LINK_FULL_DUPLEX) ?
					("full-duplex") : ("half-duplex\n"));
				else
					printf("Port %d Link Down\n",
						(uint8_t)portid);
				continue;
			}
			/* clear all_ports_up flag if any link down */
			if (link.link_status == 0) {
				all_ports_up = 0;
				break;
			}
		}
		/* after finally printing all link status, get out */
		if (print_flag == 1)
			break;

		if (all_ports_up == 0) {
			printf(".");
			fflush(stdout);
			rte_delay_ms(CHECK_INTERVAL);
		}

		/* set the print_flag if all ports up or timeout */
		if (all_ports_up == 1 || count == (MAX_CHECK_TIME - 1)) {
			print_flag = 1;
			printf("done\n");
		}
	}
}
/*----------------------------------------------------------------------------*/
#if 0
static void
dpdk_enable_fdir(int portid, uint8_t is_master)
{
	struct rte_fdir_masks fdir_masks;
	struct rte_fdir_filter fdir_filter;
	int ret;

	memset(&fdir_filter, 0, sizeof(struct rte_fdir_filter));
	fdir_filter.iptype = RTE_FDIR_IPTYPE_IPV4;
	fdir_filter.l4type = RTE_FDIR_L4TYPE_TCP;
	fdir_filter.ip_dst.ipv4_addr = g_config.mos->netdev_table->ent[portid]->ip_addr;

	if (is_master) {
		memset(&fdir_masks, 0, sizeof(struct rte_fdir_masks));
		fdir_masks.src_ipv4_mask = 0x0;
		fdir_masks.dst_ipv4_mask = 0xFFFFFFFF;
		fdir_masks.src_port_mask = 0x0;
		fdir_masks.dst_port_mask = 0x0;

		/*
		 * enable the following if the filter is IP-only
		 * (non-TCP, non-UDP)
		 */
		/* fdir_masks.only_ip_flow = 1; */
		rte_eth_dev_fdir_set_masks(portid, &fdir_masks);
		ret = rte_eth_dev_fdir_add_perfect_filter(portid,
							  &fdir_filter,
							  0,
							  g_config.mos->multiprocess_curr_core,
							  0);
	} else {
		ret = rte_eth_dev_fdir_update_perfect_filter(portid,
							     &fdir_filter,
							     0,
							     g_config.mos->multiprocess_curr_core,
							     0);
	}
	if (ret < 0) {
		rte_exit(EXIT_FAILURE,
			 "fdir_add_perfect_filter_t call failed!: %d\n",
			 ret);
	}
	fprintf(stderr, "Filter for device ifidx: %d added\n", portid);
}
#endif
/*----------------------------------------------------------------------------*/
int32_t
dpdk_dev_ioctl(struct mtcp_thread_context *ctx, int nif, int cmd, void *argp)
{
	struct dpdk_private_context *dpc;
	struct rte_mbuf *m;
	int len_of_mbuf;
	struct iphdr *iph;
	struct tcphdr *tcph;
	RssInfo *rss_i;

	iph = (struct iphdr *)argp;
	dpc = (struct dpdk_private_context *)ctx->io_private_context;
	len_of_mbuf = dpc->wmbufs[nif].len;
	rss_i = NULL;

	switch (cmd) {
	case PKT_TX_IP_CSUM:
		m = dpc->wmbufs[nif].m_table[len_of_mbuf - 1];
		m->ol_flags = PKT_TX_IP_CKSUM | PKT_TX_IPV4;
		m->l2_len = sizeof(struct ether_hdr);
		m->l3_len = (iph->ihl<<2);
		break;
	case PKT_TX_TCP_CSUM:
		m = dpc->wmbufs[nif].m_table[len_of_mbuf - 1];
		tcph = (struct tcphdr *)((unsigned char *)iph + (iph->ihl<<2));
		m->ol_flags |= PKT_TX_TCP_CKSUM;
		tcph->check = rte_ipv4_phdr_cksum((struct ipv4_hdr *)iph, m->ol_flags);
		break;
	case PKT_RX_RSS:
		rss_i = (RssInfo *)argp;
		m = dpc->pkts_burst[rss_i->pktidx];
		rss_i->hash_value = m->hash.rss;
		break;
	default:
		goto dev_ioctl_err;
	}

	return 0;
 dev_ioctl_err:
	return -1;
}
/*----------------------------------------------------------------------------*/
void
dpdk_load_module_upper_half(void)
{
	int cpu = g_config.mos->num_cores, ret;
	uint32_t cpumask = 0;
	char cpumaskbuf[10];
	char mem_channels[5];

	/* set the log level */
	rte_set_log_type(RTE_LOGTYPE_PMD, 0);
	rte_set_log_type(RTE_LOGTYPE_MALLOC, 0);
	rte_set_log_type(RTE_LOGTYPE_MEMPOOL, 0);
	rte_set_log_type(RTE_LOGTYPE_RING, 0);
	rte_set_log_level(RTE_LOG_WARNING);

	/* get the cpu mask */
	for (ret = 0; ret < cpu; ret++)
		cpumask = (cpumask | (1 << ret));
	sprintf(cpumaskbuf, "%X", cpumask);

	/* get the mem channels per socket */
	if (g_config.mos->nb_mem_channels == 0) {
		TRACE_ERROR("DPDK module requires # of memory channels "
				"per socket parameter!\n");
		exit(EXIT_FAILURE);
	}
	sprintf(mem_channels, "%d", g_config.mos->nb_mem_channels);

	/* initialize the rte env first, what a waste of implementation effort!  */
	char *argv[] = {"",
			"-c",
			cpumaskbuf,
			"-n",
			mem_channels,
			"--proc-type=auto",
			""
	};
	const int argc = 6;

	/*
	 * re-set getopt extern variable optind.
	 * this issue was a bitch to debug
	 * rte_eal_init() internally uses getopt() syscall
	 * mtcp applications that also use an `external' getopt
	 * will cause a violent crash if optind is not reset to zero
	 * prior to calling the func below...
	 * see man getopt(3) for more details
	 */
	optind = 0;

	/* initialize the dpdk eal env */
	ret = rte_eal_init(argc, argv);
	if (ret < 0)
		rte_exit(EXIT_FAILURE, "Invalid EAL args!\n");

}
/*----------------------------------------------------------------------------*/
void
dpdk_load_module_lower_half(void)
{
	int portid, rxlcore_id, ret;
	struct rte_eth_fc_conf fc_conf;	/* for Ethernet flow control settings */
	/* setting the rss key */
	static const uint8_t key[] = {
		0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05,
		0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05,
		0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05,
		0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05
	};

	port_conf.rx_adv_conf.rss_conf.rss_key = (uint8_t *)&key;
	port_conf.rx_adv_conf.rss_conf.rss_key_len = sizeof(key);

	/* resetting cpu_qid mapping */
	memset(cpu_qid_map, 0xFF, sizeof(cpu_qid_map));

	if (!g_config.mos->multiprocess
			|| (g_config.mos->multiprocess && g_config.mos->multiprocess_is_master)) {
		for (rxlcore_id = 0; rxlcore_id < g_config.mos->num_cores; rxlcore_id++) {
			char name[20];
			sprintf(name, "mbuf_pool-%d", rxlcore_id);
			/* create the mbuf pools */
			pktmbuf_pool[rxlcore_id] =
				rte_mempool_create(name, NB_MBUF,
						   MBUF_SIZE, MEMPOOL_CACHE_SIZE,
						   sizeof(struct rte_pktmbuf_pool_private),
						   rte_pktmbuf_pool_init, NULL,
						   rte_pktmbuf_init, NULL,
						   rte_lcore_to_socket_id(rxlcore_id), 0);
			if (pktmbuf_pool[rxlcore_id] == NULL)
				rte_exit(EXIT_FAILURE, "Cannot init mbuf pool\n");
		}

		/* Initialise each port */
		for (portid = 0; portid < g_config.mos->netdev_table->num; portid++) {
			int num_queue = 0, eth_idx, i, queue_id;
			for (eth_idx = 0; eth_idx < g_config.mos->netdev_table->num; eth_idx++)
				if (portid == g_config.mos->netdev_table->ent[eth_idx]->ifindex)
					break;
			if (eth_idx == g_config.mos->netdev_table->num)
				continue;
			for (i = 0; i < sizeof(uint64_t) * 8; i++)
				if (g_config.mos->netdev_table->ent[eth_idx]->cpu_mask & (1L << i))
					num_queue++;

			/* set 'num_queues' (used for GetRSSCPUCore() in util.c) */
			num_queues = num_queue;

			/* init port */
			printf("Initializing port %u... ", (unsigned) portid);
			fflush(stdout);
			ret = rte_eth_dev_configure(portid, num_queue, num_queue,
										&port_conf);
			if (ret < 0)
				rte_exit(EXIT_FAILURE, "Cannot configure device:"
									   "err=%d, port=%u\n",
									   ret, (unsigned) portid);

			/* init one RX queue per CPU */
			fflush(stdout);
#ifdef DEBUG
			rte_eth_macaddr_get(portid, &ports_eth_addr[portid]);
#endif
			queue_id = 0;
			for (rxlcore_id = 0; rxlcore_id < g_config.mos->num_cores; rxlcore_id++) {
				if (!(g_config.mos->netdev_table->ent[eth_idx]->cpu_mask & (1L << rxlcore_id)))
					continue;
				ret = rte_eth_rx_queue_setup(portid, queue_id, nb_rxd,
						rte_eth_dev_socket_id(portid), &rx_conf,
						pktmbuf_pool[rxlcore_id]);
				if (ret < 0)
					rte_exit(EXIT_FAILURE, "rte_eth_rx_queue_setup:"
										   "err=%d, port=%u, queueid: %d\n",
										   ret, (unsigned) portid, rxlcore_id);
				cpu_qid_map[portid][rxlcore_id] = queue_id++;
			}

			/* init one TX queue on each port per CPU (this is redundant for
			 * this app) */
			fflush(stdout);
			queue_id = 0;
			for (rxlcore_id = 0; rxlcore_id < g_config.mos->num_cores; rxlcore_id++) {
				if (!(g_config.mos->netdev_table->ent[eth_idx]->cpu_mask & (1L << rxlcore_id)))
					continue;
				ret = rte_eth_tx_queue_setup(portid, queue_id++, nb_txd,
						rte_eth_dev_socket_id(portid), &tx_conf);
				if (ret < 0)
					rte_exit(EXIT_FAILURE, "rte_eth_tx_queue_setup:"
										   "err=%d, port=%u, queueid: %d\n",
										   ret, (unsigned) portid, rxlcore_id);
			}

			/* Start device */
			ret = rte_eth_dev_start(portid);
			if (ret < 0)
				rte_exit(EXIT_FAILURE, "rte_eth_dev_start:err=%d, port=%u\n",
									   ret, (unsigned) portid);

			printf("done: \n");
			rte_eth_promiscuous_enable(portid);

			/* retrieve current flow control settings per port */
			memset(&fc_conf, 0, sizeof(fc_conf));
			ret = rte_eth_dev_flow_ctrl_get(portid, &fc_conf);
			if (ret != 0) {
				rte_exit(EXIT_FAILURE, "Failed to get flow control info!\n");
			}

			/* and just disable the rx/tx flow control */
			fc_conf.mode = RTE_FC_NONE;
			ret = rte_eth_dev_flow_ctrl_set(portid, &fc_conf);
			if (ret != 0) {
				rte_exit(EXIT_FAILURE, "Failed to set flow control info!: errno: %d\n",
					 ret);
			}

#ifdef DEBUG
			printf("Port %u, MAC address: %02X:%02X:%02X:%02X:%02X:%02X\n\n",
					(unsigned) portid,
					ports_eth_addr[portid].addr_bytes[0],
					ports_eth_addr[portid].addr_bytes[1],
					ports_eth_addr[portid].addr_bytes[2],
					ports_eth_addr[portid].addr_bytes[3],
					ports_eth_addr[portid].addr_bytes[4],
					ports_eth_addr[portid].addr_bytes[5]);
#endif
#if 0
			/* if multi-process support is enabled, then turn on FDIR */
			if (g_config.mos->multiprocess)
				dpdk_enable_fdir(portid, g_config.mos->multiprocess_is_master);
#endif
		}
	} else { /* g_config.mos->multiprocess && !g_config.mos->multiprocess_is_master */
		for (rxlcore_id = 0; rxlcore_id < g_config.mos->num_cores; rxlcore_id++) {
			char name[20];
			sprintf(name, "mbuf_pool-%d", rxlcore_id);
			/* initialize the mbuf pools */
			pktmbuf_pool[rxlcore_id] =
				rte_mempool_lookup(name);
			if (pktmbuf_pool[rxlcore_id] == NULL)
				rte_exit(EXIT_FAILURE, "Cannot init mbuf pool\n");
		}
#if 0
		for (portid = 0; portid < g_config.mos->netdev_table->num; portid++)
			dpdk_enable_fdir(portid, g_config.mos->multiprocess_is_master);
#endif
	}

	check_all_ports_link_status(g_config.mos->netdev_table->num, 0xFFFFFFFF);
}
/*----------------------------------------------------------------------------*/
io_module_func dpdk_module_func = {
	.load_module_upper_half		   = dpdk_load_module_upper_half,
	.load_module_lower_half		   = dpdk_load_module_lower_half,
	.init_handle		   = dpdk_init_handle,
	.link_devices		   = NULL,
	.release_pkt		   = NULL,
	.send_pkts		   = dpdk_send_pkts,
	.get_wptr   		   = dpdk_get_wptr,
	.recv_pkts		   = dpdk_recv_pkts,
	.get_rptr	   	   = dpdk_get_rptr,
	.get_nif		   = dpdk_get_nif,
	.select			   = NULL,
	.destroy_handle		   = dpdk_destroy_handle,
	.dev_ioctl		   = dpdk_dev_ioctl,
	.set_wptr		   = dpdk_set_wptr,
};
/*----------------------------------------------------------------------------*/