static void app_lcore_arp_tx_gratuitous (struct app_lcore_params_io *lp) {
uint32_t i;
for (i = 0; i < lp->tx.n_nic_queues; i++) {
uint8_t port = lp->tx.nic_queues[i].port;
uint8_t queue = lp->tx.nic_queues[i].queue;
struct rte_mbuf *tmpbuf = rte_ctrlmbuf_alloc (app.pools[0]);
if (!tmpbuf) {
puts ("Error creating gratuitous ARP");
exit (-1);
}
tmpbuf->pkt_len = arppktlen;
tmpbuf->data_len = arppktlen;
tmpbuf->port = port;
memcpy (rte_ctrlmbuf_data (tmpbuf), arppkt, arppktlen);
rte_eth_macaddr_get (port, (struct ether_addr *)(rte_ctrlmbuf_data (tmpbuf) + 6));
rte_eth_macaddr_get (port,
(struct ether_addr *)(rte_ctrlmbuf_data (tmpbuf) + 6 + 6 + 2 + 8));
memcpy (rte_ctrlmbuf_data (tmpbuf) + 6 + 6 + 2 + 14, icmppkt + 6 + 6 + 2 + 4 * 4, 4);
if (!rte_eth_tx_burst (port, queue, &tmpbuf, 1)) {
puts ("Error sending gratuitous ARP");
exit (-1);
}
}
}
/* Update device bond info */
static void
dpdk_ethdev_bond_info_update(struct vr_dpdk_ethdev *ethdev)
{
int i, slave_port_id;
int port_id = ethdev->ethdev_port_id;
uint16_t mtu = 0;
struct rte_pci_addr *pci_addr;
struct ether_addr bond_mac, mac_addr;
struct ether_addr lacp_mac = { .addr_bytes = {0x01, 0x80, 0xc2, 0, 0, 0x02} };
if (rte_eth_bond_mode_get(port_id) == -1) {
ethdev->ethdev_nb_slaves = -1;
} else {
ethdev->ethdev_nb_slaves = rte_eth_bond_slaves_get(port_id,
ethdev->ethdev_slaves, sizeof(ethdev->ethdev_slaves));
memset(&mac_addr, 0, sizeof(bond_mac));
rte_eth_macaddr_get(port_id, &bond_mac);
RTE_LOG(INFO, VROUTER, " bond eth device %" PRIu8
" configured MAC " MAC_FORMAT "\n",
port_id, MAC_VALUE(bond_mac.addr_bytes));
/* log out and configure bond members */
for (i = 0; i < ethdev->ethdev_nb_slaves; i++) {
slave_port_id = ethdev->ethdev_slaves[i];
if (!rte_eth_devices[port_id].data->mtu) {
rte_eth_dev_get_mtu(slave_port_id, &mtu);
rte_eth_devices[port_id].data->mtu = mtu;
}
memset(&mac_addr, 0, sizeof(mac_addr));
rte_eth_macaddr_get(slave_port_id, &mac_addr);
pci_addr = &rte_eth_devices[slave_port_id].pci_dev->addr;
RTE_LOG(INFO, VROUTER, " bond member eth device %" PRIu8
" PCI " PCI_PRI_FMT " MAC " MAC_FORMAT "\n",
slave_port_id, pci_addr->domain, pci_addr->bus,
pci_addr->devid, pci_addr->function,
MAC_VALUE(mac_addr.addr_bytes));
/* try to add bond mac and LACP multicast MACs */
if (rte_eth_dev_mac_addr_add(slave_port_id, &bond_mac, 0) == 0
&& rte_eth_dev_set_mc_addr_list(slave_port_id, &lacp_mac, 1) == 0) {
/* disable the promisc mode enabled by default */
rte_eth_promiscuous_disable(ethdev->ethdev_port_id);
RTE_LOG(INFO, VROUTER, " bond member eth device %" PRIu8
" promisc mode disabled\n", slave_port_id);
} else {
RTE_LOG(INFO, VROUTER, " bond member eth device %" PRIu8
": unable to add MAC addresses\n", slave_port_id);
}
}
/* In LACP mode all the bond members are in the promisc mode
* by default (see bond_mode_8023ad_activate_slave()
* But we need also to put the bond interface in promisc to get
* the broadcasts. Seems to be a bug in bond_ethdev_rx_burst_8023ad()?
*/
rte_eth_promiscuous_enable(port_id);
}
}
/*
* Returns MAC address for port in a string
*/
static const char *
get_printable_mac_addr(uint8_t port)
{
static const char err_address[] = "00:00:00:00:00:00";
static char addresses[RTE_MAX_ETHPORTS][sizeof(err_address)] = {{0}};
struct ether_addr mac = {{0}};
int ether_addr_len = sizeof(err_address) - NEWLINE_CHAR_OFFSET;
int i = 0;
int j = 0;
if (unlikely(port >= RTE_MAX_ETHPORTS))
return err_address;
/* first time run for this port so we populate addresses */
if (unlikely(addresses[port][0] == '\0')) {
rte_eth_macaddr_get(port, &mac);
while(j < NUM_BYTES_MAC_ADDR) {
rte_snprintf(&addresses[port][0] + i,
MAC_ADDR_STR_INT_L + NEWLINE_CHAR_OFFSET,
"%02x:",
mac.addr_bytes[j]);
i += MAC_ADDR_STR_INT_L;
j++;
}
/* Overwrite last ":" and null terminate the string */
addresses[port][ether_addr_len] = '\0';
}
return addresses[port];
}
static void
bond_port_init(struct rte_mempool *mbuf_pool)
{
int retval;
uint8_t i;
retval = rte_eth_bond_create("bond0", BONDING_MODE_ALB,
0 /*SOCKET_ID_ANY*/);
if (retval < 0)
rte_exit(EXIT_FAILURE,
"Faled to create bond port\n");
BOND_PORT = (uint8_t)retval;
retval = rte_eth_dev_configure(BOND_PORT, 1, 1, &port_conf);
if (retval != 0)
rte_exit(EXIT_FAILURE, "port %u: configuration failed (res=%d)\n",
BOND_PORT, retval);
/* RX setup */
retval = rte_eth_rx_queue_setup(BOND_PORT, 0, RTE_RX_DESC_DEFAULT,
rte_eth_dev_socket_id(BOND_PORT), NULL,
mbuf_pool);
if (retval < 0)
rte_exit(retval, " port %u: RX queue 0 setup failed (res=%d)",
BOND_PORT, retval);
/* TX setup */
retval = rte_eth_tx_queue_setup(BOND_PORT, 0, RTE_TX_DESC_DEFAULT,
rte_eth_dev_socket_id(BOND_PORT), NULL);
if (retval < 0)
rte_exit(retval, "port %u: TX queue 0 setup failed (res=%d)",
BOND_PORT, retval);
for (i = 0; i < slaves_count; i++) {
if (rte_eth_bond_slave_add(BOND_PORT, slaves[i]) == -1)
rte_exit(-1, "Oooops! adding slave (%u) to bond (%u) failed!\n",
slaves[i], BOND_PORT);
}
retval = rte_eth_dev_start(BOND_PORT);
if (retval < 0)
rte_exit(retval, "Start port %d failed (res=%d)", BOND_PORT, retval);
rte_eth_promiscuous_enable(BOND_PORT);
struct ether_addr addr;
rte_eth_macaddr_get(BOND_PORT, &addr);
printf("Port %u MAC: ", (unsigned)BOND_PORT);
PRINT_MAC(addr);
printf("\n");
}
int
VIFHYPER_CREATE(const char *devstr, struct virtif_sc *vif_sc, uint8_t *enaddr,
struct virtif_user **viup)
{
struct rte_eth_conf portconf;
struct rte_eth_link link;
struct ether_addr ea;
struct virtif_user *viu;
int rv = EINVAL; /* XXX: not very accurate ;) */
viu = malloc(sizeof(*viu));
memset(viu, 0, sizeof(*viu));
viu->viu_devstr = strdup(devstr);
viu->viu_virtifsc = vif_sc;
/* this is here only for simplicity */
if ((rv = globalinit(viu)) != 0)
goto out;
memset(&portconf, 0, sizeof(portconf));
if ((rv = rte_eth_dev_configure(IF_PORTID,
NQUEUE, NQUEUE, &portconf)) < 0)
OUT("configure device");
if ((rv = rte_eth_rx_queue_setup(IF_PORTID,
0, NDESCRX, 0, &rxconf, mbpool_rx)) <0)
OUT("rx queue setup");
if ((rv = rte_eth_tx_queue_setup(IF_PORTID, 0, NDESCTX, 0, &txconf)) < 0)
OUT("tx queue setup");
if ((rv = rte_eth_dev_start(IF_PORTID)) < 0)
OUT("device start");
rte_eth_link_get(IF_PORTID, &link);
if (!link.link_status) {
ifwarn(viu, "link down");
}
rte_eth_promiscuous_enable(IF_PORTID);
rte_eth_macaddr_get(IF_PORTID, &ea);
memcpy(enaddr, ea.addr_bytes, ETHER_ADDR_LEN);
rv = pthread_create(&viu->viu_rcvpt, NULL, receiver, viu);
out:
/* XXX: well this isn't much of an unrolling ... */
if (rv != 0)
free(viu);
else
*viup = viu;
return rumpuser_component_errtrans(-rv);
}
void mg_distributor_apply_src_mac(struct rte_mbuf **pkts, uint8_t port_id, uint16_t n){
// TODO: it might be faster to read mac addr only once at output registration
struct ether_addr addr;
rte_eth_macaddr_get(port_id, &addr);
uint16_t i;
for(i=0;i<n;i++){
struct ether_hdr * ethhdr = rte_pktmbuf_mtod(*pkts, struct ether_hdr *);
ether_addr_copy(&addr, ðhdr->s_addr);
pkts++;
}
}
static inline int
port_init(uint16_t port, struct rte_mempool *mbuf_pool)
{
struct rte_eth_conf port_conf = port_conf_default;
const uint16_t rx_rings = 1, 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++) {
retval = rte_eth_tx_queue_setup(port, q, TX_RING_SIZE,
rte_eth_dev_socket_id(port), NULL);
if (retval < 0)
return retval;
}
/* Start the Ethernet port. */
retval = rte_eth_dev_start(port);
if (retval < 0)
return retval;
/* Display the port MAC address. */
struct ether_addr addr;
rte_eth_macaddr_get(port, &addr);
printf("Port %u MAC: %02" PRIx8 " %02" PRIx8 " %02" PRIx8
" %02" PRIx8 " %02" PRIx8 " %02" PRIx8 "\n",
(unsigned int)port,
addr.addr_bytes[0], addr.addr_bytes[1],
addr.addr_bytes[2], addr.addr_bytes[3],
addr.addr_bytes[4], addr.addr_bytes[5]);
/* Enable RX in promiscuous mode for the Ethernet device. */
rte_eth_promiscuous_enable(port);
return 0;
}
/**
* @brief Set device default configuration
*
* @param devId uint8_t, ID of DPDK device
* @param macAddr DevMacAddr_t*, pointer to buffer where MAC address will be stored
*
* @return DevMacAddr_t* pointer to buffer where MAC address is stored and NULL otherwice
*/
DevMacAddr_t* DPDKAdapter::getDevMACAddress(uint8_t devId, DevMacAddr_t* macAddr)
{
if(devId > RTE_MAX_ETHPORTS)
{
qCritical("Device ID is out of range");
return NULL;
}
rte_eth_macaddr_get(devId, macAddr);
return macAddr;
}
/*
* Initialises a given port using global settings and with the rx buffers
* coming from the mbuf_pool passed as parameter
*/
static inline int
port_init(uint8_t port, struct rte_mempool *mbuf_pool)
{
struct rte_eth_conf port_conf;
const uint16_t rxRings = ETH_VMDQ_DCB_NUM_QUEUES,
txRings = (uint16_t)rte_lcore_count();
const uint16_t rxRingSize = 128, txRingSize = 512;
int retval;
uint16_t q;
retval = get_eth_conf(&port_conf, num_pools);
if (retval < 0)
return retval;
if (port >= rte_eth_dev_count()) return -1;
retval = rte_eth_dev_configure(port, rxRings, txRings, &port_conf);
if (retval != 0)
return retval;
for (q = 0; q < rxRings; q ++) {
retval = rte_eth_rx_queue_setup(port, q, rxRingSize,
rte_eth_dev_socket_id(port),
NULL,
mbuf_pool);
if (retval < 0)
return retval;
}
for (q = 0; q < txRings; q ++) {
retval = rte_eth_tx_queue_setup(port, q, txRingSize,
rte_eth_dev_socket_id(port),
NULL);
if (retval < 0)
return retval;
}
retval = rte_eth_dev_start(port);
if (retval < 0)
return retval;
struct ether_addr addr;
rte_eth_macaddr_get(port, &addr);
printf("Port %u MAC: %02"PRIx8" %02"PRIx8" %02"PRIx8
" %02"PRIx8" %02"PRIx8" %02"PRIx8"\n",
(unsigned)port,
addr.addr_bytes[0], addr.addr_bytes[1], addr.addr_bytes[2],
addr.addr_bytes[3], addr.addr_bytes[4], addr.addr_bytes[5]);
return 0;
}
static void
slave_port_init(uint16_t portid, struct rte_mempool *mbuf_pool)
{
int retval;
uint16_t nb_rxd = RTE_RX_DESC_DEFAULT;
uint16_t nb_txd = RTE_TX_DESC_DEFAULT;
if (portid >= rte_eth_dev_count())
rte_exit(EXIT_FAILURE, "Invalid port\n");
retval = rte_eth_dev_configure(portid, 1, 1, &port_conf);
if (retval != 0)
rte_exit(EXIT_FAILURE, "port %u: configuration failed (res=%d)\n",
portid, retval);
retval = rte_eth_dev_adjust_nb_rx_tx_desc(portid, &nb_rxd, &nb_txd);
if (retval != 0)
rte_exit(EXIT_FAILURE, "port %u: rte_eth_dev_adjust_nb_rx_tx_desc "
"failed (res=%d)\n", portid, retval);
/* RX setup */
retval = rte_eth_rx_queue_setup(portid, 0, nb_rxd,
rte_eth_dev_socket_id(portid), NULL,
mbuf_pool);
if (retval < 0)
rte_exit(retval, " port %u: RX queue 0 setup failed (res=%d)",
portid, retval);
/* TX setup */
retval = rte_eth_tx_queue_setup(portid, 0, nb_txd,
rte_eth_dev_socket_id(portid), NULL);
if (retval < 0)
rte_exit(retval, "port %u: TX queue 0 setup failed (res=%d)",
portid, retval);
retval = rte_eth_dev_start(portid);
if (retval < 0)
rte_exit(retval,
"Start port %d failed (res=%d)",
portid, retval);
struct ether_addr addr;
rte_eth_macaddr_get(portid, &addr);
printf("Port %u MAC: ", portid);
PRINT_MAC(addr);
printf("\n");
}
void
init_receiver(unsigned core_id, unsigned in_port,
struct receiver_t *receiver) {
receiver->core_id = core_id;
receiver->in_port = in_port;
receiver->nb_handler = 0;
receiver->nb_polls = 0;
receiver->nb_rec = 0;
receiver->pkts_received = 0;
receiver->burst_buffer = rte_malloc(NULL, BURST_SIZE * sizeof(void*), 64);
rte_eth_macaddr_get(receiver->in_port, &receiver->mac);
}
/*
* Initialises a given port using global settings and with the rx buffers
* coming from the mbuf_pool passed as parameter
*/
static inline int
port_init(uint8_t port, struct rte_mempool *mbuf_pool)
{
struct rte_eth_conf port_conf = port_conf_default;
const uint16_t rx_rings = 1, tx_rings = 1;
int retval;
uint16_t q;
if (port >= rte_eth_dev_count())
return -1;
retval = rte_eth_dev_configure(port, rx_rings, tx_rings, &port_conf);
if (retval != 0)
return retval;
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;
}
for (q = 0; q < tx_rings; q++) {
retval = rte_eth_tx_queue_setup(port, q, TX_RING_SIZE,
rte_eth_dev_socket_id(port), NULL);
if (retval < 0)
return retval;
}
retval = rte_eth_dev_start(port);
if (retval < 0)
return retval;
struct ether_addr addr;
rte_eth_macaddr_get(port, &addr);
printf("Port %u MAC: %02"PRIx8" %02"PRIx8" %02"PRIx8
" %02"PRIx8" %02"PRIx8" %02"PRIx8"\n",
(unsigned)port,
addr.addr_bytes[0], addr.addr_bytes[1],
addr.addr_bytes[2], addr.addr_bytes[3],
addr.addr_bytes[4], addr.addr_bytes[5]);
rte_eth_promiscuous_enable(port);
return 0;
}
static inline int
configure_eth_port(uint8_t port_id)
{
struct ether_addr addr;
const uint16_t rxRings = 1, txRings = 1;
const uint8_t nb_ports = rte_eth_dev_count();
int ret;
uint16_t q;
if (port_id > nb_ports)
return -1;
ret = rte_eth_dev_configure(port_id, rxRings, txRings, &port_conf_default);
if (ret != 0)
return ret;
for (q = 0; q < rxRings; q++) {
ret = rte_eth_rx_queue_setup(port_id, q, RX_DESC_PER_QUEUE,
rte_eth_dev_socket_id(port_id), NULL,
mbuf_pool);
if (ret < 0)
return ret;
}
for (q = 0; q < txRings; q++) {
ret = rte_eth_tx_queue_setup(port_id, q, TX_DESC_PER_QUEUE,
rte_eth_dev_socket_id(port_id), NULL);
if (ret < 0)
return ret;
}
ret = rte_eth_dev_start(port_id);
if (ret < 0)
return ret;
rte_eth_macaddr_get(port_id, &addr);
printf("Port %u MAC: %02"PRIx8" %02"PRIx8" %02"PRIx8
" %02"PRIx8" %02"PRIx8" %02"PRIx8"\n",
(unsigned)port_id,
addr.addr_bytes[0], addr.addr_bytes[1],
addr.addr_bytes[2], addr.addr_bytes[3],
addr.addr_bytes[4], addr.addr_bytes[5]);
rte_eth_promiscuous_enable(port_id);
return 0;
}
void
port_infos_display(struct cmdline *cl, portid_t port_id)
{
struct ether_addr mac_addr;
struct rte_eth_link link;
int vlan_offload;
static const char *info_border = "=====================";
rte_eth_link_get_nowait(port_id, &link);
cmdline_printf(cl, "\n%s Infos for port %-2d %s\n", info_border,
port_id, info_border);
rte_eth_macaddr_get(port_id, &mac_addr);
print_ethaddr(cl, "MAC address: ", &mac_addr);
cmdline_printf(cl, "\nLink status: %s\n",
(link.link_status) ? ("up") : ("down"));
cmdline_printf(cl, "Link speed: %u Mbps\n", (unsigned)link.link_speed);
cmdline_printf(cl, "Link duplex: %s\n",
(link.link_duplex == ETH_LINK_FULL_DUPLEX)
? ("full-duplex")
: ("half-duplex"));
cmdline_printf(cl, "Promiscuous mode: %s\n",
rte_eth_promiscuous_get(port_id) ? "enabled"
: "disabled");
cmdline_printf(cl, "Allmulticast mode: %s\n",
rte_eth_allmulticast_get(port_id) ? "enabled"
: "disabled");
vlan_offload = rte_eth_dev_get_vlan_offload(port_id);
if (vlan_offload >= 0) {
cmdline_printf(cl, "VLAN offload: \n");
if (vlan_offload & ETH_VLAN_STRIP_OFFLOAD)
cmdline_printf(cl, " strip on \n");
else
cmdline_printf(cl, " strip off \n");
if (vlan_offload & ETH_VLAN_FILTER_OFFLOAD)
cmdline_printf(cl, " filter on \n");
else
cmdline_printf(cl, " filter off \n");
if (vlan_offload & ETH_VLAN_EXTEND_OFFLOAD)
cmdline_printf(cl, " qinq(extend) on \n");
else
cmdline_printf(cl, " qinq(extend) off \n");
}
}
static const char *
get_printable_mac_addr(uint16_t port)
{
static const char err_address[] = "00:00:00:00:00:00";
static char addresses[RTE_MAX_ETHPORTS][sizeof(err_address)];
if (unlikely(port >= RTE_MAX_ETHPORTS))
return err_address;
if (unlikely(addresses[port][0]=='\0')){
struct ether_addr mac;
rte_eth_macaddr_get(port, &mac);
snprintf(addresses[port], sizeof(addresses[port]),
"%02x:%02x:%02x:%02x:%02x:%02x\n",
mac.addr_bytes[0], mac.addr_bytes[1], mac.addr_bytes[2],
mac.addr_bytes[3], mac.addr_bytes[4], mac.addr_bytes[5]);
}
return addresses[port];
}
static int dpdk_main(int port_id, int argc, char* argv[])
{
struct rte_eth_dev_info dev_info;
unsigned nb_queues;
FILE* lfile;
uint8_t core_id;
int ret;
printf("In dpdk_main\n");
// Open the log file
lfile = fopen("./vrouter.log", "w");
// Program the rte log
rte_openlog_stream(lfile);
ret = rte_eal_init(argc, argv);
if (ret < 0) {
log_crit( "Invalid EAL parameters\n");
return -1;
}
log_info( "Programming cmd rings now!\n");
rx_event_fd = (int *) malloc(sizeof(int *) * rte_lcore_count());
if (!rx_event_fd) {
log_crit("Failed to allocate memory for rx event fd arrays\n");
return -ENOMEM;
}
rte_eth_macaddr_get(port_id, &port_eth_addr);
log_info("Port%d: MAC Address: ", port_id);
print_ethaddr(&port_eth_addr);
/* Determine the number of RX/TX pairs supported by NIC */
rte_eth_dev_info_get(port_id, &dev_info);
dev_info.pci_dev->intr_handle.type = RTE_INTR_HANDLE_VFIO_MSIX;
dev_info.pci_dev->intr_handle.max_intr =
dev_info.max_rx_queues + dev_info.max_tx_queues;
ret = rte_intr_efd_enable(&dev_info.pci_dev->intr_handle,
dev_info.max_rx_queues);
if (ret < 0) {
rte_exit(EXIT_FAILURE, "Failed to enable rx interrupts\n");
}
ret = rte_intr_enable(&dev_info.pci_dev->intr_handle);
if (ret < 0) {
rte_exit(EXIT_FAILURE, "Failed to enable interrupts\n");
}
ret = rte_eth_dev_configure(port_id, dev_info.max_rx_queues,
dev_info.max_tx_queues, &port_conf);
if (ret < 0) {
rte_exit(EXIT_FAILURE, "Failed to configure ethernet device\n");
}
/* For each RX/TX pair */
nb_queues = dev_info.max_tx_queues;
for (core_id = 0; core_id < nb_queues; core_id++) {
char s[64];
if (rte_lcore_is_enabled(core_id) == 0)
continue;
/* NUMA socket number */
unsigned socketid = rte_lcore_to_socket_id(core_id);
if (socketid >= NB_SOCKETS) {
log_crit( "Socket %d of lcore %u is out of range %d\n",
socketid, core_id, NB_SOCKETS);
return -EBADF;
}
/* Create memory pool */
if (pktmbuf_pool[socketid] == NULL) {
log_info("Creating mempool on %d of ~%lx bytes\n",
socketid, NB_MBUF * MBUF_SIZE);
printf("Creating mempool on %d of ~%lx bytes\n",
socketid, NB_MBUF * MBUF_SIZE);
snprintf(s, sizeof(s), "mbuf_pool_%d", socketid);
pktmbuf_pool[socketid] = rte_mempool_create(s,
NB_MBUF,
MBUF_SIZE,
MEMPOOL_CACHE_SIZE,
PKTMBUF_PRIV_SZ,
rte_pktmbuf_pool_init,
NULL,
rte_pktmbuf_init,
NULL,
socketid,
0);
if (!pktmbuf_pool[socketid]) {
log_crit( "Cannot init mbuf pool on socket %d\n", socketid);
return -ENOMEM;
}
}
/* Setup the TX queue */
ret = rte_eth_tx_queue_setup(port_id,
core_id,
RTE_TX_DESC_DEFAULT,
socketid,
&tx_conf);
if (ret < 0) {
log_crit( "Cannot initialize TX queue (%d)\n", core_id);
return -ENODEV;
}
/* Setup the RX queue */
ret = rte_eth_rx_queue_setup(port_id,
core_id,
RTE_RX_DESC_DEFAULT,
socketid,
&rx_conf,
pktmbuf_pool[socketid]);
if (ret < 0) {
log_crit( "Cannot initialize RX queue (%d)\n", core_id);
return -ENODEV;
}
/* Create the event fds for event notification */
lcore_cmd_event_fd[core_id] = eventfd(0, 0);
}
// Start the eth device
ret = rte_eth_dev_start(port_id);
if (ret < 0) {
log_crit( "rte_eth_dev_start: err=%d, port=%d\n", ret, core_id);
return -ENODEV;
}
// Put the device in promiscuous mode
rte_eth_promiscuous_enable(port_id);
// Wait for link up
//check_all_ports_link_status(1, 1u << port_id);
log_info( "Starting engines on every core\n");
rte_eal_mp_remote_launch(engine_loop, &dev_info, CALL_MASTER);
return 0;
}
static inline int
port_init_common(uint8_t port, const struct rte_eth_conf *port_conf,
struct rte_mempool *mp)
{
const uint16_t rx_ring_size = 512, tx_ring_size = 512;
int retval;
uint16_t q;
struct rte_eth_dev_info dev_info;
if (!rte_eth_dev_is_valid_port(port))
return -1;
retval = rte_eth_dev_configure(port, 0, 0, port_conf);
rte_eth_dev_info_get(port, &dev_info);
default_params.rx_rings = RTE_MIN(dev_info.max_rx_queues,
MAX_NUM_RX_QUEUE);
default_params.tx_rings = 1;
/* Configure the Ethernet device. */
retval = rte_eth_dev_configure(port, default_params.rx_rings,
default_params.tx_rings, port_conf);
if (retval != 0)
return retval;
for (q = 0; q < default_params.rx_rings; q++) {
retval = rte_eth_rx_queue_setup(port, q, rx_ring_size,
rte_eth_dev_socket_id(port), NULL, mp);
if (retval < 0)
return retval;
}
/* Allocate and set up 1 TX queue per Ethernet port. */
for (q = 0; q < default_params.tx_rings; q++) {
retval = rte_eth_tx_queue_setup(port, q, tx_ring_size,
rte_eth_dev_socket_id(port), NULL);
if (retval < 0)
return retval;
}
/* Start the Ethernet port. */
retval = rte_eth_dev_start(port);
if (retval < 0)
return retval;
/* Display the port MAC address. */
struct ether_addr addr;
rte_eth_macaddr_get(port, &addr);
printf("Port %u MAC: %02" PRIx8 " %02" PRIx8 " %02" PRIx8
" %02" PRIx8 " %02" PRIx8 " %02" PRIx8 "\n",
(unsigned int)port,
addr.addr_bytes[0], addr.addr_bytes[1],
addr.addr_bytes[2], addr.addr_bytes[3],
addr.addr_bytes[4], addr.addr_bytes[5]);
/* Enable RX in promiscuous mode for the Ethernet device. */
rte_eth_promiscuous_enable(port);
return 0;
}
static inline int
ntuple_filter_add(uint8_t port, const char *addr, uint8_t queue_id)
{
int ret = 0;
uint32_t ntuple_ip_addr;
ret = inet_pton(AF_INET, addr, &ntuple_ip_addr);
if (ret <= 0) {
if (ret == 0) {
printf("Error: %s is not in presentation format\n", addr);
} else if (ret == -1) {
perror("inet_pton");
}
return ret;
}
struct rte_eth_ntuple_filter filter = {
.flags = RTE_5TUPLE_FLAGS,
.dst_ip = ntuple_ip_addr, /* Big endian */
.dst_ip_mask = UINT32_MAX, /* Enable */
.src_ip = 0,
.src_ip_mask = 0, /* Disable */
.dst_port = 0,
.dst_port_mask = 0, /* Disable */
.src_port = 0,
.src_port_mask = 0, /* Disable */
.proto = 0,
.proto_mask = 0, /* Disable */
.tcp_flags = 0,
.priority = 1, /* Lowest */
.queue = queue_id,
};
return rte_eth_dev_filter_ctrl(port,
RTE_ETH_FILTER_NTUPLE,
RTE_ETH_FILTER_ADD,
&filter);
}
/*
* Initialises a given port using global settings and with the rx buffers
* coming from the mbuf_pool passed as parameter
*/
static inline int
port_init(uint8_t port, struct rte_mempool *mbuf_pool)
{
port_conf_default.fdir_conf = fdir_conf;
struct rte_eth_conf port_conf = port_conf_default;
const uint16_t rx_rings = 1, tx_rings = 1;
int retval;
uint16_t q;
if (port >= rte_eth_dev_count())
return -1;
port_conf.rxmode.hw_vlan_strip = 0;
retval = rte_eth_dev_configure(port, rx_rings, tx_rings, &port_conf);
if (retval != 0)
return retval;
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;
}
for (q = 0; q < tx_rings; q++) {
retval = rte_eth_tx_queue_setup(port, q, TX_RING_SIZE,
rte_eth_dev_socket_id(port), NULL);
if (retval < 0)
return retval;
}
retval = rte_eth_dev_start(port);
if (retval < 0)
return retval;
struct ether_addr addr;
rte_eth_macaddr_get(port, &addr);
printf("Port %u MAC: %02"PRIx8" %02"PRIx8" %02"PRIx8
" %02"PRIx8" %02"PRIx8" %02"PRIx8"\n",
(unsigned)port,
addr.addr_bytes[0], addr.addr_bytes[1],
addr.addr_bytes[2], addr.addr_bytes[3],
addr.addr_bytes[4], addr.addr_bytes[5]);
rte_eth_promiscuous_enable(port);
return 0;
}
/*----------------------------------------------------------------------------*/
int
SetNetEnv(char *dev_name_list, char *port_stat_list)
{
int eidx = 0;
int i, j;
int set_all_inf = (strncmp(dev_name_list, ALL_STRING, sizeof(ALL_STRING))==0);
TRACE_CONFIG("Loading interface setting\n");
CONFIG.eths = (struct eth_table *)
calloc(MAX_DEVICES, sizeof(struct eth_table));
if (!CONFIG.eths) {
TRACE_ERROR("Can't allocate space for CONFIG.eths\n");
exit(EXIT_FAILURE);
}
if (current_iomodule_func == &ps_module_func) {
#ifndef DISABLE_PSIO
struct ifreq ifr;
/* calculate num_devices now! */
num_devices = ps_list_devices(devices);
if (num_devices == -1) {
perror("ps_list_devices");
exit(EXIT_FAILURE);
}
/* Create socket */
int sock = socket(AF_INET, SOCK_DGRAM, IPPROTO_IP);
if (sock == -1) {
TRACE_ERROR("socket");
exit(EXIT_FAILURE);
}
/* To Do: Parse dev_name_list rather than use strstr */
for (i = 0; i < num_devices; i++) {
strcpy(ifr.ifr_name, devices[i].name);
/* getting interface information */
if (ioctl(sock, SIOCGIFFLAGS, &ifr) == 0) {
if (!set_all_inf && strstr(dev_name_list, ifr.ifr_name) == NULL)
continue;
/* Setting informations */
eidx = CONFIG.eths_num++;
strcpy(CONFIG.eths[eidx].dev_name, ifr.ifr_name);
CONFIG.eths[eidx].ifindex = devices[i].ifindex;
/* getting address */
if (ioctl(sock, SIOCGIFADDR, &ifr) == 0 ) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].ip_addr = *(uint32_t *)&sin;
}
if (ioctl(sock, SIOCGIFHWADDR, &ifr) == 0 ) {
for (j = 0; j < ETH_ALEN; j ++) {
CONFIG.eths[eidx].haddr[j] = ifr.ifr_addr.sa_data[j];
}
}
/* Net MASK */
if (ioctl(sock, SIOCGIFNETMASK, &ifr) == 0) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].netmask = *(uint32_t *)&sin;
}
/* add to attached devices */
for (j = 0; j < num_devices_attached; j++) {
if (devices_attached[j] == devices[i].ifindex) {
break;
}
}
devices_attached[num_devices_attached] = devices[i].ifindex;
num_devices_attached++;
} else {
perror("SIOCGIFFLAGS");
}
}
num_queues = GetNumQueues();
if (num_queues <= 0) {
TRACE_CONFIG("Failed to find NIC queues!\n");
close(sock);
return -1;
}
if (num_queues > num_cpus) {
TRACE_CONFIG("Too many NIC queues available.\n");
close(sock);
return -1;
}
close(sock);
#endif /* !PSIO_MODULE */
} else if (current_iomodule_func == &dpdk_module_func) {
#ifndef DISABLE_DPDK
int cpu = CONFIG.num_cores;
mpz_t _cpumask;
char cpumaskbuf[32] = "";
char mem_channels[8] = "";
char socket_mem_str[32] = "";
// int i;
int ret, socket_mem;
static struct ether_addr ports_eth_addr[RTE_MAX_ETHPORTS];
/* STEP 1: first determine CPU mask */
mpz_init(_cpumask);
if (!mpz_cmp(_cpumask, CONFIG._cpumask)) {
/* get the cpu mask */
for (ret = 0; ret < cpu; ret++)
mpz_setbit(_cpumask, ret);
gmp_sprintf(cpumaskbuf, "%ZX", _cpumask);
} else
gmp_sprintf(cpumaskbuf, "%ZX", CONFIG._cpumask);
mpz_clear(_cpumask);
/* STEP 2: determine memory channels per socket */
/* get the mem channels per socket */
if (CONFIG.num_mem_ch == 0) {
TRACE_ERROR("DPDK module requires # of memory channels "
"per socket parameter!\n");
exit(EXIT_FAILURE);
}
sprintf(mem_channels, "%d", CONFIG.num_mem_ch);
/* STEP 3: determine socket memory */
/* get socket memory threshold (in MB) */
socket_mem =
RTE_ALIGN_CEIL((unsigned long)ceil((CONFIG.num_cores *
(CONFIG.rcvbuf_size +
CONFIG.sndbuf_size +
sizeof(struct tcp_stream) +
sizeof(struct tcp_recv_vars) +
sizeof(struct tcp_send_vars) +
sizeof(struct fragment_ctx)) *
CONFIG.max_concurrency)/RTE_SOCKET_MEM_SHIFT),
RTE_CACHE_LINE_SIZE);
/* initialize the rte env, what a waste of implementation effort! */
int argc = 6;//8;
char *argv[RTE_ARGC_MAX] = {"",
"-c",
cpumaskbuf,
"-n",
mem_channels,
#if 0
"--socket-mem",
socket_mem_str,
#endif
"--proc-type=auto"
};
ret = probe_all_rte_devices(argv, &argc, dev_name_list);
/* STEP 4: build up socket mem parameter */
sprintf(socket_mem_str, "%d", socket_mem);
#if 0
char *smsptr = socket_mem_str + strlen(socket_mem_str);
for (i = 1; i < ret + 1; i++) {
sprintf(smsptr, ",%d", socket_mem);
smsptr += strlen(smsptr);
}
TRACE_DBG("socket_mem: %s\n", socket_mem_str);
#endif
/*
* 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;
#ifdef DEBUG
/* print argv's */
for (i = 0; i < argc; i++)
TRACE_INFO("argv[%d]: %s\n", i, argv[i]);
#endif
/* initialize the dpdk eal env */
ret = rte_eal_init(argc, argv);
if (ret < 0) {
TRACE_ERROR("Invalid EAL args!\n");
exit(EXIT_FAILURE);
}
/* give me the count of 'detected' ethernet ports */
#if RTE_VERSION < RTE_VERSION_NUM(18, 5, 0, 0)
num_devices = rte_eth_dev_count();
#else
num_devices = rte_eth_dev_count_avail();
#endif
if (num_devices == 0) {
TRACE_ERROR("No Ethernet port!\n");
exit(EXIT_FAILURE);
}
/* get mac addr entries of 'detected' dpdk ports */
for (ret = 0; ret < num_devices; ret++)
rte_eth_macaddr_get(ret, &ports_eth_addr[ret]);
num_queues = MIN(CONFIG.num_cores, MAX_CPUS);
struct ifaddrs *ifap;
struct ifaddrs *iter_if;
char *seek;
if (getifaddrs(&ifap) != 0) {
perror("getifaddrs: ");
exit(EXIT_FAILURE);
}
iter_if = ifap;
do {
if (iter_if->ifa_addr && iter_if->ifa_addr->sa_family == AF_INET &&
!set_all_inf &&
(seek=strstr(dev_name_list, iter_if->ifa_name)) != NULL &&
/* check if the interface was not aliased */
*(seek + strlen(iter_if->ifa_name)) != ':') {
struct ifreq ifr;
/* Setting informations */
eidx = CONFIG.eths_num++;
strcpy(CONFIG.eths[eidx].dev_name, iter_if->ifa_name);
strcpy(ifr.ifr_name, iter_if->ifa_name);
/* Create socket */
int sock = socket(AF_INET, SOCK_DGRAM, IPPROTO_IP);
if (sock == -1) {
perror("socket");
exit(EXIT_FAILURE);
}
/* getting address */
if (ioctl(sock, SIOCGIFADDR, &ifr) == 0 ) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].ip_addr = *(uint32_t *)&sin;
}
if (ioctl(sock, SIOCGIFHWADDR, &ifr) == 0 ) {
for (j = 0; j < ETH_ALEN; j ++) {
CONFIG.eths[eidx].haddr[j] = ifr.ifr_addr.sa_data[j];
}
}
/* Net MASK */
if (ioctl(sock, SIOCGIFNETMASK, &ifr) == 0) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].netmask = *(uint32_t *)&sin;
}
close(sock);
for (j = 0; j < num_devices; j++) {
if (!memcmp(&CONFIG.eths[eidx].haddr[0], &ports_eth_addr[j],
ETH_ALEN))
CONFIG.eths[eidx].ifindex = j;
}
/* add to attached devices */
for (j = 0; j < num_devices_attached; j++) {
if (devices_attached[j] == CONFIG.eths[eidx].ifindex) {
break;
}
}
devices_attached[num_devices_attached] = CONFIG.eths[eidx].ifindex;
num_devices_attached++;
fprintf(stderr, "Total number of attached devices: %d\n",
num_devices_attached);
fprintf(stderr, "Interface name: %s\n",
iter_if->ifa_name);
}
iter_if = iter_if->ifa_next;
} while (iter_if != NULL);
freeifaddrs(ifap);
#if 0
/*
* XXX: It seems that there is a bug in the RTE SDK.
* The dynamically allocated rte_argv params are left
* as dangling pointers. Freeing them causes program
* to crash.
*/
/* free up all resources */
for (; rte_argc >= 9; rte_argc--) {
if (rte_argv[rte_argc] != NULL) {
fprintf(stderr, "Cleaning up rte_argv[%d]: %s (%p)\n",
rte_argc, rte_argv[rte_argc], rte_argv[rte_argc]);
free(rte_argv[rte_argc]);
rte_argv[rte_argc] = NULL;
}
}
#endif
/* check if process is primary or secondary */
CONFIG.multi_process_is_master = (eal_proc_type_detect() == RTE_PROC_PRIMARY) ?
1 : 0;
#endif /* !DISABLE_DPDK */
} else if (current_iomodule_func == &netmap_module_func) {
#ifndef DISABLE_NETMAP
struct ifaddrs *ifap;
struct ifaddrs *iter_if;
char *seek;
num_queues = MIN(CONFIG.num_cores, MAX_CPUS);
if (getifaddrs(&ifap) != 0) {
perror("getifaddrs: ");
exit(EXIT_FAILURE);
}
iter_if = ifap;
do {
if (iter_if->ifa_addr && iter_if->ifa_addr->sa_family == AF_INET &&
!set_all_inf &&
(seek=strstr(dev_name_list, iter_if->ifa_name)) != NULL &&
/* check if the interface was not aliased */
*(seek + strlen(iter_if->ifa_name)) != ':') {
struct ifreq ifr;
/* Setting informations */
eidx = CONFIG.eths_num++;
strcpy(CONFIG.eths[eidx].dev_name, iter_if->ifa_name);
strcpy(ifr.ifr_name, iter_if->ifa_name);
/* Create socket */
int sock = socket(AF_INET, SOCK_DGRAM, IPPROTO_IP);
if (sock == -1) {
perror("socket");
exit(EXIT_FAILURE);
}
/* getting address */
if (ioctl(sock, SIOCGIFADDR, &ifr) == 0 ) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].ip_addr = *(uint32_t *)&sin;
}
if (ioctl(sock, SIOCGIFHWADDR, &ifr) == 0 ) {
for (j = 0; j < ETH_ALEN; j ++) {
CONFIG.eths[eidx].haddr[j] = ifr.ifr_addr.sa_data[j];
}
}
/* Net MASK */
if (ioctl(sock, SIOCGIFNETMASK, &ifr) == 0) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].netmask = *(uint32_t *)&sin;
}
close(sock);
#if 0
for (j = 0; j < num_devices; j++) {
if (!memcmp(&CONFIG.eths[eidx].haddr[0], &ports_eth_addr[j],
ETH_ALEN))
CONFIG.eths[eidx].ifindex = ifr.ifr_ifindex;
#endif
CONFIG.eths[eidx].ifindex = eidx;
TRACE_INFO("Ifindex of interface %s is: %d\n",
ifr.ifr_name, CONFIG.eths[eidx].ifindex);
#if 0
}
#endif
/* add to attached devices */
for (j = 0; j < num_devices_attached; j++) {
if (devices_attached[j] == CONFIG.eths[eidx].ifindex) {
break;
}
}
devices_attached[num_devices_attached] = if_nametoindex(ifr.ifr_name);
num_devices_attached++;
fprintf(stderr, "Total number of attached devices: %d\n",
num_devices_attached);
fprintf(stderr, "Interface name: %s\n",
iter_if->ifa_name);
}
iter_if = iter_if->ifa_next;
} while (iter_if != NULL);
freeifaddrs(ifap);
#endif /* !DISABLE_NETMAP */
} else if (current_iomodule_func == &onvm_module_func) {
#ifdef ENABLE_ONVM
int cpu = CONFIG.num_cores;
mpz_t cpumask;
char cpumaskbuf[32];
char mem_channels[8];
char service[6];
char instance[6];
int ret;
mpz_init(cpumask);
/* get the cpu mask */
for (ret = 0; ret < cpu; ret++)
mpz_setbit(cpumask, ret);
gmp_sprintf(cpumaskbuf, "%ZX", cpumask);
mpz_clear(cpumask);
/* get the mem channels per socket */
if (CONFIG.num_mem_ch == 0) {
TRACE_ERROR("DPDK module requires # of memory channels "
"per socket parameter!\n");
exit(EXIT_FAILURE);
}
sprintf(mem_channels, "%d", CONFIG.num_mem_ch);
sprintf(service, "%d", CONFIG.onvm_serv);
sprintf(instance, "%d", CONFIG.onvm_inst);
/* initialize the rte env first, what a waste of implementation effort! */
char *argv[] = {"",
"-c",
cpumaskbuf,
"-n",
mem_channels,
"--proc-type=secondary",
"--",
"-r",
service,
instance,
""
};
const int argc = 10;
/*
* 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 = onvm_nflib_init(argc, argv, "mtcp_nf", &CONFIG.nf_info);
if (ret < 0) {
TRACE_ERROR("Invalid EAL args!\n");
exit(EXIT_FAILURE);
}
/* give me the count of 'detected' ethernet ports */
num_devices = ports->num_ports;
if (num_devices == 0) {
TRACE_ERROR("No Ethernet port!\n");
exit(EXIT_FAILURE);
}
num_queues = MIN(CONFIG.num_cores, MAX_CPUS);
struct ifaddrs *ifap;
struct ifaddrs *iter_if;
char *seek;
if (getifaddrs(&ifap) != 0) {
perror("getifaddrs: ");
exit(EXIT_FAILURE);
}
iter_if = ifap;
do {
if (iter_if->ifa_addr && iter_if->ifa_addr->sa_family == AF_INET &&
!set_all_inf &&
(seek=strstr(dev_name_list, iter_if->ifa_name)) != NULL &&
/* check if the interface was not aliased */
*(seek + strlen(iter_if->ifa_name)) != ':') {
struct ifreq ifr;
/* Setting informations */
eidx = CONFIG.eths_num++;
strcpy(CONFIG.eths[eidx].dev_name, iter_if->ifa_name);
strcpy(ifr.ifr_name, iter_if->ifa_name);
/* Create socket */
int sock = socket(AF_INET, SOCK_DGRAM, IPPROTO_IP);
if (sock == -1) {
perror("socket");
}
/* getting address */
if (ioctl(sock, SIOCGIFADDR, &ifr) == 0 ) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].ip_addr = *(uint32_t *)&sin;
}
for (j = 0; j < ETH_ALEN; j ++) {
CONFIG.eths[eidx].haddr[j] = ports->mac[eidx].addr_bytes[j];
};
/* Net MASK */
if (ioctl(sock, SIOCGIFNETMASK, &ifr) == 0) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].netmask = *(uint32_t *)&sin;
}
close(sock);
CONFIG.eths[eidx].ifindex = ports->id[eidx];
devices_attached[num_devices_attached] = CONFIG.eths[eidx].ifindex;
num_devices_attached++;
fprintf(stderr, "Total number of attached devices: %d\n",
num_devices_attached);
fprintf(stderr, "Interface name: %s\n",
iter_if->ifa_name);
}
iter_if = iter_if->ifa_next;
} while (iter_if != NULL);
freeifaddrs(ifap);
#endif /* ENABLE_ONVM */
}
CONFIG.nif_to_eidx = (int*)calloc(MAX_DEVICES, sizeof(int));
if (!CONFIG.nif_to_eidx) {
exit(EXIT_FAILURE);
}
for (i = 0; i < MAX_DEVICES; ++i) {
CONFIG.nif_to_eidx[i] = -1;
}
for (i = 0; i < CONFIG.eths_num; ++i) {
j = CONFIG.eths[i].ifindex;
if (j >= MAX_DEVICES) {
TRACE_ERROR("ifindex of eths_%d exceed the limit: %d\n", i, j);
exit(EXIT_FAILURE);
}
/* the physic port index of the i-th port listed in the config file is j*/
CONFIG.nif_to_eidx[j] = i;
/* finally set the port stats option `on' */
if (strcmp(CONFIG.eths[i].dev_name, port_stat_list) == 0)
CONFIG.eths[i].stat_print = TRUE;
}
return 0;
}
int
VIFHYPER_CREATE(const char *devstr, struct virtif_sc *vif_sc, uint8_t *enaddr,
struct virtif_user **viup)
{
struct rte_eth_conf portconf;
struct rte_eth_link link;
struct ether_addr ea;
struct virtif_user *viu;
unsigned long tmp;
char *ep;
int rv = EINVAL; /* XXX: not very accurate ;) */
viu = malloc(sizeof(*viu));
memset(viu, 0, sizeof(*viu));
viu->viu_devstr = strdup(devstr);
viu->viu_virtifsc = vif_sc;
tmp = strtoul(devstr, &ep, 10);
if (*ep != '\0')
OUT("invalid dev string");
if (tmp > 255)
OUT("DPDK port id out of range");
viu->viu_port_id = tmp;
if (viu->viu_port_id >= rte_eth_dev_count())
{
rv = -ENODEV;
OUT("DPDK port not initialized");
}
memset(&portconf, 0, sizeof(portconf));
if ((rv = rte_eth_dev_configure(viu->viu_port_id,
NQUEUE, NQUEUE, &portconf)) < 0)
OUT("configure device");
if ((rv = rte_eth_rx_queue_setup(viu->viu_port_id,
0, NDESCRX, 0, &rxconf, mbpool_rx)) <0)
OUT("rx queue setup");
if ((rv = rte_eth_tx_queue_setup(viu->viu_port_id, 0, NDESCTX, 0, &txconf)) < 0)
OUT("tx queue setup");
if ((rv = rte_eth_dev_start(viu->viu_port_id)) < 0)
OUT("device start");
rte_eth_link_get(viu->viu_port_id, &link);
if (!link.link_status) {
ifwarn(viu, "link down");
}
rte_eth_promiscuous_enable(viu->viu_port_id);
rte_eth_macaddr_get(viu->viu_port_id, &ea);
memcpy(enaddr, ea.addr_bytes, ETHER_ADDR_LEN);
rv = pthread_create(&viu->viu_rcvpt, NULL, receiver, viu);
out:
/* XXX: well this isn't much of an unrolling ... */
if (rv != 0)
free(viu);
else
*viup = viu;
return rumpuser_component_errtrans(-rv);
}
/*
* Initialises a given port using global settings and with the rx buffers
* coming from the mbuf_pool passed as parameter
*/
static inline int
port_init(uint16_t port, struct rte_mempool *mbuf_pool)
{
struct rte_eth_dev_info dev_info;
struct rte_eth_conf port_conf = {0};
uint16_t rxRingSize = RTE_TEST_RX_DESC_DEFAULT;
uint16_t txRingSize = RTE_TEST_TX_DESC_DEFAULT;
int retval;
uint16_t q;
uint16_t queues_per_pool;
uint32_t max_nb_pools;
/*
* The max pool number from dev_info will be used to validate the pool
* number specified in cmd line
*/
rte_eth_dev_info_get(port, &dev_info);
max_nb_pools = (uint32_t)dev_info.max_vmdq_pools;
/*
* We allow to process part of VMDQ pools specified by num_pools in
* command line.
*/
if (num_pools > max_nb_pools) {
printf("num_pools %d >max_nb_pools %d\n",
num_pools, max_nb_pools);
return -1;
}
/*
* NIC queues are divided into pf queues and vmdq queues.
* There is assumption here all ports have the same configuration!
*/
vmdq_queue_base = dev_info.vmdq_queue_base;
vmdq_pool_base = dev_info.vmdq_pool_base;
printf("vmdq queue base: %d pool base %d\n",
vmdq_queue_base, vmdq_pool_base);
if (vmdq_pool_base == 0) {
num_vmdq_queues = dev_info.max_rx_queues;
num_queues = dev_info.max_rx_queues;
if (num_tcs != num_vmdq_queues / num_pools) {
printf("nb_tcs %d is invalid considering with"
" nb_pools %d, nb_tcs * nb_pools should = %d\n",
num_tcs, num_pools, num_vmdq_queues);
return -1;
}
} else {
queues_per_pool = dev_info.vmdq_queue_num /
dev_info.max_vmdq_pools;
if (num_tcs > queues_per_pool) {
printf("num_tcs %d > num of queues per pool %d\n",
num_tcs, queues_per_pool);
return -1;
}
num_vmdq_queues = num_pools * queues_per_pool;
num_queues = vmdq_queue_base + num_vmdq_queues;
printf("Configured vmdq pool num: %u,"
" each vmdq pool has %u queues\n",
num_pools, queues_per_pool);
}
if (port >= rte_eth_dev_count())
return -1;
retval = get_eth_conf(&port_conf);
if (retval < 0)
return retval;
/*
* Though in this example, all queues including pf queues are setup.
* This is because VMDQ queues doesn't always start from zero, and the
* PMD layer doesn't support selectively initialising part of rx/tx
* queues.
*/
retval = rte_eth_dev_configure(port, num_queues, num_queues, &port_conf);
if (retval != 0)
return retval;
retval = rte_eth_dev_adjust_nb_rx_tx_desc(port, &rxRingSize,
&txRingSize);
if (retval != 0)
return retval;
if (RTE_MAX(rxRingSize, txRingSize) >
RTE_MAX(RTE_TEST_RX_DESC_DEFAULT, RTE_TEST_TX_DESC_DEFAULT)) {
printf("Mbuf pool has an insufficient size for port %u.\n",
port);
return -1;
}
for (q = 0; q < num_queues; q++) {
retval = rte_eth_rx_queue_setup(port, q, rxRingSize,
rte_eth_dev_socket_id(port),
NULL,
mbuf_pool);
if (retval < 0) {
printf("initialize rx queue %d failed\n", q);
return retval;
}
}
for (q = 0; q < num_queues; q++) {
retval = rte_eth_tx_queue_setup(port, q, txRingSize,
rte_eth_dev_socket_id(port),
NULL);
if (retval < 0) {
printf("initialize tx queue %d failed\n", q);
return retval;
}
}
retval = rte_eth_dev_start(port);
if (retval < 0) {
printf("port %d start failed\n", port);
return retval;
}
rte_eth_macaddr_get(port, &vmdq_ports_eth_addr[port]);
printf("Port %u MAC: %02"PRIx8" %02"PRIx8" %02"PRIx8
" %02"PRIx8" %02"PRIx8" %02"PRIx8"\n",
(unsigned)port,
vmdq_ports_eth_addr[port].addr_bytes[0],
vmdq_ports_eth_addr[port].addr_bytes[1],
vmdq_ports_eth_addr[port].addr_bytes[2],
vmdq_ports_eth_addr[port].addr_bytes[3],
vmdq_ports_eth_addr[port].addr_bytes[4],
vmdq_ports_eth_addr[port].addr_bytes[5]);
/* Set mac for each pool.*/
for (q = 0; q < num_pools; q++) {
struct ether_addr mac;
mac = pool_addr_template;
mac.addr_bytes[4] = port;
mac.addr_bytes[5] = q;
printf("Port %u vmdq pool %u set mac %02x:%02x:%02x:%02x:%02x:%02x\n",
port, q,
mac.addr_bytes[0], mac.addr_bytes[1],
mac.addr_bytes[2], mac.addr_bytes[3],
mac.addr_bytes[4], mac.addr_bytes[5]);
retval = rte_eth_dev_mac_addr_add(port, &mac,
q + vmdq_pool_base);
if (retval) {
printf("mac addr add failed at pool %d\n", q);
return retval;
}
}
return 0;
}
/**
* Initialises a given port using global settings and with the rx buffers
* coming from the mbuf_pool passed as parameter
*/
int
vxlan_port_init(uint8_t port, struct rte_mempool *mbuf_pool)
{
int retval;
uint16_t q;
struct rte_eth_dev_info dev_info;
uint16_t rx_rings, tx_rings = (uint16_t)rte_lcore_count();
const uint16_t rx_ring_size = RTE_TEST_RX_DESC_DEFAULT;
const uint16_t tx_ring_size = RTE_TEST_TX_DESC_DEFAULT;
struct rte_eth_udp_tunnel tunnel_udp;
struct rte_eth_rxconf *rxconf;
struct rte_eth_txconf *txconf;
struct vxlan_conf *pconf = &vxdev;
pconf->dst_port = udp_port;
rte_eth_dev_info_get(port, &dev_info);
if (dev_info.max_rx_queues > MAX_QUEUES) {
rte_exit(EXIT_FAILURE,
"please define MAX_QUEUES no less than %u in %s\n",
dev_info.max_rx_queues, __FILE__);
}
rxconf = &dev_info.default_rxconf;
txconf = &dev_info.default_txconf;
txconf->txq_flags = 0;
if (port >= rte_eth_dev_count())
return -1;
rx_rings = nb_devices;
/* Configure ethernet device. */
retval = rte_eth_dev_configure(port, rx_rings, tx_rings, &port_conf);
if (retval != 0)
return retval;
/* Setup the queues. */
for (q = 0; q < rx_rings; q++) {
retval = rte_eth_rx_queue_setup(port, q, rx_ring_size,
rte_eth_dev_socket_id(port),
rxconf,
mbuf_pool);
if (retval < 0)
return retval;
}
for (q = 0; q < tx_rings; q++) {
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 device. */
retval = rte_eth_dev_start(port);
if (retval < 0)
return retval;
/* Configure UDP port for UDP tunneling */
tunnel_udp.udp_port = udp_port;
tunnel_udp.prot_type = RTE_TUNNEL_TYPE_VXLAN;
retval = rte_eth_dev_udp_tunnel_port_add(port, &tunnel_udp);
if (retval < 0)
return retval;
rte_eth_macaddr_get(port, &ports_eth_addr[port]);
RTE_LOG(INFO, PORT, "Port %u MAC: %02"PRIx8" %02"PRIx8" %02"PRIx8
" %02"PRIx8" %02"PRIx8" %02"PRIx8"\n",
(unsigned)port,
ports_eth_addr[port].addr_bytes[0],
ports_eth_addr[port].addr_bytes[1],
ports_eth_addr[port].addr_bytes[2],
ports_eth_addr[port].addr_bytes[3],
ports_eth_addr[port].addr_bytes[4],
ports_eth_addr[port].addr_bytes[5]);
if (tso_segsz != 0) {
struct rte_eth_dev_info dev_info;
rte_eth_dev_info_get(port, &dev_info);
if ((dev_info.tx_offload_capa & DEV_TX_OFFLOAD_TCP_TSO) == 0)
RTE_LOG(WARNING, PORT,
"hardware TSO offload is not supported\n");
}
return 0;
}
static int
paxos_rx_process(struct rte_mbuf *pkt, struct proposer* proposer)
{
int ret = 0;
uint8_t l4_proto = 0;
uint16_t outer_header_len;
union tunnel_offload_info info = { .data = 0 };
struct udp_hdr *udp_hdr;
struct paxos_hdr *paxos_hdr;
struct ether_hdr *phdr = rte_pktmbuf_mtod(pkt, struct ether_hdr *);
parse_ethernet(phdr, &info, &l4_proto);
if (l4_proto != IPPROTO_UDP)
return -1;
udp_hdr = (struct udp_hdr *)((char *)phdr +
info.outer_l2_len + info.outer_l3_len);
/* if UDP dst port is not either PROPOSER or LEARNER port */
if (!(udp_hdr->dst_port == rte_cpu_to_be_16(PROPOSER_PORT) ||
udp_hdr->dst_port == rte_cpu_to_be_16(LEARNER_PORT)) &&
(pkt->packet_type & RTE_PTYPE_TUNNEL_MASK) == 0)
return -1;
paxos_hdr = (struct paxos_hdr *)((char *)udp_hdr + sizeof(struct udp_hdr));
if (rte_get_log_level() == RTE_LOG_DEBUG) {
//rte_hexdump(stdout, "udp", udp_hdr, sizeof(struct udp_hdr));
//rte_hexdump(stdout, "paxos", paxos_hdr, sizeof(struct paxos_hdr));
print_paxos_hdr(paxos_hdr);
}
int value_len = rte_be_to_cpu_16(paxos_hdr->value_len);
struct paxos_value *v = paxos_value_new((char *)paxos_hdr->paxosval, value_len);
switch(rte_be_to_cpu_16(paxos_hdr->msgtype)) {
case PAXOS_PROMISE: {
struct paxos_promise promise = {
.iid = rte_be_to_cpu_32(paxos_hdr->inst),
.ballot = rte_be_to_cpu_16(paxos_hdr->rnd),
.value_ballot = rte_be_to_cpu_16(paxos_hdr->vrnd),
.aid = rte_be_to_cpu_16(paxos_hdr->acptid),
.value = *v
};
proposer_handle_promise(proposer, &promise);
break;
}
case PAXOS_ACCEPT: {
if (first_time) {
proposer_preexecute(proposer);
first_time = false;
}
struct paxos_accept acpt = {
.iid = rte_be_to_cpu_32(paxos_hdr->inst),
.ballot = rte_be_to_cpu_16(paxos_hdr->rnd),
.value_ballot = rte_be_to_cpu_16(paxos_hdr->vrnd),
.aid = rte_be_to_cpu_16(paxos_hdr->acptid),
.value = *v
};
proposer_handle_accept(proposer, &acpt);
break;
}
case PAXOS_ACCEPTED: {
struct paxos_accepted ack = {
.iid = rte_be_to_cpu_32(paxos_hdr->inst),
.ballot = rte_be_to_cpu_16(paxos_hdr->rnd),
.value_ballot = rte_be_to_cpu_16(paxos_hdr->vrnd),
.aid = rte_be_to_cpu_16(paxos_hdr->acptid),
.value = *v
};
proposer_handle_accepted(proposer, &ack);
break;
}
default:
break;
}
outer_header_len = info.outer_l2_len + info.outer_l3_len
+ sizeof(struct udp_hdr) + sizeof(struct paxos_hdr);
rte_pktmbuf_adj(pkt, outer_header_len);
return ret;
}
static uint16_t
add_timestamps(uint8_t port __rte_unused, uint16_t qidx __rte_unused,
struct rte_mbuf **pkts, uint16_t nb_pkts,
uint16_t max_pkts __rte_unused, void *user_param)
{
struct proposer* proposer = (struct proposer *)user_param;
unsigned i;
uint64_t now = rte_rdtsc();
for (i = 0; i < nb_pkts; i++) {
pkts[i]->udata64 = now;
paxos_rx_process(pkts[i], proposer);
}
return nb_pkts;
}
static inline int
port_init(uint8_t port, struct rte_mempool *mbuf_pool, struct proposer* proposer)
{
struct rte_eth_dev_info dev_info;
struct rte_eth_txconf *txconf;
struct rte_eth_rxconf *rxconf;
struct rte_eth_conf port_conf = port_conf_default;
const uint16_t rx_rings = 1, tx_rings = 1;
int retval;
uint16_t q;
rte_eth_dev_info_get(port, &dev_info);
rxconf = &dev_info.default_rxconf;
txconf = &dev_info.default_txconf;
txconf->txq_flags &= PKT_TX_IPV4;
txconf->txq_flags &= PKT_TX_UDP_CKSUM;
if (port >= rte_eth_dev_count())
return -1;
retval = rte_eth_dev_configure(port, rx_rings, tx_rings, &port_conf);
if (retval != 0)
return retval;
for (q = 0; q < rx_rings; q++) {
retval = rte_eth_rx_queue_setup(port, q, RX_RING_SIZE,
rte_eth_dev_socket_id(port), rxconf, mbuf_pool);
if (retval < 0)
return retval;
}
for (q = 0; q < tx_rings; q++) {
retval = rte_eth_tx_queue_setup(port, q, TX_RING_SIZE,
rte_eth_dev_socket_id(port), txconf);
if (retval < 0)
return retval;
}
retval = rte_eth_dev_start(port);
if (retval < 0)
return retval;
struct ether_addr addr;
rte_eth_macaddr_get(port, &addr);
rte_eth_promiscuous_enable(port);
rte_eth_add_rx_callback(port, 0, add_timestamps, proposer);
rte_eth_add_tx_callback(port, 0, calc_latency, NULL);
return 0;
}
static void
lcore_main(uint8_t port, __rte_unused struct proposer *p)
{
proposer_preexecute(p);
for (;;) {
// Check if signal is received
if (force_quit)
break;
struct rte_mbuf *bufs[BURST_SIZE];
const uint16_t nb_rx = rte_eth_rx_burst(port, 0, bufs, BURST_SIZE);
if (unlikely(nb_rx == 0))
continue;
uint16_t buf;
for (buf = 0; buf < nb_rx; buf++)
rte_pktmbuf_free(bufs[buf]);
}
}
static __attribute__((noreturn)) int
lcore_mainloop(__attribute__((unused)) void *arg)
{
uint64_t prev_tsc = 0, cur_tsc, diff_tsc;
unsigned lcore_id;
lcore_id = rte_lcore_id();
rte_log(RTE_LOG_DEBUG, RTE_LOGTYPE_TIMER,
"Starting mainloop on core %u\n", lcore_id);
while(1) {
cur_tsc = rte_rdtsc();
diff_tsc = cur_tsc - prev_tsc;
if (diff_tsc > TIMER_RESOLUTION_CYCLES) {
rte_timer_manage();
prev_tsc = cur_tsc;
}
}
}
static void
report_stat(struct rte_timer *tim, __attribute((unused)) void *arg)
{
/* print stat */
uint32_t count = rte_atomic32_read(&stat);
rte_log(RTE_LOG_INFO, RTE_LOGTYPE_USER8,
"Throughput = %8u msg/s\n", count);
/* reset stat */
rte_atomic32_set(&stat, 0);
/* this timer is automatically reloaded until we decide to stop it */
if (force_quit)
rte_timer_stop(tim);
}
static void
check_timeout(struct rte_timer *tim, void *arg)
{
struct proposer* p = (struct proposer *) arg;
unsigned lcore_id = rte_lcore_id();
rte_log(RTE_LOG_DEBUG, RTE_LOGTYPE_USER8, "%s() on lcore_id %i\n", __func__, lcore_id);
struct paxos_message out;
out.type = PAXOS_PREPARE;
struct timeout_iterator* iter = proposer_timeout_iterator(p);
while(timeout_iterator_prepare(iter, &out.u.prepare)) {
rte_log(RTE_LOG_DEBUG, RTE_LOGTYPE_USER8,
"%s Send PREPARE inst %d ballot %d\n",
__func__, out.u.prepare.iid, out.u.prepare.ballot);
send_paxos_message(&out);
}
out.type = PAXOS_ACCEPT;
while(timeout_iterator_accept(iter, &out.u.accept)) {
rte_log(RTE_LOG_DEBUG, RTE_LOGTYPE_USER8,
"%s: Send ACCEPT inst %d ballot %d\n",
__func__, out.u.prepare.iid, out.u.prepare.ballot);
send_paxos_message(&out);
}
timeout_iterator_free(iter);
/* this timer is automatically reloaded until we decide to stop it */
if (force_quit)
rte_timer_stop(tim);
}
int
main(int argc, char *argv[])
{
uint8_t portid = 0;
unsigned master_core, lcore_id;
signal(SIGTERM, signal_handler);
signal(SIGINT, signal_handler);
force_quit = false;
int proposer_id = 0;
if (rte_get_log_level() == RTE_LOG_DEBUG) {
paxos_config.verbosity = PAXOS_LOG_DEBUG;
}
struct proposer *proposer = proposer_new(proposer_id, NUM_ACCEPTORS);
first_time = true;
/* init EAL */
int ret = rte_eal_init(argc, argv);
if (ret < 0)
rte_exit(EXIT_FAILURE, "Error with EAL initialization\n");
/* init timer structure */
rte_timer_init(&timer);
rte_timer_init(&stat_timer);
/* load deliver_timer, every 1 s, on a slave lcore, reloaded automatically */
uint64_t hz = rte_get_timer_hz();
/* Call rte_timer_manage every 10ms */
TIMER_RESOLUTION_CYCLES = hz / 100;
rte_log(RTE_LOG_INFO, RTE_LOGTYPE_USER1, "Clock: %"PRIu64"\n", hz);
/* master core */
master_core = rte_lcore_id();
/* slave core */
lcore_id = rte_get_next_lcore(master_core, 0, 1);
rte_log(RTE_LOG_DEBUG, RTE_LOGTYPE_USER1, "lcore_id: %d\n", lcore_id);
rte_timer_reset(&timer, hz, PERIODICAL, lcore_id, check_timeout, proposer);
/* reset timer */
rte_eal_remote_launch(lcore_mainloop, NULL, lcore_id);
/* stat core */
lcore_id = rte_get_next_lcore(lcore_id , 0, 1);
rte_log(RTE_LOG_DEBUG, RTE_LOGTYPE_USER1, "lcore_id: %d\n", lcore_id);
rte_timer_reset(&stat_timer, hz, PERIODICAL, lcore_id,
report_stat, NULL);
/* init RTE timer library */
rte_timer_subsystem_init();
mbuf_pool = rte_pktmbuf_pool_create("MBUF_POOL",
NUM_MBUFS, 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");
/* reset timer */
rte_eal_remote_launch(lcore_mainloop, NULL, lcore_id);
if (port_init(portid, mbuf_pool, proposer) != 0)
rte_exit(EXIT_FAILURE, "Cannot init port %"PRIu8"\n", portid);
lcore_main(portid, proposer);
rte_log(RTE_LOG_DEBUG, RTE_LOGTYPE_USER8, "Free proposer\n");
proposer_free(proposer);
return 0;
}
/*
* Initialises a given port using global settings and with the rx buffers
* coming from the mbuf_pool passed as parameter
*/
static inline int
port_init(uint8_t port, struct rte_mempool *mbuf_pool)
{
struct rte_eth_dev_info dev_info;
struct rte_eth_rxconf *rxconf;
struct rte_eth_conf port_conf;
uint16_t rxRings, txRings;
const uint16_t rxRingSize = RTE_TEST_RX_DESC_DEFAULT, txRingSize = RTE_TEST_TX_DESC_DEFAULT;
int retval;
uint16_t q;
uint16_t queues_per_pool;
uint32_t max_nb_pools;
/*
* The max pool number from dev_info will be used to validate the pool
* number specified in cmd line
*/
rte_eth_dev_info_get(port, &dev_info);
max_nb_pools = (uint32_t)dev_info.max_vmdq_pools;
/*
* We allow to process part of VMDQ pools specified by num_pools in
* command line.
*/
if (num_pools > max_nb_pools) {
printf("num_pools %d >max_nb_pools %d\n",
num_pools, max_nb_pools);
return -1;
}
retval = get_eth_conf(&port_conf, max_nb_pools);
if (retval < 0)
return retval;
/*
* NIC queues are divided into pf queues and vmdq queues.
*/
/* There is assumption here all ports have the same configuration! */
num_pf_queues = dev_info.max_rx_queues - dev_info.vmdq_queue_num;
queues_per_pool = dev_info.vmdq_queue_num / dev_info.max_vmdq_pools;
num_vmdq_queues = num_pools * queues_per_pool;
num_queues = num_pf_queues + num_vmdq_queues;
vmdq_queue_base = dev_info.vmdq_queue_base;
vmdq_pool_base = dev_info.vmdq_pool_base;
printf("pf queue num: %u, configured vmdq pool num: %u,"
" each vmdq pool has %u queues\n",
num_pf_queues, num_pools, queues_per_pool);
printf("vmdq queue base: %d pool base %d\n",
vmdq_queue_base, vmdq_pool_base);
if (port >= rte_eth_dev_count())
return -1;
/*
* Though in this example, we only receive packets from the first queue
* of each pool and send packets through first rte_lcore_count() tx
* queues of vmdq queues, all queues including pf queues are setup.
* This is because VMDQ queues doesn't always start from zero, and the
* PMD layer doesn't support selectively initialising part of rx/tx
* queues.
*/
rxRings = (uint16_t)dev_info.max_rx_queues;
txRings = (uint16_t)dev_info.max_tx_queues;
retval = rte_eth_dev_configure(port, rxRings, txRings, &port_conf);
if (retval != 0)
return retval;
rte_eth_dev_info_get(port, &dev_info);
rxconf = &dev_info.default_rxconf;
rxconf->rx_drop_en = 1;
for (q = 0; q < rxRings; q++) {
retval = rte_eth_rx_queue_setup(port, q, rxRingSize,
rte_eth_dev_socket_id(port),
rxconf,
mbuf_pool);
if (retval < 0) {
printf("initialise rx queue %d failed\n", q);
return retval;
}
}
for (q = 0; q < txRings; q++) {
retval = rte_eth_tx_queue_setup(port, q, txRingSize,
rte_eth_dev_socket_id(port),
NULL);
if (retval < 0) {
printf("initialise tx queue %d failed\n", q);
return retval;
}
}
retval = rte_eth_dev_start(port);
if (retval < 0) {
printf("port %d start failed\n", port);
return retval;
}
rte_eth_macaddr_get(port, &vmdq_ports_eth_addr[port]);
printf("Port %u MAC: %02"PRIx8" %02"PRIx8" %02"PRIx8
" %02"PRIx8" %02"PRIx8" %02"PRIx8"\n",
(unsigned)port,
vmdq_ports_eth_addr[port].addr_bytes[0],
vmdq_ports_eth_addr[port].addr_bytes[1],
vmdq_ports_eth_addr[port].addr_bytes[2],
vmdq_ports_eth_addr[port].addr_bytes[3],
vmdq_ports_eth_addr[port].addr_bytes[4],
vmdq_ports_eth_addr[port].addr_bytes[5]);
/*
* Set mac for each pool.
* There is no default mac for the pools in i40.
* Removes this after i40e fixes this issue.
*/
for (q = 0; q < num_pools; q++) {
struct ether_addr mac;
mac = pool_addr_template;
mac.addr_bytes[4] = port;
mac.addr_bytes[5] = q;
printf("Port %u vmdq pool %u set mac %02x:%02x:%02x:%02x:%02x:%02x\n",
port, q,
mac.addr_bytes[0], mac.addr_bytes[1],
mac.addr_bytes[2], mac.addr_bytes[3],
mac.addr_bytes[4], mac.addr_bytes[5]);
retval = rte_eth_dev_mac_addr_add(port, &mac,
q + vmdq_pool_base);
if (retval) {
printf("mac addr add failed at pool %d\n", q);
return retval;
}
}
return 0;
}
/**********************************************************************
*@description:
*
*
*@parameters:
* [in]:
* [in]:
*
*@return values:
*
**********************************************************************/
static int odp_init_ports(unsigned short nb_ports, struct odp_user_config *user_conf, struct odp_lcore_config *lcore_conf)
{
int ret;
uint8_t portid;
uint16_t queueid;
unsigned lcore_id;
uint8_t nb_rx_queue =0;
uint8_t max_rx_queue =0;
uint8_t queue, socketid;
uint32_t n_tx_queue, nb_lcores, nb_mbuf;
struct ether_addr eth_addr;
struct rte_eth_dev_info dev_info;
struct rte_eth_txconf *txconf;
nb_lcores = rte_lcore_count();
n_tx_queue = nb_lcores;
if (n_tx_queue > MAX_TX_QUEUE_PER_PORT)
n_tx_queue = MAX_TX_QUEUE_PER_PORT;
printf("\nStart to Init port \n" );
/* initialize all ports */
for (portid = 0; portid < nb_ports; portid++)
{
/* skip ports that are not enabled */
if ((user_conf->port_mask & (1 << portid)) == 0)
{
printf("\nSkipping disabled port %d\n", portid);
continue;
}
/* init port */
printf("\t port %d: \n", portid );
nb_rx_queue = odp_get_port_rx_queues_nb(portid, user_conf);
if(max_rx_queue < nb_rx_queue)
max_rx_queue = nb_rx_queue;
printf("\t Creating queues: rx queue number=%d tx queue number=%u... \n", nb_rx_queue, (unsigned)n_tx_queue );
ret = rte_eth_dev_configure(portid, nb_rx_queue, (uint16_t)n_tx_queue, &odp_port_conf);
if (ret < 0)
rte_exit(EXIT_FAILURE, "Cannot configure device: err=%d, port=%d\n", ret, portid);
rte_eth_macaddr_get(portid, ð_addr);
printf ("\t MAC Address:%02X:%02X:%02X:%02X:%02X:%02X \n",
eth_addr.addr_bytes[0], eth_addr.addr_bytes[1],
eth_addr.addr_bytes[2], eth_addr.addr_bytes[3],
eth_addr.addr_bytes[4], eth_addr.addr_bytes[5]);
/* init one TX queue per couple (lcore,port) */
queueid = 0;
for (lcore_id = 0; lcore_id < RTE_MAX_LCORE; lcore_id++)
{
if (rte_lcore_is_enabled(lcore_id) == 0)
continue;
if (user_conf->numa_on)
socketid = (uint8_t)rte_lcore_to_socket_id(lcore_id);
else
socketid = 0;
printf("\t lcore id:%u, tx queue id:%d, socket id:%d \n", lcore_id, queueid, socketid);
ret = rte_eth_tx_queue_setup(portid, queueid, ODP_TX_DESC_DEFAULT, socketid, &odp_tx_conf);
if (ret < 0)
rte_exit(EXIT_FAILURE, "rte_eth_tx_queue_setup: err=%d, " "port=%d\n", ret, portid);
lcore_conf[lcore_id].tx_queue_id[portid] = queueid;
queueid++;
}
printf("\n");
}
nb_mbuf = RTE_MAX((nb_ports*nb_rx_queue*ODP_RX_DESC_DEFAULT +
nb_ports*nb_lcores*MAX_PKT_BURST +
nb_ports*n_tx_queue*ODP_TX_DESC_DEFAULT +
nb_lcores*MEMPOOL_CACHE_SIZE), (unsigned)8192);
/* init memory */
ret = odp_init_mbuf_pool(nb_mbuf, user_conf);
if (ret < 0)
rte_exit(EXIT_FAILURE, "init_mem failed\n");
for (lcore_id = 0; lcore_id < RTE_MAX_LCORE; lcore_id++)
{
if (rte_lcore_is_enabled(lcore_id) == 0)
continue;
printf("\nInitializing rx queues on lcore %u ... \n", lcore_id );
/* init RX queues */
for(queue = 0; queue < lcore_conf[lcore_id].n_rx_queue; ++queue)
{
portid = lcore_conf[lcore_id].rx_queue_list[queue].port_id;
queueid = lcore_conf[lcore_id].rx_queue_list[queue].queue_id;
if (user_conf->numa_on)
socketid = (uint8_t)rte_lcore_to_socket_id(lcore_id);
else
socketid = 0;
printf("port id:%d, rx queue id: %d, socket id:%d \n", portid, queueid, socketid);
ret = rte_eth_rx_queue_setup(portid, queueid, ODP_RX_DESC_DEFAULT, socketid, &odp_rx_conf, odp_pktmbuf_pool[socketid]);
if (ret < 0)
rte_exit(EXIT_FAILURE, "rte_eth_rx_queue_setup: err=%d," "port=%d\n", ret, portid);
}
}
return 0;
}
/*
* Initialises a given port using global settings and with the rx buffers
* coming from the mbuf_pool passed as 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;
uint16_t rx_rings, tx_rings = (uint16_t)rte_lcore_count();
const uint16_t rx_ring_size = RTE_TEST_RX_DESC_DEFAULT, tx_ring_size = RTE_TEST_TX_DESC_DEFAULT;
int retval;
uint16_t q;
/* The max pool number from dev_info will be used to validate the pool number specified in cmd line */
rte_eth_dev_info_get (port, &dev_info);
/*configure the number of supported virtio devices based on VMDQ limits */
num_devices = dev_info.max_vmdq_pools;
num_queues = dev_info.max_rx_queues;
retval = validate_num_devices(MAX_DEVICES);
if (retval < 0)
return retval;
/* Get port configuration. */
retval = get_eth_conf(&port_conf, num_devices);
if (retval < 0)
return retval;
if (port >= rte_eth_dev_count()) return -1;
rx_rings = (uint16_t)num_queues,
/* Configure ethernet device. */
retval = rte_eth_dev_configure(port, rx_rings, tx_rings, &port_conf);
if (retval != 0)
return retval;
/* Setup the queues. */
for (q = 0; q < rx_rings; q ++) {
retval = rte_eth_rx_queue_setup(port, q, rx_ring_size,
rte_eth_dev_socket_id(port), &rx_conf_default,
mbuf_pool);
if (retval < 0)
return retval;
}
for (q = 0; q < tx_rings; q ++) {
retval = rte_eth_tx_queue_setup(port, q, tx_ring_size,
rte_eth_dev_socket_id(port), &tx_conf_default);
if (retval < 0)
return retval;
}
/* Start the device. */
retval = rte_eth_dev_start(port);
if (retval < 0)
return retval;
rte_eth_macaddr_get(port, &vmdq_ports_eth_addr[port]);
RTE_LOG(INFO, VHOST_PORT, "Max virtio devices supported: %u\n", num_devices);
RTE_LOG(INFO, VHOST_PORT, "Port %u MAC: %02"PRIx8" %02"PRIx8" %02"PRIx8
" %02"PRIx8" %02"PRIx8" %02"PRIx8"\n",
(unsigned)port,
vmdq_ports_eth_addr[port].addr_bytes[0],
vmdq_ports_eth_addr[port].addr_bytes[1],
vmdq_ports_eth_addr[port].addr_bytes[2],
vmdq_ports_eth_addr[port].addr_bytes[3],
vmdq_ports_eth_addr[port].addr_bytes[4],
vmdq_ports_eth_addr[port].addr_bytes[5]);
return 0;
}
/*----------------------------------------------------------------------------*/
int
SetInterfaceInfo(char* dev_name_list)
{
struct ifreq ifr;
int eidx = 0;
int i, j;
int set_all_inf = (strncmp(dev_name_list, ALL_STRING, sizeof(ALL_STRING))==0);
TRACE_CONFIG("Loading interface setting\n");
CONFIG.eths = (struct eth_table *)
calloc(MAX_DEVICES, sizeof(struct eth_table));
if (!CONFIG.eths)
exit(EXIT_FAILURE);
if (current_iomodule_func == &ps_module_func) {
/* calculate num_devices now! */
num_devices = ps_list_devices(devices);
if (num_devices == -1) {
perror("ps_list_devices");
exit(EXIT_FAILURE);
}
/* Create socket */
int sock = socket(AF_INET, SOCK_DGRAM, IPPROTO_IP);
if (sock == -1) {
perror("socket");
}
/* To Do: Parse dev_name_list rather than use strstr */
for (i = 0; i < num_devices; i++) {
strcpy(ifr.ifr_name, devices[i].name);
/* getting interface information */
if (ioctl(sock, SIOCGIFFLAGS, &ifr) == 0) {
if (!set_all_inf && strstr(dev_name_list, ifr.ifr_name) == NULL)
continue;
/* Setting informations */
eidx = CONFIG.eths_num++;
strcpy(CONFIG.eths[eidx].dev_name, ifr.ifr_name);
CONFIG.eths[eidx].ifindex = devices[i].ifindex;
/* getting address */
if (ioctl(sock, SIOCGIFADDR, &ifr) == 0 ) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].ip_addr = *(uint32_t *)&sin;
}
if (ioctl(sock, SIOCGIFHWADDR, &ifr) == 0 ) {
for (j = 0; j < ETH_ALEN; j ++) {
CONFIG.eths[eidx].haddr[j] = ifr.ifr_addr.sa_data[j];
}
}
/* Net MASK */
if (ioctl(sock, SIOCGIFNETMASK, &ifr) == 0) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].netmask = *(uint32_t *)&sin;
}
/* add to attached devices */
for (j = 0; j < num_devices_attached; j++) {
if (devices_attached[j] == devices[i].ifindex) {
break;
}
}
devices_attached[num_devices_attached] = devices[i].ifindex;
num_devices_attached++;
} else {
perror("SIOCGIFFLAGS");
}
}
num_queues = GetNumQueues();
if (num_queues <= 0) {
TRACE_CONFIG("Failed to find NIC queues!\n");
return -1;
}
if (num_queues > num_cpus) {
TRACE_CONFIG("Too many NIC queues available.\n");
return -1;
}
} else if (current_iomodule_func == &dpdk_module_func) {
#ifndef DISABLE_DPDK
int cpu = CONFIG.num_cores;
uint32_t cpumask = 0;
char cpumaskbuf[10];
char mem_channels[5];
int ret;
static struct ether_addr ports_eth_addr[RTE_MAX_ETHPORTS];
/* 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 (CONFIG.num_mem_ch == 0) {
TRACE_ERROR("DPDK module requires # of memory channels "
"per socket parameter!\n");
exit(EXIT_FAILURE);
}
sprintf(mem_channels, "%d", CONFIG.num_mem_ch);
/* 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");
/* give me the count of 'detected' ethernet ports */
num_devices = rte_eth_dev_count();
if (num_devices == 0) {
rte_exit(EXIT_FAILURE, "No Ethernet port!\n");
}
/* get mac addr entries of 'detected' dpdk ports */
for (ret = 0; ret < num_devices; ret++)
rte_eth_macaddr_get(ret, &ports_eth_addr[ret]);
num_queues = MIN(CONFIG.num_cores, MAX_CPUS);
struct ifaddrs *ifap;
struct ifaddrs *iter_if;
char *seek;
if (getifaddrs(&ifap) != 0) {
perror("getifaddrs: ");
exit(EXIT_FAILURE);
}
iter_if = ifap;
do {
if (iter_if->ifa_addr->sa_family == AF_INET &&
!set_all_inf &&
(seek=strstr(dev_name_list, iter_if->ifa_name)) != NULL &&
/* check if the interface was not aliased */
*(seek + strlen(iter_if->ifa_name)) != ':') {
struct ifreq ifr;
/* Setting informations */
eidx = CONFIG.eths_num++;
strcpy(CONFIG.eths[eidx].dev_name, iter_if->ifa_name);
strcpy(ifr.ifr_name, iter_if->ifa_name);
/* Create socket */
int sock = socket(AF_INET, SOCK_DGRAM, IPPROTO_IP);
if (sock == -1) {
perror("socket");
}
/* getting address */
if (ioctl(sock, SIOCGIFADDR, &ifr) == 0 ) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].ip_addr = *(uint32_t *)&sin;
}
if (ioctl(sock, SIOCGIFHWADDR, &ifr) == 0 ) {
for (j = 0; j < ETH_ALEN; j ++) {
CONFIG.eths[eidx].haddr[j] = ifr.ifr_addr.sa_data[j];
}
}
/* Net MASK */
if (ioctl(sock, SIOCGIFNETMASK, &ifr) == 0) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].netmask = *(uint32_t *)&sin;
}
close(sock);
for (j = 0; j < num_devices; j++) {
if (!memcmp(&CONFIG.eths[eidx].haddr[0], &ports_eth_addr[j],
ETH_ALEN))
CONFIG.eths[eidx].ifindex = j;
}
/* add to attached devices */
for (j = 0; j < num_devices_attached; j++) {
if (devices_attached[j] == CONFIG.eths[eidx].ifindex) {
break;
}
}
devices_attached[num_devices_attached] = CONFIG.eths[eidx].ifindex;
num_devices_attached++;
fprintf(stderr, "Total number of attached devices: %d\n",
num_devices_attached);
fprintf(stderr, "Interface name: %s\n",
iter_if->ifa_name);
}
iter_if = iter_if->ifa_next;
} while (iter_if != NULL);
freeifaddrs(ifap);
#endif /* !DISABLE_DPDK */
} else if (current_iomodule_func == &netmap_module_func) {
#ifndef DISABLE_NETMAP
struct ifaddrs *ifap;
struct ifaddrs *iter_if;
char *seek;
num_queues = MIN(CONFIG.num_cores, MAX_CPUS);
if (getifaddrs(&ifap) != 0) {
perror("getifaddrs: ");
exit(EXIT_FAILURE);
}
iter_if = ifap;
do {
if (iter_if->ifa_addr->sa_family == AF_INET &&
!set_all_inf &&
(seek=strstr(dev_name_list, iter_if->ifa_name)) != NULL &&
/* check if the interface was not aliased */
*(seek + strlen(iter_if->ifa_name)) != ':') {
struct ifreq ifr;
/* Setting informations */
eidx = CONFIG.eths_num++;
strcpy(CONFIG.eths[eidx].dev_name, iter_if->ifa_name);
strcpy(ifr.ifr_name, iter_if->ifa_name);
/* Create socket */
int sock = socket(AF_INET, SOCK_DGRAM, IPPROTO_IP);
if (sock == -1) {
perror("socket");
}
/* getting address */
if (ioctl(sock, SIOCGIFADDR, &ifr) == 0 ) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].ip_addr = *(uint32_t *)&sin;
}
if (ioctl(sock, SIOCGIFHWADDR, &ifr) == 0 ) {
for (j = 0; j < ETH_ALEN; j ++) {
CONFIG.eths[eidx].haddr[j] = ifr.ifr_addr.sa_data[j];
}
}
/* Net MASK */
if (ioctl(sock, SIOCGIFNETMASK, &ifr) == 0) {
struct in_addr sin = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr;
CONFIG.eths[eidx].netmask = *(uint32_t *)&sin;
}
close(sock);
#if 0
for (j = 0; j < num_devices; j++) {
if (!memcmp(&CONFIG.eths[eidx].haddr[0], &ports_eth_addr[j],
ETH_ALEN))
CONFIG.eths[eidx].ifindex = ifr.ifr_ifindex;
#endif
CONFIG.eths[eidx].ifindex = eidx;//if_nametoindex(ifr.ifr_name);
TRACE_INFO("Ifindex of interface %s is: %d\n",
ifr.ifr_name, CONFIG.eths[eidx].ifindex);
#if 0
}
#endif
/* add to attached devices */
for (j = 0; j < num_devices_attached; j++) {
if (devices_attached[j] == CONFIG.eths[eidx].ifindex) {
break;
}
}
devices_attached[num_devices_attached] = if_nametoindex(ifr.ifr_name);//CONFIG.eths[eidx].ifindex;
num_devices_attached++;
fprintf(stderr, "Total number of attached devices: %d\n",
num_devices_attached);
fprintf(stderr, "Interface name: %s\n",
iter_if->ifa_name);
}
iter_if = iter_if->ifa_next;
} while (iter_if != NULL);
freeifaddrs(ifap);
#endif /* !DISABLE_NETMAP */
}
return 0;
}
int
main(int argc, char **argv)
{
//struct lcore_queue_conf *qconf = NULL;
//struct rte_eth_dev_info dev_info;
struct lcore_env** envs;
int ret;
uint8_t n_ports;
unsigned lcore_count;
ret = rte_eal_init(argc, argv);
if (ret < 0)
rte_exit(EXIT_FAILURE, "Invalid EAL arguments\n");
argc -= ret;
argv += ret;
ret = l2sw_parse_args(argc, argv);
if (ret < 0)
rte_exit(EXIT_FAILURE, "Invalid MARIO arguments\n");
lcore_count = rte_lcore_count();
n_ports = rte_eth_dev_count();
//RTE_LOG(INFO, MARIO, "Find %u logical cores\n" , lcore_count);
mbuf_pool = rte_mempool_create("mbuf_pool", NB_MBUF, MBUF_SIZE,
32, sizeof(struct rte_pktmbuf_pool_private),
rte_pktmbuf_pool_init, NULL,
rte_pktmbuf_init, NULL,
rte_socket_id(), 0);
if (mbuf_pool == NULL)
rte_exit(EXIT_FAILURE, "Cannot init mbuf pool\n");
// init route_table
route_table = create_route_table(ROUTE_ENTRY_SIZE);
add_staticroute(route_table);
// init arp_table
arp_table = create_arp_table(ARP_ENTRY_SIZE);
n_ports = rte_eth_dev_count();
if (n_ports == 0)
rte_exit(EXIT_FAILURE, "No Ethernet ports - byte\n");
//RTE_LOG(INFO, MARIO, "Find %u ethernet ports\n", n_ports);
if (n_ports > RTE_MAX_ETHPORTS)
n_ports = RTE_MAX_ETHPORTS;
/* Each logical core is assigned a dedicated TX queue on each port. */
/*
for(uint8_t port_id = 0; port_id < n_ports; port_id++) {
rte_eth_dev_info_get(port_id, &dev_info);
}
*/
/* Initialize the port/queue configuration of each logical core */
/*
for(uint8_t port_id = 0; port_id < n_ports; port_id++) {
;
}
*/
/* Initialize lcore_env */
envs = (struct lcore_env**) rte_malloc(NULL,sizeof(struct lcore_env*),0);
if (envs == NULL)
rte_exit(EXIT_FAILURE, "Cannot allocate memory for core envs\n");
uint8_t lcore_id;
for (lcore_id = 0; lcore_id < lcore_count; lcore_id++) {
struct lcore_env* env;
env = (struct lcore_env*) rte_malloc(NULL,sizeof(struct lcore_env) +
sizeof(struct mbuf_table) *n_ports,0);
if (env == NULL)
rte_exit(EXIT_FAILURE,
"Cannot allocate memory for %u core env\n", lcore_id);
env->n_port = n_ports;
env->lcore_id = lcore_id;
memset(env->tx_mbufs, 0, sizeof(struct mbuf_table) * n_ports);
envs[lcore_id] = env;
}
/* Initialise each port */
uint8_t port_id;
for(port_id = 0; port_id < n_ports; port_id++) {
//RTE_LOG(INFO, MARIO, "Initializing port %u...", port_id);
fflush(stdout);
ret = rte_eth_dev_configure(port_id, lcore_count, lcore_count, &port_conf);
if (ret < 0)
rte_exit(EXIT_FAILURE, "Cannot configure device: err=%d, port=%u\n",
ret, (unsigned)port_id);
//RTE_LOG(INFO, MARIO, "done\n");
rte_eth_macaddr_get(port_id, &port2eth[port_id]);
/* init one RX queue */
uint8_t core_id;
for (core_id = 0; core_id < lcore_count; core_id++) {
ret = rte_eth_rx_queue_setup(port_id, core_id, nb_rxd,
rte_eth_dev_socket_id(port_id),
NULL,
mbuf_pool);
if (ret < 0)
rte_exit(EXIT_FAILURE,
"rte_eth_rx_queue_setup:err=%d, port=%u queue=%u\n",
ret, (unsigned) port_id, (unsigned) core_id);
}
/* init one TX queue */
for (core_id = 0; core_id < lcore_count; core_id++) {
ret = rte_eth_tx_queue_setup(port_id, core_id, nb_txd,
rte_eth_dev_socket_id(port_id), NULL);
if (ret < 0)
rte_exit(EXIT_FAILURE,
"rte_eth_tx_queue_setup:err=%d, port=%u queue=%u\n",
ret, (unsigned) port_id, (unsigned) core_id);
}
/* Start device */
ret = rte_eth_dev_start(port_id);
if (ret < 0)
rte_exit(EXIT_FAILURE, "rte_eth_dev_start:err=%d, port=%u\n",
ret, (unsigned) port_id);
rte_eth_promiscuous_enable(port_id);
/*RTE_LOG(INFO, MARIO,
"Port %u, MAC address %02x:%02x:%02x:%02x:%02x:%02x\n\n",
port_id,
port2eth[port_id].addr_bytes[0],
port2eth[port_id].addr_bytes[1],
port2eth[port_id].addr_bytes[2],
port2eth[port_id].addr_bytes[3],
port2eth[port_id].addr_bytes[4],
port2eth[port_id].addr_bytes[5]);
*/
memset(&port_statistics, 0, sizeof(port_statistics));
}
check_all_ports_link_status(n_ports);
/* launch per-lcore init on every lcore */
rte_eal_mp_remote_launch(l2sw_launch_one_lcore, envs, CALL_MASTER);
{
uint8_t lcore_id;
RTE_LCORE_FOREACH_SLAVE(lcore_id) {
if (rte_eal_wait_lcore(lcore_id) < 0)
return -1;
}
}
rte_free(arp_table);
rte_free(route_table);
return 0;
}
void pktgen_config_ports(void)
{
uint32_t lid, pid, i, s, q, sid;
rxtx_t rt;
pkt_seq_t * pkt;
port_info_t * info;
char buff[RTE_MEMZONE_NAMESIZE];
int32_t ret, cache_size;
char output_buff[256] = { 0 };
// Find out the total number of ports in the system.
// We have already blacklisted the ones we needed to in main routine.
pktgen.nb_ports = rte_eth_dev_count();
if (pktgen.nb_ports > RTE_MAX_ETHPORTS)
pktgen.nb_ports = RTE_MAX_ETHPORTS;
if ( pktgen.nb_ports == 0 )
pktgen_log_panic("*** Did not find any ports to use ***");
pktgen.starting_port = 0;
// Setup the number of ports to display at a time
if ( pktgen.nb_ports > pktgen.nb_ports_per_page )
pktgen.ending_port = pktgen.starting_port + pktgen.nb_ports_per_page;
else
pktgen.ending_port = pktgen.starting_port + pktgen.nb_ports;
wr_port_matrix_dump(pktgen.l2p);
pktgen_log_info("Configuring %d ports, MBUF Size %d, MBUF Cache Size %d",
pktgen.nb_ports, MBUF_SIZE, MBUF_CACHE_SIZE);
// For each lcore setup each port that is handled by that lcore.
for(lid = 0; lid < RTE_MAX_LCORE; lid++) {
if ( wr_get_map(pktgen.l2p, RTE_MAX_ETHPORTS, lid) == 0 )
continue;
// For each port attached or handled by the lcore
for(pid = 0; pid < pktgen.nb_ports; pid++) {
// If non-zero then this port is handled by this lcore.
if ( wr_get_map(pktgen.l2p, pid, lid) == 0 )
continue;
wr_set_port_private(pktgen.l2p, pid, &pktgen.info[pid]);
pktgen.info[pid].pid = pid;
}
}
wr_dump_l2p(pktgen.l2p);
pktgen.total_mem_used = 0;
for(pid = 0; pid < pktgen.nb_ports; pid++) {
// Skip if we do not have any lcores attached to a port.
if ( (rt.rxtx = wr_get_map(pktgen.l2p, pid, RTE_MAX_LCORE)) == 0 )
continue;
pktgen.port_cnt++;
snprintf(output_buff, sizeof(output_buff),
"Initialize Port %d -- TxQ %d, RxQ %d", pid, rt.tx, rt.rx);
info = wr_get_port_private(pktgen.l2p, pid);
// Create the pkt header structures for transmitting sequence of packets.
snprintf(buff, sizeof(buff), "seq_hdr_%d", pid);
info->seq_pkt = (pkt_seq_t *)rte_zmalloc(buff, (sizeof(pkt_seq_t) * NUM_TOTAL_PKTS), RTE_CACHE_LINE_SIZE);
if ( info->seq_pkt == NULL )
pktgen_log_panic("Unable to allocate %d pkt_seq_t headers", NUM_TOTAL_PKTS);
info->seqIdx = 0;
info->seqCnt = 0;
info->nb_mbufs = MAX_MBUFS_PER_PORT;
cache_size = (info->nb_mbufs > RTE_MEMPOOL_CACHE_MAX_SIZE)?
RTE_MEMPOOL_CACHE_MAX_SIZE : info->nb_mbufs;
pktgen_port_conf_setup(pid, &rt, &default_port_conf);
if ( (ret = rte_eth_dev_configure(pid, rt.rx, rt.tx, &info->port_conf)) < 0)
pktgen_log_panic("Cannot configure device: port=%d, Num queues %d,%d (%d)%s",
pid, rt.rx, rt.tx, errno, rte_strerror(-ret));
pkt = &info->seq_pkt[SINGLE_PKT];
// Grab the source MAC addresses */
rte_eth_macaddr_get(pid, &pkt->eth_src_addr);
pktgen_log_info("%s, Src MAC %02x:%02x:%02x:%02x:%02x:%02x", output_buff,
pkt->eth_src_addr.addr_bytes[0],
pkt->eth_src_addr.addr_bytes[1],
pkt->eth_src_addr.addr_bytes[2],
pkt->eth_src_addr.addr_bytes[3],
pkt->eth_src_addr.addr_bytes[4],
pkt->eth_src_addr.addr_bytes[5]);
// Copy the first Src MAC address in SINGLE_PKT to the rest of the sequence packets.
for (i = 0; i < NUM_SEQ_PKTS; i++)
ethAddrCopy( &info->seq_pkt[i].eth_src_addr, &pkt->eth_src_addr );
pktgen.mem_used = 0;
for(q = 0; q < rt.rx; q++) {
// grab the socket id value based on the lcore being used.
sid = rte_lcore_to_socket_id(wr_get_port_lid(pktgen.l2p, pid, q));
// Create and initialize the default Receive buffers.
info->q[q].rx_mp = pktgen_mbuf_pool_create("Default RX", pid, q, info->nb_mbufs, sid, cache_size);
if ( info->q[q].rx_mp == NULL )
pktgen_log_panic("Cannot init port %d for Default RX mbufs", pid);
ret = rte_eth_rx_queue_setup(pid, q, pktgen.nb_rxd, sid, &info->rx_conf, pktgen.info[pid].q[q].rx_mp);
if (ret < 0)
pktgen_log_panic("rte_eth_rx_queue_setup: err=%d, port=%d, %s", ret, pid, rte_strerror(-ret));
}
pktgen_log_info("");
for(q = 0; q < rt.tx; q++) {
// grab the socket id value based on the lcore being used.
sid = rte_lcore_to_socket_id(wr_get_port_lid(pktgen.l2p, pid, q));
// Create and initialize the default Transmit buffers.
info->q[q].tx_mp = pktgen_mbuf_pool_create("Default TX", pid, q, MAX_MBUFS_PER_PORT, sid, cache_size);
if ( info->q[q].tx_mp == NULL )
pktgen_log_panic("Cannot init port %d for Default TX mbufs", pid);
// Create and initialize the range Transmit buffers.
info->q[q].range_mp = pktgen_mbuf_pool_create("Range TX", pid, q, MAX_MBUFS_PER_PORT, sid, 0);
if ( info->q[q].range_mp == NULL )
pktgen_log_panic("Cannot init port %d for Range TX mbufs", pid);
// Create and initialize the sequence Transmit buffers.
info->q[q].seq_mp = pktgen_mbuf_pool_create("Sequence TX", pid, q, MAX_MBUFS_PER_PORT, sid, cache_size);
if ( info->q[q].seq_mp == NULL )
pktgen_log_panic("Cannot init port %d for Sequence TX mbufs", pid);
// Used for sending special packets like ARP requests
info->q[q].special_mp = pktgen_mbuf_pool_create("Special TX", pid, q, MAX_SPECIAL_MBUFS, sid, cache_size);
if (info->q[q].special_mp == NULL)
pktgen_log_panic("Cannot init port %d for Special TX mbufs", pid);
// Setup the PCAP file for each port
if ( pktgen.info[pid].pcap != NULL ) {
if ( pktgen_pcap_parse(pktgen.info[pid].pcap, info, q) == -1 )
pktgen_log_panic("Cannot load PCAP file for port %d", pid);
}
// Find out the link speed to program the WTHRESH value correctly.
pktgen_get_link_status(info, pid, 0);
//info->tx_conf.tx_thresh.wthresh = (info->link.link_speed == 1000)? TX_WTHRESH_1GB : TX_WTHRESH;
ret = rte_eth_tx_queue_setup(pid, q, pktgen.nb_txd, sid, &info->tx_conf);
if (ret < 0)
pktgen_log_panic("rte_eth_tx_queue_setup: err=%d, port=%d, %s", ret, pid, rte_strerror(-ret));
#if 0
ret = rte_eth_dev_flow_ctrl_set(pid, &fc_conf);
if (ret < 0)
pktgen_log_panic("rte_eth_dev_flow_ctrl_set: err=%d, port=%d, %s", ret, pid, rte_strerror(-ret));
#endif
pktgen_log_info("");
}
pktgen_log_info("%*sPort memory used = %6lu KB", 71, " ", (pktgen.mem_used + 1023)/1024);
}
pktgen_log_info("%*sTotal memory used = %6lu KB", 70, " ", (pktgen.total_mem_used + 1023)/1024);
// Start up the ports and display the port Link status
for(pid = 0; pid < pktgen.nb_ports; pid++) {
if ( wr_get_map(pktgen.l2p, pid, RTE_MAX_LCORE) == 0 )
continue;
info = wr_get_port_private(pktgen.l2p, pid);
/* Start device */
if ( (ret = rte_eth_dev_start(pid)) < 0 )
pktgen_log_panic("rte_eth_dev_start: port=%d, %s", pid, rte_strerror(-ret));
pktgen_get_link_status(info, pid, 1);
if (info->link.link_status) {
snprintf(output_buff, sizeof(output_buff), "Port %2d: Link Up - speed %u Mbps - %s", pid,
(uint32_t) info->link.link_speed,
(info->link.link_duplex == ETH_LINK_FULL_DUPLEX) ?
("full-duplex") : ("half-duplex"));
} else
snprintf(output_buff, sizeof(output_buff), "Port %2d: Link Down", pid);
// If enabled, put device in promiscuous mode.
if (pktgen.flags & PROMISCUOUS_ON_FLAG) {
strncatf(output_buff, " <Enable promiscuous mode>");
rte_eth_promiscuous_enable(pid);
}
pktgen_log_info("%s", output_buff);
pktgen.info[pid].seq_pkt[SINGLE_PKT].pktSize = MIN_PKT_SIZE;
// Setup the port and packet defaults. (must be after link speed is found)
for (s = 0; s < NUM_TOTAL_PKTS; s++)
pktgen_port_defaults(pid, s);
pktgen_range_setup(info);
pktgen_rnd_bits_init(&pktgen.info[pid].rnd_bitfields);
}
pktgen_log_info("");
for (sid = 0; sid < wr_coremap_cnt(pktgen.core_info, pktgen.core_cnt, 0); sid++)
pktgen_packet_capture_init(&pktgen.capture[sid], sid);
}
static inline int
app_link_filter_arp_add(struct app_link_params *link)
{
struct rte_eth_ethertype_filter filter = {
.ether_type = ETHER_TYPE_ARP,
.flags = 0,
.queue = link->arp_q,
};
return rte_eth_dev_filter_ctrl(link->pmd_id,
RTE_ETH_FILTER_ETHERTYPE,
RTE_ETH_FILTER_ADD,
&filter);
}
static inline int
app_link_filter_tcp_syn_add(struct app_link_params *link)
{
struct rte_eth_syn_filter filter = {
.hig_pri = 1,
.queue = link->tcp_syn_q,
};
return rte_eth_dev_filter_ctrl(link->pmd_id,
RTE_ETH_FILTER_SYN,
RTE_ETH_FILTER_ADD,
&filter);
}
static inline int
app_link_filter_ip_add(struct app_link_params *l1, struct app_link_params *l2)
{
struct rte_eth_ntuple_filter filter = {
.flags = RTE_5TUPLE_FLAGS,
.dst_ip = rte_bswap32(l2->ip),
.dst_ip_mask = UINT32_MAX, /* Enable */
.src_ip = 0,
.src_ip_mask = 0, /* Disable */
.dst_port = 0,
.dst_port_mask = 0, /* Disable */
.src_port = 0,
.src_port_mask = 0, /* Disable */
.proto = 0,
.proto_mask = 0, /* Disable */
.tcp_flags = 0,
.priority = 1, /* Lowest */
.queue = l1->ip_local_q,
};
return rte_eth_dev_filter_ctrl(l1->pmd_id,
RTE_ETH_FILTER_NTUPLE,
RTE_ETH_FILTER_ADD,
&filter);
}
static inline int
app_link_filter_ip_del(struct app_link_params *l1, struct app_link_params *l2)
{
struct rte_eth_ntuple_filter filter = {
.flags = RTE_5TUPLE_FLAGS,
.dst_ip = rte_bswap32(l2->ip),
.dst_ip_mask = UINT32_MAX, /* Enable */
.src_ip = 0,
.src_ip_mask = 0, /* Disable */
.dst_port = 0,
.dst_port_mask = 0, /* Disable */
.src_port = 0,
.src_port_mask = 0, /* Disable */
.proto = 0,
.proto_mask = 0, /* Disable */
.tcp_flags = 0,
.priority = 1, /* Lowest */
.queue = l1->ip_local_q,
};
return rte_eth_dev_filter_ctrl(l1->pmd_id,
RTE_ETH_FILTER_NTUPLE,
RTE_ETH_FILTER_DELETE,
&filter);
}
static inline int
app_link_filter_tcp_add(struct app_link_params *l1, struct app_link_params *l2)
{
struct rte_eth_ntuple_filter filter = {
.flags = RTE_5TUPLE_FLAGS,
.dst_ip = rte_bswap32(l2->ip),
.dst_ip_mask = UINT32_MAX, /* Enable */
.src_ip = 0,
.src_ip_mask = 0, /* Disable */
.dst_port = 0,
.dst_port_mask = 0, /* Disable */
.src_port = 0,
.src_port_mask = 0, /* Disable */
.proto = IPPROTO_TCP,
.proto_mask = UINT8_MAX, /* Enable */
.tcp_flags = 0,
.priority = 2, /* Higher priority than IP */
.queue = l1->tcp_local_q,
};
return rte_eth_dev_filter_ctrl(l1->pmd_id,
RTE_ETH_FILTER_NTUPLE,
RTE_ETH_FILTER_ADD,
&filter);
}
static inline int
app_link_filter_tcp_del(struct app_link_params *l1, struct app_link_params *l2)
{
struct rte_eth_ntuple_filter filter = {
.flags = RTE_5TUPLE_FLAGS,
.dst_ip = rte_bswap32(l2->ip),
.dst_ip_mask = UINT32_MAX, /* Enable */
.src_ip = 0,
.src_ip_mask = 0, /* Disable */
.dst_port = 0,
.dst_port_mask = 0, /* Disable */
.src_port = 0,
.src_port_mask = 0, /* Disable */
.proto = IPPROTO_TCP,
.proto_mask = UINT8_MAX, /* Enable */
.tcp_flags = 0,
.priority = 2, /* Higher priority than IP */
.queue = l1->tcp_local_q,
};
return rte_eth_dev_filter_ctrl(l1->pmd_id,
RTE_ETH_FILTER_NTUPLE,
RTE_ETH_FILTER_DELETE,
&filter);
}
static inline int
app_link_filter_udp_add(struct app_link_params *l1, struct app_link_params *l2)
{
struct rte_eth_ntuple_filter filter = {
.flags = RTE_5TUPLE_FLAGS,
.dst_ip = rte_bswap32(l2->ip),
.dst_ip_mask = UINT32_MAX, /* Enable */
.src_ip = 0,
.src_ip_mask = 0, /* Disable */
.dst_port = 0,
.dst_port_mask = 0, /* Disable */
.src_port = 0,
.src_port_mask = 0, /* Disable */
.proto = IPPROTO_UDP,
.proto_mask = UINT8_MAX, /* Enable */
.tcp_flags = 0,
.priority = 2, /* Higher priority than IP */
.queue = l1->udp_local_q,
};
return rte_eth_dev_filter_ctrl(l1->pmd_id,
RTE_ETH_FILTER_NTUPLE,
RTE_ETH_FILTER_ADD,
&filter);
}
static inline int
app_link_filter_udp_del(struct app_link_params *l1, struct app_link_params *l2)
{
struct rte_eth_ntuple_filter filter = {
.flags = RTE_5TUPLE_FLAGS,
.dst_ip = rte_bswap32(l2->ip),
.dst_ip_mask = UINT32_MAX, /* Enable */
.src_ip = 0,
.src_ip_mask = 0, /* Disable */
.dst_port = 0,
.dst_port_mask = 0, /* Disable */
.src_port = 0,
.src_port_mask = 0, /* Disable */
.proto = IPPROTO_UDP,
.proto_mask = UINT8_MAX, /* Enable */
.tcp_flags = 0,
.priority = 2, /* Higher priority than IP */
.queue = l1->udp_local_q,
};
return rte_eth_dev_filter_ctrl(l1->pmd_id,
RTE_ETH_FILTER_NTUPLE,
RTE_ETH_FILTER_DELETE,
&filter);
}
static inline int
app_link_filter_sctp_add(struct app_link_params *l1, struct app_link_params *l2)
{
struct rte_eth_ntuple_filter filter = {
.flags = RTE_5TUPLE_FLAGS,
.dst_ip = rte_bswap32(l2->ip),
.dst_ip_mask = UINT32_MAX, /* Enable */
.src_ip = 0,
.src_ip_mask = 0, /* Disable */
.dst_port = 0,
.dst_port_mask = 0, /* Disable */
.src_port = 0,
.src_port_mask = 0, /* Disable */
.proto = IPPROTO_SCTP,
.proto_mask = UINT8_MAX, /* Enable */
.tcp_flags = 0,
.priority = 2, /* Higher priority than IP */
.queue = l1->sctp_local_q,
};
return rte_eth_dev_filter_ctrl(l1->pmd_id,
RTE_ETH_FILTER_NTUPLE,
RTE_ETH_FILTER_ADD,
&filter);
}
static inline int
app_link_filter_sctp_del(struct app_link_params *l1, struct app_link_params *l2)
{
struct rte_eth_ntuple_filter filter = {
.flags = RTE_5TUPLE_FLAGS,
.dst_ip = rte_bswap32(l2->ip),
.dst_ip_mask = UINT32_MAX, /* Enable */
.src_ip = 0,
.src_ip_mask = 0, /* Disable */
.dst_port = 0,
.dst_port_mask = 0, /* Disable */
.src_port = 0,
.src_port_mask = 0, /* Disable */
.proto = IPPROTO_SCTP,
.proto_mask = UINT8_MAX, /* Enable */
.tcp_flags = 0,
.priority = 2, /* Higher priority than IP */
.queue = l1->sctp_local_q,
};
return rte_eth_dev_filter_ctrl(l1->pmd_id,
RTE_ETH_FILTER_NTUPLE,
RTE_ETH_FILTER_DELETE,
&filter);
}
static void
app_link_set_arp_filter(struct app_params *app, struct app_link_params *cp)
{
if (cp->arp_q != 0) {
int status = app_link_filter_arp_add(cp);
APP_LOG(app, LOW, "%s (%" PRIu32 "): "
"Adding ARP filter (queue = %" PRIu32 ")",
cp->name, cp->pmd_id, cp->arp_q);
if (status)
rte_panic("%s (%" PRIu32 "): "
"Error adding ARP filter "
"(queue = %" PRIu32 ") (%" PRId32 ")\n",
cp->name, cp->pmd_id, cp->arp_q, status);
}
}
static void
app_link_set_tcp_syn_filter(struct app_params *app, struct app_link_params *cp)
{
if (cp->tcp_syn_q != 0) {
int status = app_link_filter_tcp_syn_add(cp);
APP_LOG(app, LOW, "%s (%" PRIu32 "): "
"Adding TCP SYN filter (queue = %" PRIu32 ")",
cp->name, cp->pmd_id, cp->tcp_syn_q);
if (status)
rte_panic("%s (%" PRIu32 "): "
"Error adding TCP SYN filter "
"(queue = %" PRIu32 ") (%" PRId32 ")\n",
cp->name, cp->pmd_id, cp->tcp_syn_q,
status);
}
}
void
app_link_up_internal(struct app_params *app, struct app_link_params *cp)
{
uint32_t i;
int status;
/* For each link, add filters for IP of current link */
if (cp->ip != 0) {
for (i = 0; i < app->n_links; i++) {
struct app_link_params *p = &app->link_params[i];
/* IP */
if (p->ip_local_q != 0) {
int status = app_link_filter_ip_add(p, cp);
APP_LOG(app, LOW, "%s (%" PRIu32 "): "
"Adding IP filter (queue= %" PRIu32
", IP = 0x%08" PRIx32 ")",
p->name, p->pmd_id, p->ip_local_q,
cp->ip);
if (status)
rte_panic("%s (%" PRIu32 "): "
"Error adding IP "
"filter (queue= %" PRIu32 ", "
"IP = 0x%08" PRIx32
") (%" PRId32 ")\n",
p->name, p->pmd_id,
p->ip_local_q, cp->ip, status);
}
/* TCP */
if (p->tcp_local_q != 0) {
int status = app_link_filter_tcp_add(p, cp);
APP_LOG(app, LOW, "%s (%" PRIu32 "): "
"Adding TCP filter "
"(queue = %" PRIu32
", IP = 0x%08" PRIx32 ")",
p->name, p->pmd_id, p->tcp_local_q,
cp->ip);
if (status)
rte_panic("%s (%" PRIu32 "): "
"Error adding TCP "
"filter (queue = %" PRIu32 ", "
"IP = 0x%08" PRIx32
") (%" PRId32 ")\n",
p->name, p->pmd_id,
p->tcp_local_q, cp->ip, status);
}
/* UDP */
if (p->udp_local_q != 0) {
int status = app_link_filter_udp_add(p, cp);
APP_LOG(app, LOW, "%s (%" PRIu32 "): "
"Adding UDP filter "
"(queue = %" PRIu32
", IP = 0x%08" PRIx32 ")",
p->name, p->pmd_id, p->udp_local_q,
cp->ip);
if (status)
rte_panic("%s (%" PRIu32 "): "
"Error adding UDP "
"filter (queue = %" PRIu32 ", "
"IP = 0x%08" PRIx32
") (%" PRId32 ")\n",
p->name, p->pmd_id,
p->udp_local_q, cp->ip, status);
}
/* SCTP */
if (p->sctp_local_q != 0) {
int status = app_link_filter_sctp_add(p, cp);
APP_LOG(app, LOW, "%s (%" PRIu32
"): Adding SCTP filter "
"(queue = %" PRIu32
", IP = 0x%08" PRIx32 ")",
p->name, p->pmd_id, p->sctp_local_q,
cp->ip);
if (status)
rte_panic("%s (%" PRIu32 "): "
"Error adding SCTP "
"filter (queue = %" PRIu32 ", "
"IP = 0x%08" PRIx32
") (%" PRId32 ")\n",
p->name, p->pmd_id,
p->sctp_local_q, cp->ip,
status);
}
}
}
/* PMD link up */
status = rte_eth_dev_set_link_up(cp->pmd_id);
/* Do not panic if PMD does not provide link up functionality */
if (status < 0 && status != -ENOTSUP)
rte_panic("%s (%" PRIu32 "): PMD set link up error %"
PRId32 "\n", cp->name, cp->pmd_id, status);
/* Mark link as UP */
cp->state = 1;
}
void
app_link_down_internal(struct app_params *app, struct app_link_params *cp)
{
uint32_t i;
int status;
/* PMD link down */
status = rte_eth_dev_set_link_down(cp->pmd_id);
/* Do not panic if PMD does not provide link down functionality */
if (status < 0 && status != -ENOTSUP)
rte_panic("%s (%" PRIu32 "): PMD set link down error %"
PRId32 "\n", cp->name, cp->pmd_id, status);
/* Mark link as DOWN */
cp->state = 0;
/* Return if current link IP is not valid */
if (cp->ip == 0)
return;
/* For each link, remove filters for IP of current link */
for (i = 0; i < app->n_links; i++) {
struct app_link_params *p = &app->link_params[i];
/* IP */
if (p->ip_local_q != 0) {
int status = app_link_filter_ip_del(p, cp);
APP_LOG(app, LOW, "%s (%" PRIu32
"): Deleting IP filter "
"(queue = %" PRIu32 ", IP = 0x%" PRIx32 ")",
p->name, p->pmd_id, p->ip_local_q, cp->ip);
if (status)
rte_panic("%s (%" PRIu32
"): Error deleting IP filter "
"(queue = %" PRIu32
", IP = 0x%" PRIx32
") (%" PRId32 ")\n",
p->name, p->pmd_id, p->ip_local_q,
cp->ip, status);
}
/* TCP */
if (p->tcp_local_q != 0) {
int status = app_link_filter_tcp_del(p, cp);
APP_LOG(app, LOW, "%s (%" PRIu32
"): Deleting TCP filter "
"(queue = %" PRIu32
", IP = 0x%" PRIx32 ")",
p->name, p->pmd_id, p->tcp_local_q, cp->ip);
if (status)
rte_panic("%s (%" PRIu32
"): Error deleting TCP filter "
"(queue = %" PRIu32
", IP = 0x%" PRIx32
") (%" PRId32 ")\n",
p->name, p->pmd_id, p->tcp_local_q,
cp->ip, status);
}
/* UDP */
if (p->udp_local_q != 0) {
int status = app_link_filter_udp_del(p, cp);
APP_LOG(app, LOW, "%s (%" PRIu32
"): Deleting UDP filter "
"(queue = %" PRIu32 ", IP = 0x%" PRIx32 ")",
p->name, p->pmd_id, p->udp_local_q, cp->ip);
if (status)
rte_panic("%s (%" PRIu32
"): Error deleting UDP filter "
"(queue = %" PRIu32
", IP = 0x%" PRIx32
") (%" PRId32 ")\n",
p->name, p->pmd_id, p->udp_local_q,
cp->ip, status);
}
/* SCTP */
if (p->sctp_local_q != 0) {
int status = app_link_filter_sctp_del(p, cp);
APP_LOG(app, LOW, "%s (%" PRIu32
"): Deleting SCTP filter "
"(queue = %" PRIu32
", IP = 0x%" PRIx32 ")",
p->name, p->pmd_id, p->sctp_local_q, cp->ip);
if (status)
rte_panic("%s (%" PRIu32
"): Error deleting SCTP filter "
"(queue = %" PRIu32
", IP = 0x%" PRIx32
") (%" PRId32 ")\n",
p->name, p->pmd_id, p->sctp_local_q,
cp->ip, status);
}
}
}
static void
app_check_link(struct app_params *app)
{
uint32_t all_links_up, i;
all_links_up = 1;
for (i = 0; i < app->n_links; i++) {
struct app_link_params *p = &app->link_params[i];
struct rte_eth_link link_params;
memset(&link_params, 0, sizeof(link_params));
rte_eth_link_get(p->pmd_id, &link_params);
APP_LOG(app, HIGH, "%s (%" PRIu32 ") (%" PRIu32 " Gbps) %s",
p->name,
p->pmd_id,
link_params.link_speed / 1000,
link_params.link_status ? "UP" : "DOWN");
if (link_params.link_status == ETH_LINK_DOWN)
all_links_up = 0;
}
if (all_links_up == 0)
rte_panic("Some links are DOWN\n");
}
static uint32_t
is_any_swq_frag_or_ras(struct app_params *app)
{
uint32_t i;
for (i = 0; i < app->n_pktq_swq; i++) {
struct app_pktq_swq_params *p = &app->swq_params[i];
if ((p->ipv4_frag == 1) || (p->ipv6_frag == 1) ||
(p->ipv4_ras == 1) || (p->ipv6_ras == 1))
return 1;
}
return 0;
}
static void
app_init_link_frag_ras(struct app_params *app)
{
uint32_t i;
if (is_any_swq_frag_or_ras(app)) {
for (i = 0; i < app->n_pktq_hwq_out; i++) {
struct app_pktq_hwq_out_params *p_txq = &app->hwq_out_params[i];
p_txq->conf.txq_flags &= ~ETH_TXQ_FLAGS_NOMULTSEGS;
}
}
}
static inline int
app_get_cpu_socket_id(uint32_t pmd_id)
{
int status = rte_eth_dev_socket_id(pmd_id);
return (status != SOCKET_ID_ANY) ? status : 0;
}
static inline int
app_link_rss_enabled(struct app_link_params *cp)
{
return (cp->n_rss_qs) ? 1 : 0;
}
static void
app_link_rss_setup(struct app_link_params *cp)
{
struct rte_eth_dev_info dev_info;
struct rte_eth_rss_reta_entry64 reta_conf[APP_RETA_SIZE_MAX];
uint32_t i;
int status;
/* Get RETA size */
memset(&dev_info, 0, sizeof(dev_info));
rte_eth_dev_info_get(cp->pmd_id, &dev_info);
if (dev_info.reta_size == 0)
rte_panic("%s (%u): RSS setup error (null RETA size)\n",
cp->name, cp->pmd_id);
if (dev_info.reta_size > ETH_RSS_RETA_SIZE_512)
rte_panic("%s (%u): RSS setup error (RETA size too big)\n",
cp->name, cp->pmd_id);
/* Setup RETA contents */
memset(reta_conf, 0, sizeof(reta_conf));
for (i = 0; i < dev_info.reta_size; i++)
reta_conf[i / RTE_RETA_GROUP_SIZE].mask = UINT64_MAX;
for (i = 0; i < dev_info.reta_size; i++) {
uint32_t reta_id = i / RTE_RETA_GROUP_SIZE;
uint32_t reta_pos = i % RTE_RETA_GROUP_SIZE;
uint32_t rss_qs_pos = i % cp->n_rss_qs;
reta_conf[reta_id].reta[reta_pos] =
(uint16_t) cp->rss_qs[rss_qs_pos];
}
/* RETA update */
status = rte_eth_dev_rss_reta_update(cp->pmd_id,
reta_conf,
dev_info.reta_size);
if (status != 0)
rte_panic("%s (%u): RSS setup error (RETA update failed)\n",
cp->name, cp->pmd_id);
}
static void
app_init_link_set_config(struct app_link_params *p)
{
if (p->n_rss_qs) {
p->conf.rxmode.mq_mode = ETH_MQ_RX_RSS;
p->conf.rx_adv_conf.rss_conf.rss_hf = p->rss_proto_ipv4 |
p->rss_proto_ipv6 |
p->rss_proto_l2;
}
}
static void
app_init_link(struct app_params *app)
{
uint32_t i;
app_init_link_frag_ras(app);
for (i = 0; i < app->n_links; i++) {
struct app_link_params *p_link = &app->link_params[i];
uint32_t link_id, n_hwq_in, n_hwq_out, j;
int status;
sscanf(p_link->name, "LINK%" PRIu32, &link_id);
n_hwq_in = app_link_get_n_rxq(app, p_link);
n_hwq_out = app_link_get_n_txq(app, p_link);
app_init_link_set_config(p_link);
APP_LOG(app, HIGH, "Initializing %s (%" PRIu32") "
"(%" PRIu32 " RXQ, %" PRIu32 " TXQ) ...",
p_link->name,
p_link->pmd_id,
n_hwq_in,
n_hwq_out);
/* LINK */
status = rte_eth_dev_configure(
p_link->pmd_id,
n_hwq_in,
n_hwq_out,
&p_link->conf);
if (status < 0)
rte_panic("%s (%" PRId32 "): "
"init error (%" PRId32 ")\n",
p_link->name, p_link->pmd_id, status);
rte_eth_macaddr_get(p_link->pmd_id,
(struct ether_addr *) &p_link->mac_addr);
if (p_link->promisc)
rte_eth_promiscuous_enable(p_link->pmd_id);
/* RXQ */
for (j = 0; j < app->n_pktq_hwq_in; j++) {
struct app_pktq_hwq_in_params *p_rxq =
&app->hwq_in_params[j];
uint32_t rxq_link_id, rxq_queue_id;
uint16_t nb_rxd = p_rxq->size;
sscanf(p_rxq->name, "RXQ%" PRIu32 ".%" PRIu32,
&rxq_link_id, &rxq_queue_id);
if (rxq_link_id != link_id)
continue;
status = rte_eth_dev_adjust_nb_rx_tx_desc(
p_link->pmd_id,
&nb_rxd,
NULL);
if (status < 0)
rte_panic("%s (%" PRIu32 "): "
"%s adjust number of Rx descriptors "
"error (%" PRId32 ")\n",
p_link->name,
p_link->pmd_id,
p_rxq->name,
status);
status = rte_eth_rx_queue_setup(
p_link->pmd_id,
rxq_queue_id,
nb_rxd,
app_get_cpu_socket_id(p_link->pmd_id),
&p_rxq->conf,
app->mempool[p_rxq->mempool_id]);
if (status < 0)
rte_panic("%s (%" PRIu32 "): "
"%s init error (%" PRId32 ")\n",
p_link->name,
p_link->pmd_id,
p_rxq->name,
status);
}
/* TXQ */
for (j = 0; j < app->n_pktq_hwq_out; j++) {
struct app_pktq_hwq_out_params *p_txq =
&app->hwq_out_params[j];
uint32_t txq_link_id, txq_queue_id;
uint16_t nb_txd = p_txq->size;
sscanf(p_txq->name, "TXQ%" PRIu32 ".%" PRIu32,
&txq_link_id, &txq_queue_id);
if (txq_link_id != link_id)
continue;
status = rte_eth_dev_adjust_nb_rx_tx_desc(
p_link->pmd_id,
NULL,
&nb_txd);
if (status < 0)
rte_panic("%s (%" PRIu32 "): "
"%s adjust number of Tx descriptors "
"error (%" PRId32 ")\n",
p_link->name,
p_link->pmd_id,
p_txq->name,
status);
status = rte_eth_tx_queue_setup(
p_link->pmd_id,
txq_queue_id,
nb_txd,
app_get_cpu_socket_id(p_link->pmd_id),
&p_txq->conf);
if (status < 0)
rte_panic("%s (%" PRIu32 "): "
"%s init error (%" PRId32 ")\n",
p_link->name,
p_link->pmd_id,
p_txq->name,
status);
}
/* LINK START */
status = rte_eth_dev_start(p_link->pmd_id);
if (status < 0)
rte_panic("Cannot start %s (error %" PRId32 ")\n",
p_link->name, status);
/* LINK FILTERS */
app_link_set_arp_filter(app, p_link);
app_link_set_tcp_syn_filter(app, p_link);
if (app_link_rss_enabled(p_link))
app_link_rss_setup(p_link);
/* LINK UP */
app_link_up_internal(app, p_link);
}
app_check_link(app);
}
static void
app_init_swq(struct app_params *app)
{
uint32_t i;
for (i = 0; i < app->n_pktq_swq; i++) {
struct app_pktq_swq_params *p = &app->swq_params[i];
unsigned flags = 0;
if (app_swq_get_readers(app, p) == 1)
flags |= RING_F_SC_DEQ;
if (app_swq_get_writers(app, p) == 1)
flags |= RING_F_SP_ENQ;
APP_LOG(app, HIGH, "Initializing %s...", p->name);
app->swq[i] = rte_ring_create(
p->name,
p->size,
p->cpu_socket_id,
flags);
if (app->swq[i] == NULL)
rte_panic("%s init error\n", p->name);
}
}
static void
app_init_tm(struct app_params *app)
{
uint32_t i;
for (i = 0; i < app->n_pktq_tm; i++) {
struct app_pktq_tm_params *p_tm = &app->tm_params[i];
struct app_link_params *p_link;
struct rte_eth_link link_eth_params;
struct rte_sched_port *sched;
uint32_t n_subports, subport_id;
int status;
p_link = app_get_link_for_tm(app, p_tm);
/* LINK */
rte_eth_link_get(p_link->pmd_id, &link_eth_params);
/* TM */
p_tm->sched_port_params.name = p_tm->name;
p_tm->sched_port_params.socket =
app_get_cpu_socket_id(p_link->pmd_id);
p_tm->sched_port_params.rate =
(uint64_t) link_eth_params.link_speed * 1000 * 1000 / 8;
APP_LOG(app, HIGH, "Initializing %s ...", p_tm->name);
sched = rte_sched_port_config(&p_tm->sched_port_params);
if (sched == NULL)
rte_panic("%s init error\n", p_tm->name);
app->tm[i] = sched;
/* Subport */
n_subports = p_tm->sched_port_params.n_subports_per_port;
for (subport_id = 0; subport_id < n_subports; subport_id++) {
uint32_t n_pipes_per_subport, pipe_id;
status = rte_sched_subport_config(sched,
subport_id,
&p_tm->sched_subport_params[subport_id]);
if (status)
rte_panic("%s subport %" PRIu32
" init error (%" PRId32 ")\n",
p_tm->name, subport_id, status);
/* Pipe */
n_pipes_per_subport =
p_tm->sched_port_params.n_pipes_per_subport;
for (pipe_id = 0;
pipe_id < n_pipes_per_subport;
pipe_id++) {
int profile_id = p_tm->sched_pipe_to_profile[
subport_id * APP_MAX_SCHED_PIPES +
pipe_id];
if (profile_id == -1)
continue;
status = rte_sched_pipe_config(sched,
subport_id,
pipe_id,
profile_id);
if (status)
rte_panic("%s subport %" PRIu32
" pipe %" PRIu32
" (profile %" PRId32 ") "
"init error (% " PRId32 ")\n",
p_tm->name, subport_id, pipe_id,
profile_id, status);
}
}
}
}
#ifndef RTE_EXEC_ENV_LINUXAPP
static void
app_init_tap(struct app_params *app) {
if (app->n_pktq_tap == 0)
return;
rte_panic("TAP device not supported.\n");
}
#else
static void
app_init_tap(struct app_params *app)
{
uint32_t i;
for (i = 0; i < app->n_pktq_tap; i++) {
struct app_pktq_tap_params *p_tap = &app->tap_params[i];
struct ifreq ifr;
int fd, status;
APP_LOG(app, HIGH, "Initializing %s ...", p_tap->name);
fd = open("/dev/net/tun", O_RDWR | O_NONBLOCK);
if (fd < 0)
rte_panic("Cannot open file /dev/net/tun\n");
memset(&ifr, 0, sizeof(ifr));
ifr.ifr_flags = IFF_TAP | IFF_NO_PI; /* No packet information */
snprintf(ifr.ifr_name, IFNAMSIZ, "%s", p_tap->name);
status = ioctl(fd, TUNSETIFF, (void *) &ifr);
if (status < 0)
rte_panic("TAP setup error\n");
app->tap[i] = fd;
}
}
