/*
* test data manipulation in mbuf with non-ascii data
*/
static int
test_pktmbuf_with_non_ascii_data(void)
{
struct rte_mbuf *m = NULL;
char *data;
m = rte_pktmbuf_alloc(pktmbuf_pool);
if (m == NULL)
GOTO_FAIL("Cannot allocate mbuf");
if (rte_pktmbuf_pkt_len(m) != 0)
GOTO_FAIL("Bad length");
data = rte_pktmbuf_append(m, MBUF_TEST_DATA_LEN);
if (data == NULL)
GOTO_FAIL("Cannot append data");
if (rte_pktmbuf_pkt_len(m) != MBUF_TEST_DATA_LEN)
GOTO_FAIL("Bad pkt length");
if (rte_pktmbuf_data_len(m) != MBUF_TEST_DATA_LEN)
GOTO_FAIL("Bad data length");
memset(data, 0xff, rte_pktmbuf_pkt_len(m));
if (!rte_pktmbuf_is_contiguous(m))
GOTO_FAIL("Buffer should be continuous");
rte_pktmbuf_dump(m, MBUF_TEST_DATA_LEN);
rte_pktmbuf_free(m);
return 0;
fail:
if(m) {
rte_pktmbuf_free(m);
}
return -1;
}
/* Sends 'num_pkts' 'packets' and 'request' data to datapath. */
int
dpdk_link_send_bulk(struct dpif_dpdk_message *request,
const struct ofpbuf *const *packets, size_t num_pkts)
{
struct rte_mbuf *mbufs[PKT_BURST_SIZE] = {NULL};
uint8_t *mbuf_data = NULL;
int i = 0;
int ret = 0;
if (num_pkts > PKT_BURST_SIZE) {
return EINVAL;
}
DPDK_DEBUG()
for (i = 0; i < num_pkts; i++) {
mbufs[i] = rte_pktmbuf_alloc(mp);
if (!mbufs[i]) {
return ENOBUFS;
}
mbuf_data = rte_pktmbuf_mtod(mbufs[i], uint8_t *);
rte_memcpy(mbuf_data, &request[i], sizeof(request[i]));
if (request->type == DPIF_DPDK_PACKET_FAMILY) {
mbuf_data = mbuf_data + sizeof(request[i]);
if (likely(packets[i]->size <= (mbufs[i]->buf_len - sizeof(request[i])))) {
rte_memcpy(mbuf_data, packets[i]->data, packets[i]->size);
rte_pktmbuf_data_len(mbufs[i]) =
sizeof(request[i]) + packets[i]->size;
rte_pktmbuf_pkt_len(mbufs[i]) = rte_pktmbuf_data_len(mbufs[i]);
} else {
RTE_LOG(ERR, APP, "%s, %d: %s", __FUNCTION__, __LINE__,
"memcpy prevented: packet size exceeds available mbuf space");
for (i = 0; i < num_pkts; i++) {
rte_pktmbuf_free(mbufs[i]);
}
return ENOMEM;
}
} else {
rte_pktmbuf_data_len(mbufs[i]) = sizeof(request[i]);
rte_pktmbuf_pkt_len(mbufs[i]) = rte_pktmbuf_data_len(mbufs[i]);
}
}
ret = rte_ring_sp_enqueue_bulk(message_ring, (void * const *)mbufs, num_pkts);
if (ret == -ENOBUFS) {
for (i = 0; i < num_pkts; i++) {
rte_pktmbuf_free(mbufs[i]);
}
ret = ENOBUFS;
} else if (unlikely(ret == -EDQUOT)) {
ret = EDQUOT;
}
return ret;
}
void dpdk_ipDeFragment(void *handle, struct rte_mbuf *m){
struct ip * iphead = (struct ip *)m;
struct sk_buff s;
s.data = (char *)(m - sizeof(struct ip));
s.truesize = rte_pktmbuf_pkt_len(m) - sizeof(struct ip);
ipDeFragment(handle, iphead, &s);
}
int pcap_next_ex(pcap_t *p, struct pcap_pkthdr **pkt_header,
const u_char **pkt_data)
{
struct rte_mbuf* mbuf = NULL;
int len = 0;
if (p == NULL || pkt_header == NULL || pkt_data == NULL ||
p->deviceId < 0 || p->deviceId > RTE_MAX_ETHPORTS)
{
snprintf (errbuf_g, PCAP_ERRBUF_SIZE, "Invalid parameter");
return DPDKPCAP_FAILURE;
}
debug("Receiving a packet on port %d\n", p->deviceId);
while (!rte_eth_rx_burst(p->deviceId, 0, &mbuf, 1))
{
}
len = rte_pktmbuf_pkt_len(mbuf);
pktHeader_g.len = len;
*pkt_header = &pktHeader_g;
rte_memcpy((void*)data_g, rte_pktmbuf_mtod(mbuf, void*), len);
*pkt_data = data_g;
rte_pktmbuf_free(mbuf);
return 1;
}
void dump_ring(CLQManager *a_pclsCLQ, char *a_pszQName, uint32_t a_unStartIdx, uint32_t a_unEndIdx)
{
struct rte_ring *pstRing = NULL;
struct rte_mbuf *m = NULL;
int ret = 0;
uint32_t unMask = 0;
uint32_t unStartIdx = 0;
uint32_t unEndIdx = 0;
uint32_t unIdx = 0;
ret = a_pclsCLQ->CreateRing(a_pszQName, &pstRing);
if(ret != E_Q_EXIST)
{
printf("There is no Queue (%s) \n", a_pszQName);
if(ret == 0)
{
a_pclsCLQ->DeleteQueue(a_pszQName);
}
}
if(pstRing != NULL)
{
unMask = pstRing->prod.mask;
unStartIdx = pstRing->cons.tail;
unEndIdx = pstRing->prod.tail;
if(a_unEndIdx > unEndIdx)
{
printf("Invalid End idx %u\n", a_unEndIdx);
return ;
}
if(a_unEndIdx < a_unStartIdx)
{
printf("Invalid Start Idx %u, End Idx %u\n", a_unStartIdx, a_unEndIdx);
return ;
}
if(a_unStartIdx)
unStartIdx = a_unStartIdx;
if(a_unEndIdx)
unEndIdx = a_unEndIdx;
printf("Start Idx %u, End Idx %u\n", unStartIdx, unEndIdx);
for(uint32_t i = unStartIdx; i < unEndIdx ; i++)
{
unIdx = i & unMask ;
m = (struct rte_mbuf*)pstRing->ring[unIdx];
printf("idx : [%8u], total_len : [%5u], data_len : [%5u], seg_cnt : [%2u], Data : %s\n"
, i
, rte_pktmbuf_pkt_len(m)
, rte_pktmbuf_data_len(m)
, m->nb_segs
, rte_pktmbuf_mtod(m, char*)
);
}
}
static inline struct rte_mbuf *ipaugenblick_get_from_shadow(int sock)
{
struct rte_mbuf *mbuf = NULL;
if(local_socket_descriptors[sock].shadow) {
mbuf = local_socket_descriptors[sock].shadow;
local_socket_descriptors[sock].shadow = NULL;
rte_pktmbuf_data_len(mbuf) = local_socket_descriptors[sock].shadow_len_remainder;
mbuf->data_off += local_socket_descriptors[sock].shadow_len_delievered;
if(mbuf->next) {
rte_pktmbuf_pkt_len(mbuf) = rte_pktmbuf_data_len(mbuf) + rte_pktmbuf_pkt_len(mbuf->next);
}
else {
rte_pktmbuf_pkt_len(mbuf) = rte_pktmbuf_data_len(mbuf);
}
mbuf->next = local_socket_descriptors[sock].shadow_next;
local_socket_descriptors[sock].shadow_next = NULL;
}
return mbuf;
}
/*
* Function sends unmatched packets to vswitchd.
*/
void
send_packet_to_vswitchd(struct rte_mbuf *mbuf, struct dpdk_upcall *info)
{
int rslt = 0;
void *mbuf_ptr = NULL;
const uint64_t dpif_send_tsc =
(rte_get_tsc_hz() + US_PER_S - 1) / US_PER_S * DPIF_SEND_US;
uint64_t cur_tsc = 0;
uint64_t diff_tsc = 0;
static uint64_t prev_tsc = 0;
/* send one packet, delete information about segments */
rte_pktmbuf_pkt_len(mbuf) = rte_pktmbuf_data_len(mbuf);
/* allocate space before the packet for the upcall info */
mbuf_ptr = rte_pktmbuf_prepend(mbuf, sizeof(*info));
if (mbuf_ptr == NULL) {
printf("Cannot prepend upcall info\n");
rte_pktmbuf_free(mbuf);
stats_vswitch_tx_drop_increment(INC_BY_1);
stats_vport_tx_drop_increment(VSWITCHD, INC_BY_1);
return;
}
rte_memcpy(mbuf_ptr, info, sizeof(*info));
/* send the packet and the upcall info to the daemon */
rslt = rte_ring_mp_enqueue(vswitchd_packet_ring, mbuf);
if (rslt < 0) {
if (rslt == -ENOBUFS) {
rte_pktmbuf_free(mbuf);
stats_vswitch_tx_drop_increment(INC_BY_1);
stats_vport_tx_drop_increment(VSWITCHD, INC_BY_1);
return;
} else {
stats_vport_overrun_increment(VSWITCHD, INC_BY_1);
}
}
stats_vport_tx_increment(VSWITCHD, INC_BY_1);
cur_tsc = rte_rdtsc();
diff_tsc = cur_tsc - prev_tsc;
prev_tsc = cur_tsc;
/* Only signal the daemon after 100 milliseconds */
if (unlikely(diff_tsc > dpif_send_tsc))
send_signal_to_dpif();
}
static struct rte_mbuf * dpdk_replicate_packet_mb (vlib_buffer_t * b)
{
vlib_main_t * vm = vlib_get_main();
vlib_buffer_main_t * bm = vm->buffer_main;
struct rte_mbuf * first_mb = 0, * new_mb, * pkt_mb, ** prev_mb_next = 0;
u8 nb_segs, nb_segs_left;
u32 copy_bytes;
unsigned socket_id = rte_socket_id();
ASSERT (bm->pktmbuf_pools[socket_id]);
pkt_mb = ((struct rte_mbuf *)b)-1;
nb_segs = pkt_mb->nb_segs;
for (nb_segs_left = nb_segs; nb_segs_left; nb_segs_left--)
{
if (PREDICT_FALSE(pkt_mb == 0))
{
clib_warning ("Missing %d mbuf chain segment(s): "
"(nb_segs = %d, nb_segs_left = %d)!",
nb_segs - nb_segs_left, nb_segs, nb_segs_left);
if (first_mb)
rte_pktmbuf_free(first_mb);
return NULL;
}
new_mb = rte_pktmbuf_alloc (bm->pktmbuf_pools[socket_id]);
if (PREDICT_FALSE(new_mb == 0))
{
if (first_mb)
rte_pktmbuf_free(first_mb);
return NULL;
}
/*
* Copy packet info into 1st segment.
*/
if (first_mb == 0)
{
first_mb = new_mb;
rte_pktmbuf_pkt_len (first_mb) = pkt_mb->pkt_len;
first_mb->nb_segs = pkt_mb->nb_segs;
first_mb->port = pkt_mb->port;
#ifdef DAW_FIXME // TX Offload support TBD
first_mb->vlan_macip = pkt_mb->vlan_macip;
first_mb->hash = pkt_mb->hash;
first_mb->ol_flags = pkt_mb->ol_flags
#endif
}
else
{
static int
testclone_testupdate_testdetach(void)
{
#ifndef RTE_MBUF_SCATTER_GATHER
return 0;
#else
struct rte_mbuf *mc = NULL;
struct rte_mbuf *clone = NULL;
/* alloc a mbuf */
mc = rte_pktmbuf_alloc(pktmbuf_pool);
if (mc == NULL)
GOTO_FAIL("ooops not allocating mbuf");
if (rte_pktmbuf_pkt_len(mc) != 0)
GOTO_FAIL("Bad length");
/* clone the allocated mbuf */
clone = rte_pktmbuf_clone(mc, pktmbuf_pool);
if (clone == NULL)
GOTO_FAIL("cannot clone data\n");
rte_pktmbuf_free(clone);
mc->pkt.next = rte_pktmbuf_alloc(pktmbuf_pool);
if(mc->pkt.next == NULL)
GOTO_FAIL("Next Pkt Null\n");
clone = rte_pktmbuf_clone(mc, pktmbuf_pool);
if (clone == NULL)
GOTO_FAIL("cannot clone data\n");
/* free mbuf */
rte_pktmbuf_free(mc);
rte_pktmbuf_free(clone);
mc = NULL;
clone = NULL;
return 0;
fail:
if (mc)
rte_pktmbuf_free(mc);
return -1;
#endif /* RTE_MBUF_SCATTER_GATHER */
}
static int
usock_mbuf_write(struct vr_usocket *usockp, struct rte_mbuf *mbuf)
{
unsigned int i, pkt_len;
struct msghdr mhdr;
struct rte_mbuf *m;
struct iovec *iov;
if (!mbuf)
return 0;
pkt_len = rte_pktmbuf_pkt_len(mbuf);
if (!pkt_len)
return 0;
iov = usockp->usock_iovec;
m = mbuf;
for (i = 0; (m && (i < PKT0_MAX_IOV_LEN)); i++) {
iov->iov_base = rte_pktmbuf_mtod(m, unsigned char *);
iov->iov_len = rte_pktmbuf_data_len(m);
m = m->next;
iov++;
}
if ((i == PKT0_MAX_IOV_LEN) && m)
usockp->usock_pkt_truncated++;
mhdr.msg_name = NULL;
mhdr.msg_namelen = 0;
mhdr.msg_iov = usockp->usock_iovec;
mhdr.msg_iovlen = i;
mhdr.msg_control = NULL;
mhdr.msg_controllen = 0;
mhdr.msg_flags = 0;
#ifdef VR_DPDK_USOCK_DUMP
RTE_LOG(DEBUG, USOCK, "%s[%lx]: FD %d sending message\n", __func__,
pthread_self(), usockp->usock_fd);
rte_hexdump(stdout, "usock message dump:", &mhdr, sizeof(mhdr));
#endif
return sendmsg(usockp->usock_fd, &mhdr, MSG_DONTWAIT);
}
static inline int
app_pkt_handle(struct rte_mbuf *pkt, uint64_t time)
{
uint8_t input_color, output_color;
uint8_t *pkt_data = rte_pktmbuf_mtod(pkt, uint8_t *);
uint32_t pkt_len = rte_pktmbuf_pkt_len(pkt) - sizeof(struct ether_hdr);
uint8_t flow_id = (uint8_t)(pkt_data[APP_PKT_FLOW_POS] & (APP_FLOWS_MAX - 1));
input_color = pkt_data[APP_PKT_COLOR_POS];
enum policer_action action;
/* color input is not used for blind modes */
output_color = (uint8_t) FUNC_METER(&app_flows[flow_id], time, pkt_len,
(enum rte_meter_color) input_color);
/* Apply policing and set the output color */
action = policer_table[input_color][output_color];
app_set_pkt_color(pkt_data, action);
return action;
}
/*
* Function sends unmatched packets to vswitchd.
*/
static void
send_packet_to_vswitchd(struct rte_mbuf *mbuf, struct dpdk_upcall *info)
{
int rslt = 0;
struct statistics *vswd_stat = NULL;
void *mbuf_ptr = NULL;
vswd_stat = &vport_stats[VSWITCHD];
/* send one packet, delete information about segments */
rte_pktmbuf_pkt_len(mbuf) = rte_pktmbuf_data_len(mbuf);
/* allocate space before the packet for the upcall info */
mbuf_ptr = rte_pktmbuf_prepend(mbuf, sizeof(*info));
if (mbuf_ptr == NULL) {
printf("Cannot prepend upcall info\n");
rte_pktmbuf_free(mbuf);
switch_tx_drop++;
vswd_stat->tx_drop++;
return;
}
rte_memcpy(mbuf_ptr, info, sizeof(*info));
/* send the packet and the upcall info to the daemon */
rslt = rte_ring_sp_enqueue(vswitch_packet_ring, mbuf);
if (rslt < 0) {
if (rslt == -ENOBUFS) {
rte_pktmbuf_free(mbuf);
switch_tx_drop++;
vswd_stat->tx_drop++;
return;
} else {
overruns++;
}
}
vswd_stat->tx++;
}
static uint32_t send_pkts(uint8_t port, struct rte_mempool* pool) {
static uint64_t seq = 0;
// alloc bufs
struct rte_mbuf* bufs[BATCH_SIZE];
uint32_t i;
for (i = 0; i < BATCH_SIZE; i++) {
struct rte_mbuf* buf = rte_pktmbuf_alloc(pool);
rte_pktmbuf_data_len(buf) = 60;
rte_pktmbuf_pkt_len(buf) = 60;
bufs[i] = buf;
// write seq number
uint64_t* pkt = rte_pktmbuf_mtod(buf, uint64_t*);
pkt[0] = seq++;
}
// send pkts
uint32_t sent = 0;
while (1) {
sent += rte_eth_tx_burst(port, 0, bufs + sent, BATCH_SIZE - sent);
if (sent >= BATCH_SIZE) {
return sent;
}
}
}
int
qat_comp_build_request(void *in_op, uint8_t *out_msg,
void *op_cookie,
enum qat_device_gen qat_dev_gen __rte_unused)
{
struct rte_comp_op *op = in_op;
struct qat_comp_op_cookie *cookie =
(struct qat_comp_op_cookie *)op_cookie;
struct qat_comp_xform *qat_xform = op->private_xform;
const uint8_t *tmpl = (uint8_t *)&qat_xform->qat_comp_req_tmpl;
struct icp_qat_fw_comp_req *comp_req =
(struct icp_qat_fw_comp_req *)out_msg;
if (unlikely(op->op_type != RTE_COMP_OP_STATELESS)) {
QAT_DP_LOG(ERR, "QAT PMD only supports stateless compression "
"operation requests, op (%p) is not a "
"stateless operation.", op);
op->status = RTE_COMP_OP_STATUS_INVALID_ARGS;
return -EINVAL;
}
rte_mov128(out_msg, tmpl);
comp_req->comn_mid.opaque_data = (uint64_t)(uintptr_t)op;
/* common for sgl and flat buffers */
comp_req->comp_pars.comp_len = op->src.length;
comp_req->comp_pars.out_buffer_sz = rte_pktmbuf_pkt_len(op->m_dst) -
op->dst.offset;
if (op->m_src->next != NULL || op->m_dst->next != NULL) {
/* sgl */
int ret = 0;
ICP_QAT_FW_COMN_PTR_TYPE_SET(comp_req->comn_hdr.comn_req_flags,
QAT_COMN_PTR_TYPE_SGL);
ret = qat_sgl_fill_array(op->m_src,
op->src.offset,
&cookie->qat_sgl_src,
op->src.length,
RTE_PMD_QAT_COMP_SGL_MAX_SEGMENTS);
if (ret) {
QAT_DP_LOG(ERR, "QAT PMD Cannot fill source sgl array");
op->status = RTE_COMP_OP_STATUS_INVALID_ARGS;
return ret;
}
ret = qat_sgl_fill_array(op->m_dst,
op->dst.offset,
&cookie->qat_sgl_dst,
comp_req->comp_pars.out_buffer_sz,
RTE_PMD_QAT_COMP_SGL_MAX_SEGMENTS);
if (ret) {
QAT_DP_LOG(ERR, "QAT PMD Cannot fill dest. sgl array");
op->status = RTE_COMP_OP_STATUS_INVALID_ARGS;
return ret;
}
comp_req->comn_mid.src_data_addr =
cookie->qat_sgl_src_phys_addr;
comp_req->comn_mid.dest_data_addr =
cookie->qat_sgl_dst_phys_addr;
comp_req->comn_mid.src_length = 0;
comp_req->comn_mid.dst_length = 0;
} else {
/* flat aka linear buffer */
ICP_QAT_FW_COMN_PTR_TYPE_SET(comp_req->comn_hdr.comn_req_flags,
QAT_COMN_PTR_TYPE_FLAT);
comp_req->comn_mid.src_length = op->src.length;
comp_req->comn_mid.dst_length =
comp_req->comp_pars.out_buffer_sz;
comp_req->comn_mid.src_data_addr =
rte_pktmbuf_mtophys_offset(op->m_src, op->src.offset);
comp_req->comn_mid.dest_data_addr =
rte_pktmbuf_mtophys_offset(op->m_dst, op->dst.offset);
}
#if RTE_LOG_DP_LEVEL >= RTE_LOG_DEBUG
QAT_DP_LOG(DEBUG, "Direction: %s",
qat_xform->qat_comp_request_type == QAT_COMP_REQUEST_DECOMPRESS ?
"decompression" : "compression");
QAT_DP_HEXDUMP_LOG(DEBUG, "qat compression message:", comp_req,
sizeof(struct icp_qat_fw_comp_req));
#endif
return 0;
}
virtio_dev_rx(struct virtio_net *dev, uint16_t queue_id,
struct rte_mbuf **pkts, uint32_t count)
{
struct vhost_virtqueue *vq;
struct vring_desc *desc;
struct rte_mbuf *buff;
/* The virtio_hdr is initialised to 0. */
struct virtio_net_hdr_mrg_rxbuf virtio_hdr = {{0, 0, 0, 0, 0, 0}, 0};
uint64_t buff_addr = 0;
uint64_t buff_hdr_addr = 0;
uint32_t head[MAX_PKT_BURST];
uint32_t head_idx, packet_success = 0;
uint16_t avail_idx, res_cur_idx;
uint16_t res_base_idx, res_end_idx;
uint16_t free_entries;
uint8_t success = 0;
LOG_DEBUG(VHOST_DATA, "(%"PRIu64") virtio_dev_rx()\n", dev->device_fh);
if (unlikely(queue_id != VIRTIO_RXQ)) {
LOG_DEBUG(VHOST_DATA, "mq isn't supported in this version.\n");
return 0;
}
vq = dev->virtqueue[VIRTIO_RXQ];
count = (count > MAX_PKT_BURST) ? MAX_PKT_BURST : count;
/*
* As many data cores may want access to available buffers,
* they need to be reserved.
*/
do {
res_base_idx = vq->last_used_idx_res;
avail_idx = *((volatile uint16_t *)&vq->avail->idx);
free_entries = (avail_idx - res_base_idx);
/*check that we have enough buffers*/
if (unlikely(count > free_entries))
count = free_entries;
if (count == 0)
return 0;
res_end_idx = res_base_idx + count;
/* vq->last_used_idx_res is atomically updated. */
/* TODO: Allow to disable cmpset if no concurrency in application. */
success = rte_atomic16_cmpset(&vq->last_used_idx_res,
res_base_idx, res_end_idx);
} while (unlikely(success == 0));
res_cur_idx = res_base_idx;
LOG_DEBUG(VHOST_DATA, "(%"PRIu64") Current Index %d| End Index %d\n",
dev->device_fh, res_cur_idx, res_end_idx);
/* Prefetch available ring to retrieve indexes. */
rte_prefetch0(&vq->avail->ring[res_cur_idx & (vq->size - 1)]);
/* Retrieve all of the head indexes first to avoid caching issues. */
for (head_idx = 0; head_idx < count; head_idx++)
head[head_idx] = vq->avail->ring[(res_cur_idx + head_idx) &
(vq->size - 1)];
/*Prefetch descriptor index. */
rte_prefetch0(&vq->desc[head[packet_success]]);
while (res_cur_idx != res_end_idx) {
uint32_t offset = 0, vb_offset = 0;
uint32_t pkt_len, len_to_cpy, data_len, total_copied = 0;
uint8_t hdr = 0, uncompleted_pkt = 0;
/* Get descriptor from available ring */
desc = &vq->desc[head[packet_success]];
buff = pkts[packet_success];
/* Convert from gpa to vva (guest physical addr -> vhost virtual addr) */
buff_addr = gpa_to_vva(dev, desc->addr);
/* Prefetch buffer address. */
rte_prefetch0((void *)(uintptr_t)buff_addr);
/* Copy virtio_hdr to packet and increment buffer address */
buff_hdr_addr = buff_addr;
/*
* If the descriptors are chained the header and data are
* placed in separate buffers.
*/
if ((desc->flags & VRING_DESC_F_NEXT) &&
(desc->len == vq->vhost_hlen)) {
desc = &vq->desc[desc->next];
/* Buffer address translation. */
buff_addr = gpa_to_vva(dev, desc->addr);
} else {
vb_offset += vq->vhost_hlen;
hdr = 1;
}
pkt_len = rte_pktmbuf_pkt_len(buff);
data_len = rte_pktmbuf_data_len(buff);
len_to_cpy = RTE_MIN(data_len,
hdr ? desc->len - vq->vhost_hlen : desc->len);
while (total_copied < pkt_len) {
/* Copy mbuf data to buffer */
rte_memcpy((void *)(uintptr_t)(buff_addr + vb_offset),
(const void *)(rte_pktmbuf_mtod(buff, const char *) + offset),
len_to_cpy);
PRINT_PACKET(dev, (uintptr_t)(buff_addr + vb_offset),
len_to_cpy, 0);
offset += len_to_cpy;
vb_offset += len_to_cpy;
total_copied += len_to_cpy;
/* The whole packet completes */
if (total_copied == pkt_len)
break;
/* The current segment completes */
if (offset == data_len) {
buff = buff->next;
offset = 0;
data_len = rte_pktmbuf_data_len(buff);
}
/* The current vring descriptor done */
if (vb_offset == desc->len) {
if (desc->flags & VRING_DESC_F_NEXT) {
desc = &vq->desc[desc->next];
buff_addr = gpa_to_vva(dev, desc->addr);
vb_offset = 0;
} else {
/* Room in vring buffer is not enough */
uncompleted_pkt = 1;
break;
}
}
len_to_cpy = RTE_MIN(data_len - offset, desc->len - vb_offset);
};
/* Update used ring with desc information */
vq->used->ring[res_cur_idx & (vq->size - 1)].id =
head[packet_success];
/* Drop the packet if it is uncompleted */
if (unlikely(uncompleted_pkt == 1))
vq->used->ring[res_cur_idx & (vq->size - 1)].len =
vq->vhost_hlen;
else
vq->used->ring[res_cur_idx & (vq->size - 1)].len =
pkt_len + vq->vhost_hlen;
res_cur_idx++;
packet_success++;
if (unlikely(uncompleted_pkt == 1))
continue;
rte_memcpy((void *)(uintptr_t)buff_hdr_addr,
(const void *)&virtio_hdr, vq->vhost_hlen);
PRINT_PACKET(dev, (uintptr_t)buff_hdr_addr, vq->vhost_hlen, 1);
if (res_cur_idx < res_end_idx) {
/* Prefetch descriptor index. */
rte_prefetch0(&vq->desc[head[packet_success]]);
}
}
rte_compiler_barrier();
/* Wait until it's our turn to add our buffer to the used ring. */
while (unlikely(vq->last_used_idx != res_base_idx))
rte_pause();
*(volatile uint16_t *)&vq->used->idx += count;
vq->last_used_idx = res_end_idx;
/* flush used->idx update before we read avail->flags. */
rte_mb();
/* Kick the guest if necessary. */
if (!(vq->avail->flags & VRING_AVAIL_F_NO_INTERRUPT))
eventfd_write((int)vq->callfd, 1);
return count;
}
uint16_t
fm10k_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
uint16_t nb_pkts)
{
struct rte_mbuf *mbuf;
union fm10k_rx_desc desc;
struct fm10k_rx_queue *q = rx_queue;
uint16_t count = 0;
int alloc = 0;
uint16_t next_dd;
int ret;
next_dd = q->next_dd;
nb_pkts = RTE_MIN(nb_pkts, q->alloc_thresh);
for (count = 0; count < nb_pkts; ++count) {
mbuf = q->sw_ring[next_dd];
desc = q->hw_ring[next_dd];
if (!(desc.d.staterr & FM10K_RXD_STATUS_DD))
break;
#ifdef RTE_LIBRTE_FM10K_DEBUG_RX
dump_rxd(&desc);
#endif
rte_pktmbuf_pkt_len(mbuf) = desc.w.length;
rte_pktmbuf_data_len(mbuf) = desc.w.length;
mbuf->ol_flags = 0;
#ifdef RTE_LIBRTE_FM10K_RX_OLFLAGS_ENABLE
rx_desc_to_ol_flags(mbuf, &desc);
#endif
mbuf->hash.rss = desc.d.rss;
rx_pkts[count] = mbuf;
if (++next_dd == q->nb_desc) {
next_dd = 0;
alloc = 1;
}
/* Prefetch next mbuf while processing current one. */
rte_prefetch0(q->sw_ring[next_dd]);
/*
* When next RX descriptor is on a cache-line boundary,
* prefetch the next 4 RX descriptors and the next 8 pointers
* to mbufs.
*/
if ((next_dd & 0x3) == 0) {
rte_prefetch0(&q->hw_ring[next_dd]);
rte_prefetch0(&q->sw_ring[next_dd]);
}
}
q->next_dd = next_dd;
if ((q->next_dd > q->next_trigger) || (alloc == 1)) {
ret = rte_mempool_get_bulk(q->mp,
(void **)&q->sw_ring[q->next_alloc],
q->alloc_thresh);
if (unlikely(ret != 0)) {
uint8_t port = q->port_id;
PMD_RX_LOG(ERR, "Failed to alloc mbuf");
/*
* Need to restore next_dd if we cannot allocate new
* buffers to replenish the old ones.
*/
q->next_dd = (q->next_dd + q->nb_desc - count) %
q->nb_desc;
rte_eth_devices[port].data->rx_mbuf_alloc_failed++;
return 0;
}
for (; q->next_alloc <= q->next_trigger; ++q->next_alloc) {
mbuf = q->sw_ring[q->next_alloc];
/* setup static mbuf fields */
fm10k_pktmbuf_reset(mbuf, q->port_id);
/* write descriptor */
desc.q.pkt_addr = MBUF_DMA_ADDR_DEFAULT(mbuf);
desc.q.hdr_addr = MBUF_DMA_ADDR_DEFAULT(mbuf);
q->hw_ring[q->next_alloc] = desc;
}
FM10K_PCI_REG_WRITE(q->tail_ptr, q->next_trigger);
q->next_trigger += q->alloc_thresh;
if (q->next_trigger >= q->nb_desc) {
q->next_trigger = q->alloc_thresh - 1;
q->next_alloc = 0;
}
}
return count;
}
uint16_t
vmxnet3_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts,
uint16_t nb_pkts)
{
uint16_t nb_tx;
Vmxnet3_TxDesc *txd = NULL;
vmxnet3_buf_info_t *tbi = NULL;
struct vmxnet3_hw *hw;
struct rte_mbuf *txm;
vmxnet3_tx_queue_t *txq = tx_queue;
hw = txq->hw;
if (unlikely(txq->stopped)) {
PMD_TX_LOG(DEBUG, "Tx queue is stopped.");
return 0;
}
/* Free up the comp_descriptors aggressively */
vmxnet3_tq_tx_complete(txq);
nb_tx = 0;
while (nb_tx < nb_pkts) {
if (vmxnet3_cmd_ring_desc_avail(&txq->cmd_ring)) {
int copy_size = 0;
txm = tx_pkts[nb_tx];
/* Don't support scatter packets yet, free them if met */
if (txm->nb_segs != 1) {
if (vmxnet3_xmit_convert_callback ){
txm=vmxnet3_xmit_convert_callback(txm);
}else{
txq->stats.drop_total++;
nb_tx++;
rte_pktmbuf_free(txm);
continue;
}
}
if (!txm) {
txq->stats.drop_total++;
nb_tx++;
continue;
}
/* Needs to minus ether header len */
if (txm->data_len > (hw->cur_mtu + ETHER_HDR_LEN)) {
PMD_TX_LOG(DEBUG, "Packet data_len higher than MTU");
rte_pktmbuf_free(txm);
txq->stats.drop_total++;
nb_tx++;
continue;
}
txd = (Vmxnet3_TxDesc *)(txq->cmd_ring.base + txq->cmd_ring.next2fill);
if (rte_pktmbuf_pkt_len(txm) <= VMXNET3_HDR_COPY_SIZE) {
struct Vmxnet3_TxDataDesc *tdd;
tdd = txq->data_ring.base + txq->cmd_ring.next2fill;
copy_size = rte_pktmbuf_pkt_len(txm);
rte_memcpy(tdd->data, rte_pktmbuf_mtod(txm, char *), copy_size);
}
/* Fill the tx descriptor */
tbi = txq->cmd_ring.buf_info + txq->cmd_ring.next2fill;
tbi->bufPA = RTE_MBUF_DATA_DMA_ADDR(txm);
if (copy_size)
txd->addr = rte_cpu_to_le_64(txq->data_ring.basePA +
txq->cmd_ring.next2fill *
sizeof(struct Vmxnet3_TxDataDesc));
else
txd->addr = tbi->bufPA;
txd->len = txm->data_len;
/* Mark the last descriptor as End of Packet. */
txd->cq = 1;
txd->eop = 1;
/* Add VLAN tag if requested */
if (txm->ol_flags & PKT_TX_VLAN_PKT) {
txd->ti = 1;
txd->tci = rte_cpu_to_le_16(txm->vlan_tci);
}
/* Record current mbuf for freeing it later in tx complete */
#ifdef RTE_LIBRTE_VMXNET3_DEBUG_DRIVER
VMXNET3_ASSERT(txm);
#endif
tbi->m = txm;
/* Set the offloading mode to default */
txd->hlen = 0;
txd->om = VMXNET3_OM_NONE;
txd->msscof = 0;
/* finally flip the GEN bit of the SOP desc */
txd->gen = txq->cmd_ring.gen;
txq->shared->ctrl.txNumDeferred++;
/* move to the next2fill descriptor */
vmxnet3_cmd_ring_adv_next2fill(&txq->cmd_ring);
nb_tx++;
} else {
/*
* test data manipulation in mbuf
*/
static int
test_one_pktmbuf(void)
{
struct rte_mbuf *m = NULL;
char *data, *data2, *hdr;
unsigned i;
printf("Test pktmbuf API\n");
/* alloc a mbuf */
m = rte_pktmbuf_alloc(pktmbuf_pool);
if (m == NULL)
GOTO_FAIL("Cannot allocate mbuf");
if (rte_pktmbuf_pkt_len(m) != 0)
GOTO_FAIL("Bad length");
rte_pktmbuf_dump(m, 0);
/* append data */
data = rte_pktmbuf_append(m, MBUF_TEST_DATA_LEN);
if (data == NULL)
GOTO_FAIL("Cannot append data");
if (rte_pktmbuf_pkt_len(m) != MBUF_TEST_DATA_LEN)
GOTO_FAIL("Bad pkt length");
if (rte_pktmbuf_data_len(m) != MBUF_TEST_DATA_LEN)
GOTO_FAIL("Bad data length");
memset(data, 0x66, rte_pktmbuf_pkt_len(m));
if (!rte_pktmbuf_is_contiguous(m))
GOTO_FAIL("Buffer should be continuous");
rte_pktmbuf_dump(m, MBUF_TEST_DATA_LEN);
rte_pktmbuf_dump(m, 2*MBUF_TEST_DATA_LEN);
/* this append should fail */
data2 = rte_pktmbuf_append(m, (uint16_t)(rte_pktmbuf_tailroom(m) + 1));
if (data2 != NULL)
GOTO_FAIL("Append should not succeed");
/* append some more data */
data2 = rte_pktmbuf_append(m, MBUF_TEST_DATA_LEN2);
if (data2 == NULL)
GOTO_FAIL("Cannot append data");
if (rte_pktmbuf_pkt_len(m) != MBUF_TEST_DATA_LEN + MBUF_TEST_DATA_LEN2)
GOTO_FAIL("Bad pkt length");
if (rte_pktmbuf_data_len(m) != MBUF_TEST_DATA_LEN + MBUF_TEST_DATA_LEN2)
GOTO_FAIL("Bad data length");
if (!rte_pktmbuf_is_contiguous(m))
GOTO_FAIL("Buffer should be continuous");
/* trim data at the end of mbuf */
if (rte_pktmbuf_trim(m, MBUF_TEST_DATA_LEN2) < 0)
GOTO_FAIL("Cannot trim data");
if (rte_pktmbuf_pkt_len(m) != MBUF_TEST_DATA_LEN)
GOTO_FAIL("Bad pkt length");
if (rte_pktmbuf_data_len(m) != MBUF_TEST_DATA_LEN)
GOTO_FAIL("Bad data length");
if (!rte_pktmbuf_is_contiguous(m))
GOTO_FAIL("Buffer should be continuous");
/* this trim should fail */
if (rte_pktmbuf_trim(m, (uint16_t)(rte_pktmbuf_data_len(m) + 1)) == 0)
GOTO_FAIL("trim should not succeed");
/* prepend one header */
hdr = rte_pktmbuf_prepend(m, MBUF_TEST_HDR1_LEN);
if (hdr == NULL)
GOTO_FAIL("Cannot prepend");
if (data - hdr != MBUF_TEST_HDR1_LEN)
GOTO_FAIL("Prepend failed");
if (rte_pktmbuf_pkt_len(m) != MBUF_TEST_DATA_LEN + MBUF_TEST_HDR1_LEN)
GOTO_FAIL("Bad pkt length");
if (rte_pktmbuf_data_len(m) != MBUF_TEST_DATA_LEN + MBUF_TEST_HDR1_LEN)
GOTO_FAIL("Bad data length");
if (!rte_pktmbuf_is_contiguous(m))
GOTO_FAIL("Buffer should be continuous");
memset(hdr, 0x55, MBUF_TEST_HDR1_LEN);
/* prepend another header */
hdr = rte_pktmbuf_prepend(m, MBUF_TEST_HDR2_LEN);
if (hdr == NULL)
GOTO_FAIL("Cannot prepend");
if (data - hdr != MBUF_TEST_ALL_HDRS_LEN)
GOTO_FAIL("Prepend failed");
if (rte_pktmbuf_pkt_len(m) != MBUF_TEST_DATA_LEN + MBUF_TEST_ALL_HDRS_LEN)
GOTO_FAIL("Bad pkt length");
if (rte_pktmbuf_data_len(m) != MBUF_TEST_DATA_LEN + MBUF_TEST_ALL_HDRS_LEN)
GOTO_FAIL("Bad data length");
if (!rte_pktmbuf_is_contiguous(m))
GOTO_FAIL("Buffer should be continuous");
memset(hdr, 0x55, MBUF_TEST_HDR2_LEN);
rte_mbuf_sanity_check(m, RTE_MBUF_PKT, 1);
rte_mbuf_sanity_check(m, RTE_MBUF_PKT, 0);
rte_pktmbuf_dump(m, 0);
/* this prepend should fail */
hdr = rte_pktmbuf_prepend(m, (uint16_t)(rte_pktmbuf_headroom(m) + 1));
if (hdr != NULL)
GOTO_FAIL("prepend should not succeed");
/* remove data at beginning of mbuf (adj) */
if (data != rte_pktmbuf_adj(m, MBUF_TEST_ALL_HDRS_LEN))
GOTO_FAIL("rte_pktmbuf_adj failed");
if (rte_pktmbuf_pkt_len(m) != MBUF_TEST_DATA_LEN)
GOTO_FAIL("Bad pkt length");
if (rte_pktmbuf_data_len(m) != MBUF_TEST_DATA_LEN)
GOTO_FAIL("Bad data length");
if (!rte_pktmbuf_is_contiguous(m))
GOTO_FAIL("Buffer should be continuous");
/* this adj should fail */
if (rte_pktmbuf_adj(m, (uint16_t)(rte_pktmbuf_data_len(m) + 1)) != NULL)
GOTO_FAIL("rte_pktmbuf_adj should not succeed");
/* check data */
if (!rte_pktmbuf_is_contiguous(m))
GOTO_FAIL("Buffer should be continuous");
for (i=0; i<MBUF_TEST_DATA_LEN; i++) {
if (data[i] != 0x66)
GOTO_FAIL("Data corrupted at offset %u", i);
}
/* free mbuf */
rte_pktmbuf_free(m);
m = NULL;
return 0;
fail:
if (m)
rte_pktmbuf_free(m);
return -1;
}
dpdk_virtio_dev_to_vm_tx_burst(struct dpdk_virtio_writer *p,
vr_dpdk_virtioq_t *vq, struct rte_mbuf **pkts, uint32_t count)
{
struct vring_desc *desc;
struct rte_mbuf *buff;
/* The virtio_hdr is initialised to 0. */
struct virtio_net_hdr_mrg_rxbuf virtio_hdr = {{0, 0, 0, 0, 0, 0}, 0};
uint64_t buff_addr = 0;
uint64_t buff_hdr_addr = 0;
uint32_t head[VR_DPDK_VIRTIO_TX_BURST_SZ];
uint32_t head_idx, packet_success = 0;
uint16_t avail_idx, res_cur_idx;
uint16_t res_base_idx, res_end_idx;
uint16_t free_entries;
uint8_t success = 0;
vr_uvh_client_t *vru_cl;
if (unlikely(vq->vdv_ready_state == VQ_NOT_READY))
return 0;
vru_cl = vr_dpdk_virtio_get_vif_client(vq->vdv_vif_idx);
if (unlikely(vru_cl == NULL))
return 0;
/*
* As many data cores may want access to available buffers,
* they need to be reserved.
*/
do {
res_base_idx = vq->vdv_last_used_idx_res;
avail_idx = *((volatile uint16_t *)&vq->vdv_avail->idx);
free_entries = (avail_idx - res_base_idx);
/*check that we have enough buffers*/
if (unlikely(count > free_entries))
count = free_entries;
if (unlikely(count == 0))
return 0;
res_end_idx = res_base_idx + count;
/* vq->vdv_last_used_idx_res is atomically updated. */
/* TODO: Allow to disable cmpset if no concurrency in application. */
success = rte_atomic16_cmpset(&vq->vdv_last_used_idx_res,
res_base_idx, res_end_idx);
} while (unlikely(success == 0));
res_cur_idx = res_base_idx;
RTE_LOG(DEBUG, VROUTER, "%s: Current Index %d| End Index %d\n",
__func__, res_cur_idx, res_end_idx);
/* Prefetch available ring to retrieve indexes. */
rte_prefetch0(&vq->vdv_avail->ring[res_cur_idx & (vq->vdv_size - 1)]);
/* Retrieve all of the head indexes first to avoid caching issues. */
for (head_idx = 0; head_idx < count; head_idx++)
head[head_idx] = vq->vdv_avail->ring[(res_cur_idx + head_idx) &
(vq->vdv_size - 1)];
/* Prefetch descriptor index. */
rte_prefetch0(&vq->vdv_desc[head[packet_success]]);
while (res_cur_idx != res_end_idx) {
uint32_t offset = 0, vb_offset = 0;
uint32_t pkt_len, len_to_cpy, data_len, total_copied = 0;
uint8_t hdr = 0, uncompleted_pkt = 0;
/* Get descriptor from available ring */
desc = &vq->vdv_desc[head[packet_success]];
buff = pkts[packet_success];
/* Convert from gpa to vva (guest physical addr -> vhost virtual addr) */
buff_addr = (uintptr_t)vr_dpdk_guest_phys_to_host_virt(vru_cl, desc->addr);
/* Prefetch buffer address. */
rte_prefetch0((void *)(uintptr_t)buff_addr);
/* Copy virtio_hdr to packet and increment buffer address */
buff_hdr_addr = buff_addr;
/*
* If the descriptors are chained the header and data are
* placed in separate buffers.
*/
if (likely(desc->flags & VRING_DESC_F_NEXT)
&& (desc->len == sizeof(struct virtio_net_hdr))) {
/*
* TODO: verify that desc->next is sane below.
*/
desc = &vq->vdv_desc[desc->next];
/* Buffer address translation. */
buff_addr = (uintptr_t)vr_dpdk_guest_phys_to_host_virt(vru_cl, desc->addr);
} else {
vb_offset += sizeof(struct virtio_net_hdr);
hdr = 1;
}
pkt_len = rte_pktmbuf_pkt_len(buff);
data_len = rte_pktmbuf_data_len(buff);
len_to_cpy = RTE_MIN(data_len,
hdr ? desc->len - sizeof(struct virtio_net_hdr) : desc->len);
while (total_copied < pkt_len) {
/* Copy mbuf data to buffer */
rte_memcpy((void *)(uintptr_t)(buff_addr + vb_offset),
rte_pktmbuf_mtod_offset(buff, const void *, offset),
len_to_cpy);
offset += len_to_cpy;
vb_offset += len_to_cpy;
total_copied += len_to_cpy;
/* The whole packet completes */
if (likely(total_copied == pkt_len))
break;
/* The current segment completes */
if (offset == data_len) {
buff = buff->next;
offset = 0;
data_len = rte_pktmbuf_data_len(buff);
}
/* The current vring descriptor done */
if (vb_offset == desc->len) {
if (desc->flags & VRING_DESC_F_NEXT) {
desc = &vq->vdv_desc[desc->next];
buff_addr = (uintptr_t)vr_dpdk_guest_phys_to_host_virt(vru_cl, desc->addr);
vb_offset = 0;
} else {
/* Room in vring buffer is not enough */
uncompleted_pkt = 1;
break;
}
}
len_to_cpy = RTE_MIN(data_len - offset, desc->len - vb_offset);
};
/* Update used ring with desc information */
vq->vdv_used->ring[res_cur_idx & (vq->vdv_size - 1)].id =
head[packet_success];
/* Drop the packet if it is uncompleted */
if (unlikely(uncompleted_pkt == 1))
vq->vdv_used->ring[res_cur_idx & (vq->vdv_size - 1)].len =
sizeof(struct virtio_net_hdr);
else
vq->vdv_used->ring[res_cur_idx & (vq->vdv_size - 1)].len =
pkt_len + sizeof(struct virtio_net_hdr);
res_cur_idx++;
packet_success++;
/* TODO: in DPDK 2.1 we do not copy the header
if (unlikely(uncompleted_pkt == 1))
continue;
*/
rte_memcpy((void *)(uintptr_t)buff_hdr_addr,
(const void *)&virtio_hdr, sizeof(struct virtio_net_hdr));
if (likely(res_cur_idx < res_end_idx)) {
/* Prefetch descriptor index. */
rte_prefetch0(&vq->vdv_desc[head[packet_success]]);
}
}
rte_compiler_barrier();
/* Wait until it's our turn to add our buffer to the used ring. */
while (unlikely(vq->vdv_last_used_idx != res_base_idx))
rte_pause();
*(volatile uint16_t *)&vq->vdv_used->idx += count;
vq->vdv_last_used_idx = res_end_idx;
RTE_LOG(DEBUG, VROUTER, "%s: vif %d vq %p last_used_idx %d used->idx %d\n",
__func__, vq->vdv_vif_idx, vq, vq->vdv_last_used_idx, vq->vdv_used->idx);
/* flush used->idx update before we read avail->flags. */
rte_mb();
/* Kick the guest if necessary. */
if (unlikely(!(vq->vdv_avail->flags & VRING_AVAIL_F_NO_INTERRUPT))) {
p->nb_syscalls++;
eventfd_write(vq->vdv_callfd, 1);
}
return count;
}
int do_nf(void *useless)
{
(void) useless; //XXX: this line suppresses the "unused-parameter" error
int i;
unsigned int p;
mbuf_array_t pkts_received;
//Init the regex engine
if(!initializeRegEx(&re_bbc, re_extra_bbc,BBC))
return 0;
mbuf_array_t *pkts_to_send = (mbuf_array_t*)malloc(NUM_PORTS * sizeof(mbuf_array_t));
for(p = 0; p < NUM_PORTS; p++)
pkts_to_send[p].n_mbufs = 0;
while(1)
{
#ifdef ENABLE_SEMAPHORE
sem_wait(nf_params.semaphore);
#endif
/*0) Iterates on all the ports */
for(p = 0; p < NUM_PORTS; p++)
{
/*1) Receive incoming packets */
pkts_received.n_mbufs = rte_ring_sc_dequeue_burst(nf_params.ports[p].to_nf_queue,(void **)&pkts_received.array[0],PKT_TO_NF_THRESHOLD);
if(likely(pkts_received.n_mbufs > 0))
{
#ifdef ENABLE_LOG
fprintf(logFile,"[%s] Received %d pkts on port %d (%s)\n", NAME, pkts_received.n_mbufs,p,nf_params.ports[p].name);
#endif
for (i=0;i < pkts_received.n_mbufs;i++)
{
/*2) Operate on the packet */
unsigned char *pkt = rte_pktmbuf_mtod(pkts_received.array[i],unsigned char *);
#ifdef ENABLE_LOG
fprintf(logFile,"[%s] Packet size: %d\n",NAME,rte_pktmbuf_pkt_len(pkts_received.array[i]));
fprintf(logFile,"[%s] %.2x:%.2x:%.2x:%.2x:%.2x:%.2x -> %.2x:%.2x:%.2x:%.2x:%.2x:%.2x\n",NAME,pkt[6],pkt[7],pkt[8],pkt[9],pkt[10],pkt[11],pkt[0],pkt[1],pkt[2],pkt[3],pkt[4],pkt[5]);
#endif
/**
* If the packet arrives from the first port, check if it must be dropped
*/
if(p == 0)
{
#ifdef ENABLE_LOG
fprintf(logFile,"[%s] I'm going to check if the packet must be dropped.\n", NAME);
#endif
if(drop(pkt,rte_pktmbuf_pkt_len(pkts_received.array[i])))
{
//The packet must be dropped
#ifdef ENABLE_LOG
fprintf(logFile,"[%s] The packet is dropped.\n", NAME);
#endif
rte_pktmbuf_free(pkts_received.array[i]);
continue;
}
}
unsigned int output_port = (p+1) % NUM_PORTS;
pkts_to_send[output_port].array[pkts_to_send[output_port].n_mbufs] = pkts_received.array[i];
pkts_to_send[output_port].n_mbufs++;
}//end of iteration on the packets received from the current port
} //end if(likely(pkts_received.n_mbufs > 0))
}//end iteration on the ports
/*3) Send the processed packet not transmitted yet*/
for(p = 0; p < NUM_PORTS; p++)
{
if(likely(pkts_to_send[p].n_mbufs > 0))
{
#ifdef ENABLE_LOG
fprintf(logFile,"[%s] Sending %d packets on port %x (%s).\n", NAME,pkts_to_send[p].n_mbufs,p,nf_params.ports[p].name);
#endif
int ret = rte_ring_sp_enqueue_burst(nf_params.ports[p].to_xdpd_queue,(void *const*)pkts_to_send[p].array,(unsigned)pkts_to_send[p].n_mbufs);
if (unlikely(ret < pkts_to_send[p].n_mbufs))
{
fprintf(logFile,"[%s] Not enough room in port %d towards xDPD to enqueue; the packet will be dropped.\n", NAME,p);
do {
struct rte_mbuf *pkt_to_free = pkts_to_send[p].array[ret];
rte_pktmbuf_free(pkt_to_free);
} while (++ret < pkts_to_send[p].n_mbufs);
}
}
pkts_to_send[p].n_mbufs = 0;
}/* End of iteration on the ports */
}/*End of while true*/
void app_main_loop_rx_flow(void)
{
const unsigned lcore_id = rte_lcore_id();
struct rte_mbuf *bufs[RX_BURST_SIZE];
struct rte_mbuf *buf;
struct ether_hdr *eth_hdr;
struct ipv4_hdr *ipv4_hdr;
struct ipv6_hdr *ipv6_hdr;
struct tcp_hdr *tcp_hdr;
struct udp_hdr *udp_hdr;
struct pkt_info pktinfo;
int32_t ret;
uint16_t i, n_rx, queueid;
uint8_t port;
port = 0;
queueid = (uint16_t) app.lcore_conf[lcore_id].queue_id;
RTE_LOG(INFO, FLOWATCHER, "[core %u] packet RX & update flow_table Ready\n", lcore_id);
while (!app_quit_signal) {
n_rx = rte_eth_rx_burst(port, queueid, bufs, RX_BURST_SIZE);
if (unlikely(n_rx == 0)) {
port++;
if (port >= app.n_ports)
port = 0;
continue;
}
app_stat[queueid].rx_count += n_rx;
for (i = 0; i < n_rx; i++) {
buf = bufs[i];
pktinfo.timestamp = rte_rdtsc();
pktinfo.pktlen = rte_pktmbuf_pkt_len(buf);
eth_hdr = rte_pktmbuf_mtod(buf, struct ether_hdr *);
/* strip vlan_hdr */
if (eth_hdr->ether_type == rte_cpu_to_be_16(ETHER_TYPE_VLAN)) {
/* struct vlan_hdr *vh = (struct vlan_hdr *) ð_hdr[1]; */
/* buf->ol_flags |= PKT_RX_VLAN_PKT; */
/* buf->vlan_tci = rte_be_to_cpu_16(vh->vlan_tci); */
/* memmove(rte_pktmbuf_adj(buf, sizeof(struct vlan_hdr)), */
/* eth_hdr, 2 * ETHER_ADDR_LEN); */
/* eth_hdr = rte_pktmbuf_mtod(buf, struct ether_hdr *); */
eth_hdr = (struct ether_hdr *) rte_pktmbuf_adj(buf, sizeof(struct vlan_hdr));
}
if (eth_hdr->ether_type == rte_cpu_to_be_16(ETHER_TYPE_IPv4)) {
/* IPv4 */
pktinfo.type = PKT_IP_TYPE_IPV4;
ipv4_hdr = (struct ipv4_hdr *) ð_hdr[1];
pktinfo.key.v4.src_ip = rte_be_to_cpu_32(ipv4_hdr->src_addr);
pktinfo.key.v4.dst_ip = rte_be_to_cpu_32(ipv4_hdr->dst_addr);
pktinfo.key.v4.proto = ipv4_hdr->next_proto_id;
switch (ipv4_hdr->next_proto_id) {
case IPPROTO_TCP:
tcp_hdr = (struct tcp_hdr *) &ipv4_hdr[1];
pktinfo.key.v4.src_port = rte_be_to_cpu_16(tcp_hdr->src_port);
pktinfo.key.v4.dst_port = rte_be_to_cpu_16(tcp_hdr->dst_port);
break;
case IPPROTO_UDP:
udp_hdr = (struct udp_hdr *) &ipv4_hdr[1];
pktinfo.key.v4.src_port = rte_be_to_cpu_16(udp_hdr->src_port);
pktinfo.key.v4.dst_port = rte_be_to_cpu_16(udp_hdr->dst_port);
break;
default:
pktinfo.key.v4.src_port = 0;
pktinfo.key.v4.dst_port = 0;
break;
}
rte_pktmbuf_free(buf);
/* update flow_table_v4 */
ret = update_flow_entry(app.flow_table_v4[queueid], &pktinfo);
if (ret == 0)
app_stat[queueid].updated_tbl_v4_count++;
else
app_stat[queueid].miss_updated_tbl_v4_count++;
} else if (eth_hdr->ether_type == rte_cpu_to_be_16(ETHER_TYPE_IPv6)) {
/* IPv6 */
pktinfo.type = PKT_IP_TYPE_IPV6;
ipv6_hdr = (struct ipv6_hdr *) ð_hdr[1];
rte_memcpy(pktinfo.key.v6.src_ip, ipv6_hdr->src_addr, 16);
rte_memcpy(pktinfo.key.v6.dst_ip, ipv6_hdr->dst_addr, 16);
pktinfo.key.v6.proto = ipv6_hdr->proto;
switch (ipv6_hdr->proto) {
case IPPROTO_TCP:
tcp_hdr = (struct tcp_hdr *) &ipv6_hdr[1];
pktinfo.key.v6.src_port = rte_be_to_cpu_16(tcp_hdr->src_port);
pktinfo.key.v6.dst_port = rte_be_to_cpu_16(tcp_hdr->dst_port);
break;
case IPPROTO_UDP:
udp_hdr = (struct udp_hdr *) &ipv6_hdr[1];
pktinfo.key.v6.src_port = rte_be_to_cpu_16(udp_hdr->src_port);
pktinfo.key.v6.dst_port = rte_be_to_cpu_16(udp_hdr->dst_port);
break;
default:
pktinfo.key.v6.src_port = 0;
pktinfo.key.v6.dst_port = 0;
break;
}
rte_pktmbuf_free(buf);
/* update flow_table_v6 */
ret = update_flow_entry(app.flow_table_v6[queueid], &pktinfo);
if (ret == 0)
app_stat[queueid].updated_tbl_v6_count++;
else
app_stat[queueid].miss_updated_tbl_v6_count++;
} else {
/* others */
app_stat[queueid].unknown_pkt_count++;
rte_pktmbuf_free(buf);
continue;
}
}
port++;
if (port >= app.n_ports)
port = 0;
}
RTE_LOG(INFO, FLOWATCHER, "[core %u] packet RX & update flow_table finished\n", lcore_id);
}
