virtio_dev_rx(struct virtio_net *dev, 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], packet_len = 0;
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);
vq = dev->virtqueue_rx;
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. */
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) {
/* Get descriptor from available ring */
desc = &vq->desc[head[packet_success]];
/* Prefetch descriptor address. */
rte_prefetch0(desc);
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;
packet_len = rte_pktmbuf_data_len(buff) + vq->vhost_hlen;
/*
* 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);
desc->len = rte_pktmbuf_data_len(buff);
} else {
buff_addr += vq->vhost_hlen;
desc->len = packet_len;
}
}
/* Update used ring with desc information */
vq->used->ring[res_cur_idx & (vq->size - 1)].id = head[packet_success];
vq->used->ring[res_cur_idx & (vq->size - 1)].len = packet_len;
/* Copy mbuf data to buffer */
rte_memcpy((void *)(uintptr_t)buff_addr, (const void*)buff->pkt.data, rte_pktmbuf_data_len(buff));
res_cur_idx++;
packet_success++;
/* mergeable is disabled then a header is required per buffer. */
rte_memcpy((void *)(uintptr_t)buff_hdr_addr, (const void*)&virtio_hdr, vq->vhost_hlen);
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;
return count;
}
virtio_dev_merge_rx(struct virtio_net *dev, uint16_t queue_id,
struct rte_mbuf **pkts, uint32_t count)
{
struct vhost_virtqueue *vq;
uint32_t pkt_idx = 0, entry_success = 0;
uint16_t avail_idx;
uint16_t res_base_idx, res_cur_idx;
uint8_t success = 0;
LOG_DEBUG(VHOST_DATA, "(%"PRIu64") virtio_dev_merge_rx()\n",
dev->device_fh);
if (unlikely(queue_id != VIRTIO_RXQ)) {
LOG_DEBUG(VHOST_DATA, "mq isn't supported in this version.\n");
}
vq = dev->virtqueue[VIRTIO_RXQ];
count = RTE_MIN((uint32_t)MAX_PKT_BURST, count);
if (count == 0)
return 0;
for (pkt_idx = 0; pkt_idx < count; pkt_idx++) {
uint32_t pkt_len = pkts[pkt_idx]->pkt_len + vq->vhost_hlen;
do {
/*
* As many data cores may want access to available
* buffers, they need to be reserved.
*/
uint32_t secure_len = 0;
uint32_t vec_idx = 0;
res_base_idx = vq->last_used_idx_res;
res_cur_idx = res_base_idx;
do {
avail_idx = *((volatile uint16_t *)&vq->avail->idx);
if (unlikely(res_cur_idx == avail_idx)) {
LOG_DEBUG(VHOST_DATA,
"(%"PRIu64") Failed "
"to get enough desc from "
"vring\n",
dev->device_fh);
return pkt_idx;
} else {
update_secure_len(vq, res_cur_idx, &secure_len, &vec_idx);
res_cur_idx++;
}
} while (pkt_len > secure_len);
/* vq->last_used_idx_res is atomically updated. */
success = rte_atomic16_cmpset(&vq->last_used_idx_res,
res_base_idx,
res_cur_idx);
} while (success == 0);
entry_success = copy_from_mbuf_to_vring(dev, res_base_idx,
res_cur_idx, pkts[pkt_idx]);
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 += entry_success;
vq->last_used_idx = res_cur_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;
}
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;
}
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;
}