X-Git-Url: http://git.droids-corp.org/?a=blobdiff_plain;ds=sidebyside;f=lib%2Flibrte_vhost%2Fvirtio_net.c;h=5b85b832d0d6dc6ed013f7e7f4514f8cf716191d;hb=4142b06e2d60734eb7f6dda5edc212993c461f0a;hp=22c9f15666d7e52920e59511ec09aa9312d7641a;hpb=62250c1d09780e3d75ae252d9f69ca862e5bfca7;p=dpdk.git diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c index 22c9f15666..5b85b832d0 100644 --- a/lib/librte_vhost/virtio_net.c +++ b/lib/librte_vhost/virtio_net.c @@ -37,51 +37,13 @@ is_valid_virt_queue_idx(uint32_t idx, int is_tx, uint32_t nr_vring) return (is_tx ^ (idx & 1)) == 0 && idx < nr_vring; } -static __rte_always_inline void * -alloc_copy_ind_table(struct virtio_net *dev, struct vhost_virtqueue *vq, - uint64_t desc_addr, uint64_t desc_len) -{ - void *idesc; - uint64_t src, dst; - uint64_t len, remain = desc_len; - - idesc = rte_malloc(__func__, desc_len, 0); - if (unlikely(!idesc)) - return 0; - - dst = (uint64_t)(uintptr_t)idesc; - - while (remain) { - len = remain; - src = vhost_iova_to_vva(dev, vq, desc_addr, &len, - VHOST_ACCESS_RO); - if (unlikely(!src || !len)) { - rte_free(idesc); - return 0; - } - - rte_memcpy((void *)(uintptr_t)dst, (void *)(uintptr_t)src, len); - - remain -= len; - dst += len; - desc_addr += len; - } - - return idesc; -} - -static __rte_always_inline void -free_ind_table(void *idesc) -{ - rte_free(idesc); -} - static __rte_always_inline void -do_flush_shadow_used_ring(struct virtio_net *dev, struct vhost_virtqueue *vq, - uint16_t to, uint16_t from, uint16_t size) +do_flush_shadow_used_ring_split(struct virtio_net *dev, + struct vhost_virtqueue *vq, + uint16_t to, uint16_t from, uint16_t size) { rte_memcpy(&vq->used->ring[to], - &vq->shadow_used_ring[from], + &vq->shadow_used_split[from], size * sizeof(struct vring_used_elem)); vhost_log_cache_used_vring(dev, vq, offsetof(struct vring_used, ring[to]), @@ -89,22 +51,22 @@ do_flush_shadow_used_ring(struct virtio_net *dev, struct vhost_virtqueue *vq, } static __rte_always_inline void -flush_shadow_used_ring(struct virtio_net *dev, struct vhost_virtqueue *vq) +flush_shadow_used_ring_split(struct virtio_net *dev, struct vhost_virtqueue *vq) { uint16_t used_idx = vq->last_used_idx & (vq->size - 1); if (used_idx + vq->shadow_used_idx <= vq->size) { - do_flush_shadow_used_ring(dev, vq, used_idx, 0, + do_flush_shadow_used_ring_split(dev, vq, used_idx, 0, vq->shadow_used_idx); } else { uint16_t size; /* update used ring interval [used_idx, vq->size] */ size = vq->size - used_idx; - do_flush_shadow_used_ring(dev, vq, used_idx, 0, size); + do_flush_shadow_used_ring_split(dev, vq, used_idx, 0, size); /* update the left half used ring interval [0, left_size] */ - do_flush_shadow_used_ring(dev, vq, 0, size, + do_flush_shadow_used_ring_split(dev, vq, 0, size, vq->shadow_used_idx - size); } vq->last_used_idx += vq->shadow_used_idx; @@ -120,13 +82,91 @@ flush_shadow_used_ring(struct virtio_net *dev, struct vhost_virtqueue *vq) } static __rte_always_inline void -update_shadow_used_ring(struct vhost_virtqueue *vq, - uint16_t desc_idx, uint16_t len) +update_shadow_used_ring_split(struct vhost_virtqueue *vq, + uint16_t desc_idx, uint32_t len) { uint16_t i = vq->shadow_used_idx++; - vq->shadow_used_ring[i].id = desc_idx; - vq->shadow_used_ring[i].len = len; + vq->shadow_used_split[i].id = desc_idx; + vq->shadow_used_split[i].len = len; +} + +static __rte_always_inline void +flush_shadow_used_ring_packed(struct virtio_net *dev, + struct vhost_virtqueue *vq) +{ + int i; + uint16_t used_idx = vq->last_used_idx; + uint16_t head_idx = vq->last_used_idx; + uint16_t head_flags = 0; + + /* Split loop in two to save memory barriers */ + for (i = 0; i < vq->shadow_used_idx; i++) { + vq->desc_packed[used_idx].id = vq->shadow_used_packed[i].id; + vq->desc_packed[used_idx].len = vq->shadow_used_packed[i].len; + + used_idx += vq->shadow_used_packed[i].count; + if (used_idx >= vq->size) + used_idx -= vq->size; + } + + rte_smp_wmb(); + + for (i = 0; i < vq->shadow_used_idx; i++) { + uint16_t flags; + + if (vq->shadow_used_packed[i].len) + flags = VRING_DESC_F_WRITE; + else + flags = 0; + + if (vq->used_wrap_counter) { + flags |= VRING_DESC_F_USED; + flags |= VRING_DESC_F_AVAIL; + } else { + flags &= ~VRING_DESC_F_USED; + flags &= ~VRING_DESC_F_AVAIL; + } + + if (i > 0) { + vq->desc_packed[vq->last_used_idx].flags = flags; + + vhost_log_cache_used_vring(dev, vq, + vq->last_used_idx * + sizeof(struct vring_packed_desc), + sizeof(struct vring_packed_desc)); + } else { + head_idx = vq->last_used_idx; + head_flags = flags; + } + + vq->last_used_idx += vq->shadow_used_packed[i].count; + if (vq->last_used_idx >= vq->size) { + vq->used_wrap_counter ^= 1; + vq->last_used_idx -= vq->size; + } + } + + vq->desc_packed[head_idx].flags = head_flags; + + vhost_log_cache_used_vring(dev, vq, + head_idx * + sizeof(struct vring_packed_desc), + sizeof(struct vring_packed_desc)); + + vq->shadow_used_idx = 0; + vhost_log_cache_sync(dev, vq); +} + +static __rte_always_inline void +update_shadow_used_ring_packed(struct vhost_virtqueue *vq, + uint16_t desc_idx, uint32_t len, uint16_t count) +{ + uint16_t i = vq->shadow_used_idx++; + + vq->shadow_used_packed[i].id = desc_idx; + vq->shadow_used_packed[i].len = len; + vq->shadow_used_packed[i].count = count; } static inline void @@ -178,15 +218,15 @@ virtio_enqueue_offload(struct rte_mbuf *m_buf, struct virtio_net_hdr *net_hdr) switch (csum_l4) { case PKT_TX_TCP_CKSUM: - net_hdr->csum_offset = (offsetof(struct tcp_hdr, + net_hdr->csum_offset = (offsetof(struct rte_tcp_hdr, cksum)); break; case PKT_TX_UDP_CKSUM: - net_hdr->csum_offset = (offsetof(struct udp_hdr, + net_hdr->csum_offset = (offsetof(struct rte_udp_hdr, dgram_cksum)); break; case PKT_TX_SCTP_CKSUM: - net_hdr->csum_offset = (offsetof(struct sctp_hdr, + net_hdr->csum_offset = (offsetof(struct rte_sctp_hdr, cksum)); break; } @@ -198,9 +238,9 @@ virtio_enqueue_offload(struct rte_mbuf *m_buf, struct virtio_net_hdr *net_hdr) /* IP cksum verification cannot be bypassed, then calculate here */ if (m_buf->ol_flags & PKT_TX_IP_CKSUM) { - struct ipv4_hdr *ipv4_hdr; + struct rte_ipv4_hdr *ipv4_hdr; - ipv4_hdr = rte_pktmbuf_mtod_offset(m_buf, struct ipv4_hdr *, + ipv4_hdr = rte_pktmbuf_mtod_offset(m_buf, struct rte_ipv4_hdr *, m_buf->l2_len); ipv4_hdr->hdr_checksum = rte_ipv4_cksum(ipv4_hdr); } @@ -225,23 +265,68 @@ virtio_enqueue_offload(struct rte_mbuf *m_buf, struct virtio_net_hdr *net_hdr) } } +static __rte_always_inline int +map_one_desc(struct virtio_net *dev, struct vhost_virtqueue *vq, + struct buf_vector *buf_vec, uint16_t *vec_idx, + uint64_t desc_iova, uint64_t desc_len, uint8_t perm) +{ + uint16_t vec_id = *vec_idx; + + while (desc_len) { + uint64_t desc_addr; + uint64_t desc_chunck_len = desc_len; + + if (unlikely(vec_id >= BUF_VECTOR_MAX)) + return -1; + + desc_addr = vhost_iova_to_vva(dev, vq, + desc_iova, + &desc_chunck_len, + perm); + if (unlikely(!desc_addr)) + return -1; + + rte_prefetch0((void *)(uintptr_t)desc_addr); + + buf_vec[vec_id].buf_iova = desc_iova; + buf_vec[vec_id].buf_addr = desc_addr; + buf_vec[vec_id].buf_len = desc_chunck_len; + + desc_len -= desc_chunck_len; + desc_iova += desc_chunck_len; + vec_id++; + } + *vec_idx = vec_id; + + return 0; +} + static __rte_always_inline int fill_vec_buf_split(struct virtio_net *dev, struct vhost_virtqueue *vq, uint32_t avail_idx, uint16_t *vec_idx, struct buf_vector *buf_vec, uint16_t *desc_chain_head, - uint16_t *desc_chain_len, uint8_t perm) + uint32_t *desc_chain_len, uint8_t perm) { uint16_t idx = vq->avail->ring[avail_idx & (vq->size - 1)]; uint16_t vec_id = *vec_idx; uint32_t len = 0; - uint64_t dlen, desc_avail, desc_iova; + uint64_t dlen; + uint32_t nr_descs = vq->size; + uint32_t cnt = 0; struct vring_desc *descs = vq->desc; struct vring_desc *idesc = NULL; + if (unlikely(idx >= vq->size)) + return -1; + *desc_chain_head = idx; if (vq->desc[idx].flags & VRING_DESC_F_INDIRECT) { dlen = vq->desc[idx].len; + nr_descs = dlen / sizeof(struct vring_desc); + if (unlikely(nr_descs > vq->size)) + return -1; + descs = (struct vring_desc *)(uintptr_t) vhost_iova_to_vva(dev, vq, vq->desc[idx].addr, &dlen, @@ -254,7 +339,7 @@ fill_vec_buf_split(struct virtio_net *dev, struct vhost_virtqueue *vq, * The indirect desc table is not contiguous * in process VA space, we have to copy it. */ - idesc = alloc_copy_ind_table(dev, vq, + idesc = vhost_alloc_copy_ind_table(dev, vq, vq->desc[idx].addr, vq->desc[idx].len); if (unlikely(!idesc)) return -1; @@ -266,42 +351,18 @@ fill_vec_buf_split(struct virtio_net *dev, struct vhost_virtqueue *vq, } while (1) { - if (unlikely(idx >= vq->size)) { + if (unlikely(idx >= nr_descs || cnt++ >= nr_descs)) { free_ind_table(idesc); return -1; } - len += descs[idx].len; - desc_avail = descs[idx].len; - desc_iova = descs[idx].addr; - - while (desc_avail) { - uint64_t desc_addr; - uint64_t desc_chunck_len = desc_avail; - - if (unlikely(vec_id >= BUF_VECTOR_MAX)) { - free_ind_table(idesc); - return -1; - } - - desc_addr = vhost_iova_to_vva(dev, vq, - desc_iova, - &desc_chunck_len, - perm); - if (unlikely(!desc_addr)) { - free_ind_table(idesc); - return -1; - } - - buf_vec[vec_id].buf_iova = desc_iova; - buf_vec[vec_id].buf_addr = desc_addr; - buf_vec[vec_id].buf_len = desc_chunck_len; - buf_vec[vec_id].desc_idx = idx; - desc_avail -= desc_chunck_len; - desc_iova += desc_chunck_len; - vec_id++; + if (unlikely(map_one_desc(dev, vq, buf_vec, &vec_id, + descs[idx].addr, descs[idx].len, + perm))) { + free_ind_table(idesc); + return -1; } if ((descs[idx].flags & VRING_DESC_F_NEXT) == 0) @@ -333,19 +394,26 @@ reserve_avail_buf_split(struct virtio_net *dev, struct vhost_virtqueue *vq, uint16_t max_tries, tries = 0; uint16_t head_idx = 0; - uint16_t len = 0; + uint32_t len = 0; *num_buffers = 0; cur_idx = vq->last_avail_idx; if (rxvq_is_mergeable(dev)) - max_tries = vq->size; + max_tries = vq->size - 1; else max_tries = 1; while (size > 0) { if (unlikely(cur_idx == avail_head)) return -1; + /* + * if we tried all available ring items, and still + * can't get enough buf, it means something abnormal + * happened. + */ + if (unlikely(++tries > max_tries)) + return -1; if (unlikely(fill_vec_buf_split(dev, vq, cur_idx, &vec_idx, buf_vec, @@ -353,20 +421,191 @@ reserve_avail_buf_split(struct virtio_net *dev, struct vhost_virtqueue *vq, VHOST_ACCESS_RW) < 0)) return -1; len = RTE_MIN(len, size); - update_shadow_used_ring(vq, head_idx, len); + update_shadow_used_ring_split(vq, head_idx, len); size -= len; cur_idx++; - tries++; *num_buffers += 1; + } + *nr_vec = vec_idx; + + return 0; +} + +static __rte_always_inline int +fill_vec_buf_packed_indirect(struct virtio_net *dev, + struct vhost_virtqueue *vq, + struct vring_packed_desc *desc, uint16_t *vec_idx, + struct buf_vector *buf_vec, uint32_t *len, uint8_t perm) +{ + uint16_t i; + uint32_t nr_descs; + uint16_t vec_id = *vec_idx; + uint64_t dlen; + struct vring_packed_desc *descs, *idescs = NULL; + + dlen = desc->len; + descs = (struct vring_packed_desc *)(uintptr_t) + vhost_iova_to_vva(dev, vq, desc->addr, &dlen, VHOST_ACCESS_RO); + if (unlikely(!descs)) + return -1; + + if (unlikely(dlen < desc->len)) { + /* + * The indirect desc table is not contiguous + * in process VA space, we have to copy it. + */ + idescs = vhost_alloc_copy_ind_table(dev, + vq, desc->addr, desc->len); + if (unlikely(!idescs)) + return -1; + + descs = idescs; + } + + nr_descs = desc->len / sizeof(struct vring_packed_desc); + if (unlikely(nr_descs >= vq->size)) { + free_ind_table(idescs); + return -1; + } + + for (i = 0; i < nr_descs; i++) { + if (unlikely(vec_id >= BUF_VECTOR_MAX)) { + free_ind_table(idescs); + return -1; + } + + *len += descs[i].len; + if (unlikely(map_one_desc(dev, vq, buf_vec, &vec_id, + descs[i].addr, descs[i].len, + perm))) + return -1; + } + *vec_idx = vec_id; + + if (unlikely(!!idescs)) + free_ind_table(idescs); + + return 0; +} + +static __rte_always_inline int +fill_vec_buf_packed(struct virtio_net *dev, struct vhost_virtqueue *vq, + uint16_t avail_idx, uint16_t *desc_count, + struct buf_vector *buf_vec, uint16_t *vec_idx, + uint16_t *buf_id, uint32_t *len, uint8_t perm) +{ + bool wrap_counter = vq->avail_wrap_counter; + struct vring_packed_desc *descs = vq->desc_packed; + uint16_t vec_id = *vec_idx; + + if (avail_idx < vq->last_avail_idx) + wrap_counter ^= 1; + + if (unlikely(!desc_is_avail(&descs[avail_idx], wrap_counter))) + return -1; + + /* + * The ordering between desc flags and desc + * content reads need to be enforced. + */ + rte_smp_rmb(); + + *desc_count = 0; + *len = 0; + + while (1) { + if (unlikely(vec_id >= BUF_VECTOR_MAX)) + return -1; + + if (unlikely(*desc_count >= vq->size)) + return -1; + + *desc_count += 1; + *buf_id = descs[avail_idx].id; + + if (descs[avail_idx].flags & VRING_DESC_F_INDIRECT) { + if (unlikely(fill_vec_buf_packed_indirect(dev, vq, + &descs[avail_idx], + &vec_id, buf_vec, + len, perm) < 0)) + return -1; + } else { + *len += descs[avail_idx].len; + + if (unlikely(map_one_desc(dev, vq, buf_vec, &vec_id, + descs[avail_idx].addr, + descs[avail_idx].len, + perm))) + return -1; + } + + if ((descs[avail_idx].flags & VRING_DESC_F_NEXT) == 0) + break; + + if (++avail_idx >= vq->size) { + avail_idx -= vq->size; + wrap_counter ^= 1; + } + } + + *vec_idx = vec_id; + + return 0; +} + +/* + * Returns -1 on fail, 0 on success + */ +static inline int +reserve_avail_buf_packed(struct virtio_net *dev, struct vhost_virtqueue *vq, + uint32_t size, struct buf_vector *buf_vec, + uint16_t *nr_vec, uint16_t *num_buffers, + uint16_t *nr_descs) +{ + uint16_t avail_idx; + uint16_t vec_idx = 0; + uint16_t max_tries, tries = 0; + + uint16_t buf_id = 0; + uint32_t len = 0; + uint16_t desc_count; + + *num_buffers = 0; + avail_idx = vq->last_avail_idx; + + if (rxvq_is_mergeable(dev)) + max_tries = vq->size - 1; + else + max_tries = 1; + + while (size > 0) { /* * if we tried all available ring items, and still * can't get enough buf, it means something abnormal * happened. */ - if (unlikely(tries > max_tries)) + if (unlikely(++tries > max_tries)) + return -1; + + if (unlikely(fill_vec_buf_packed(dev, vq, + avail_idx, &desc_count, + buf_vec, &vec_idx, + &buf_id, &len, + VHOST_ACCESS_RW) < 0)) return -1; + + len = RTE_MIN(len, size); + update_shadow_used_ring_packed(vq, buf_id, len, desc_count); + size -= len; + + avail_idx += desc_count; + if (avail_idx >= vq->size) + avail_idx -= vq->size; + + *nr_descs += desc_count; + *num_buffers += 1; } *nr_vec = vec_idx; @@ -374,6 +613,36 @@ reserve_avail_buf_split(struct virtio_net *dev, struct vhost_virtqueue *vq, return 0; } +static __rte_noinline void +copy_vnet_hdr_to_desc(struct virtio_net *dev, struct vhost_virtqueue *vq, + struct buf_vector *buf_vec, + struct virtio_net_hdr_mrg_rxbuf *hdr) +{ + uint64_t len; + uint64_t remain = dev->vhost_hlen; + uint64_t src = (uint64_t)(uintptr_t)hdr, dst; + uint64_t iova = buf_vec->buf_iova; + + while (remain) { + len = RTE_MIN(remain, + buf_vec->buf_len); + dst = buf_vec->buf_addr; + rte_memcpy((void *)(uintptr_t)dst, + (void *)(uintptr_t)src, + len); + + PRINT_PACKET(dev, (uintptr_t)dst, + (uint32_t)len, 0); + vhost_log_cache_write(dev, vq, + iova, len); + + remain -= len; + iova += len; + src += len; + buf_vec++; + } +} + static __rte_always_inline int copy_mbuf_to_desc(struct virtio_net *dev, struct vhost_virtqueue *vq, struct rte_mbuf *m, struct buf_vector *buf_vec, @@ -399,9 +668,6 @@ copy_mbuf_to_desc(struct virtio_net *dev, struct vhost_virtqueue *vq, buf_iova = buf_vec[vec_idx].buf_iova; buf_len = buf_vec[vec_idx].buf_len; - if (nr_vec > 1) - rte_prefetch0((void *)(uintptr_t)buf_vec[1].buf_addr); - if (unlikely(buf_len < dev->vhost_hlen && nr_vec <= 1)) { error = -1; goto out; @@ -444,10 +710,6 @@ copy_mbuf_to_desc(struct virtio_net *dev, struct vhost_virtqueue *vq, buf_iova = buf_vec[vec_idx].buf_iova; buf_len = buf_vec[vec_idx].buf_len; - /* Prefetch next buffer address. */ - if (vec_idx + 1 < nr_vec) - rte_prefetch0((void *)(uintptr_t) - buf_vec[vec_idx + 1].buf_addr); buf_offset = 0; buf_avail = buf_len; } @@ -467,29 +729,7 @@ copy_mbuf_to_desc(struct virtio_net *dev, struct vhost_virtqueue *vq, num_buffers); if (unlikely(hdr == &tmp_hdr)) { - uint64_t len; - uint64_t remain = dev->vhost_hlen; - uint64_t src = (uint64_t)(uintptr_t)hdr, dst; - uint64_t iova = buf_vec[0].buf_iova; - uint16_t hdr_vec_idx = 0; - - while (remain) { - len = remain; - dst = buf_vec[hdr_vec_idx].buf_addr; - rte_memcpy((void *)(uintptr_t)dst, - (void *)(uintptr_t)src, - len); - - PRINT_PACKET(dev, (uintptr_t)dst, - (uint32_t)len, 0); - vhost_log_cache_write(dev, vq, - iova, len); - - remain -= len; - iova += len; - src += len; - hdr_vec_idx++; - } + copy_vnet_hdr_to_desc(dev, vq, buf_vec, hdr); } else { PRINT_PACKET(dev, (uintptr_t)hdr_addr, dev->vhost_hlen, 0); @@ -501,7 +741,7 @@ copy_mbuf_to_desc(struct virtio_net *dev, struct vhost_virtqueue *vq, hdr_addr = 0; } - cpy_len = RTE_MIN(buf_len, mbuf_avail); + cpy_len = RTE_MIN(buf_avail, mbuf_avail); if (likely(cpy_len > MAX_BATCH_LEN || vq->batch_copy_nb_elems >= vq->size)) { @@ -534,7 +774,7 @@ out: return error; } -static __rte_always_inline uint32_t +static __rte_noinline uint32_t virtio_dev_rx_split(struct virtio_net *dev, struct vhost_virtqueue *vq, struct rte_mbuf **pkts, uint32_t count) { @@ -543,9 +783,16 @@ virtio_dev_rx_split(struct virtio_net *dev, struct vhost_virtqueue *vq, struct buf_vector buf_vec[BUF_VECTOR_MAX]; uint16_t avail_head; - rte_prefetch0(&vq->avail->ring[vq->last_avail_idx & (vq->size - 1)]); avail_head = *((volatile uint16_t *)&vq->avail->idx); + /* + * The ordering between avail index and + * desc reads needs to be enforced. + */ + rte_smp_rmb(); + + rte_prefetch0(&vq->avail->ring[vq->last_avail_idx & (vq->size - 1)]); + for (pkt_idx = 0; pkt_idx < count; pkt_idx++) { uint32_t pkt_len = pkts[pkt_idx]->pkt_len + dev->vhost_hlen; uint16_t nr_vec = 0; @@ -560,8 +807,6 @@ virtio_dev_rx_split(struct virtio_net *dev, struct vhost_virtqueue *vq, break; } - rte_prefetch0((void *)(uintptr_t)buf_vec[0].buf_addr); - VHOST_LOG_DEBUG(VHOST_DATA, "(%d) current index %d | end index %d\n", dev->vid, vq->last_avail_idx, vq->last_avail_idx + num_buffers); @@ -579,8 +824,59 @@ virtio_dev_rx_split(struct virtio_net *dev, struct vhost_virtqueue *vq, do_data_copy_enqueue(dev, vq); if (likely(vq->shadow_used_idx)) { - flush_shadow_used_ring(dev, vq); - vhost_vring_call(dev, vq); + flush_shadow_used_ring_split(dev, vq); + vhost_vring_call_split(dev, vq); + } + + return pkt_idx; +} + +static __rte_noinline uint32_t +virtio_dev_rx_packed(struct virtio_net *dev, struct vhost_virtqueue *vq, + struct rte_mbuf **pkts, uint32_t count) +{ + uint32_t pkt_idx = 0; + uint16_t num_buffers; + struct buf_vector buf_vec[BUF_VECTOR_MAX]; + + for (pkt_idx = 0; pkt_idx < count; pkt_idx++) { + uint32_t pkt_len = pkts[pkt_idx]->pkt_len + dev->vhost_hlen; + uint16_t nr_vec = 0; + uint16_t nr_descs = 0; + + if (unlikely(reserve_avail_buf_packed(dev, vq, + pkt_len, buf_vec, &nr_vec, + &num_buffers, &nr_descs) < 0)) { + VHOST_LOG_DEBUG(VHOST_DATA, + "(%d) failed to get enough desc from vring\n", + dev->vid); + vq->shadow_used_idx -= num_buffers; + break; + } + + VHOST_LOG_DEBUG(VHOST_DATA, "(%d) current index %d | end index %d\n", + dev->vid, vq->last_avail_idx, + vq->last_avail_idx + num_buffers); + + if (copy_mbuf_to_desc(dev, vq, pkts[pkt_idx], + buf_vec, nr_vec, + num_buffers) < 0) { + vq->shadow_used_idx -= num_buffers; + break; + } + + vq->last_avail_idx += nr_descs; + if (vq->last_avail_idx >= vq->size) { + vq->last_avail_idx -= vq->size; + vq->avail_wrap_counter ^= 1; + } + } + + do_data_copy_enqueue(dev, vq); + + if (likely(vq->shadow_used_idx)) { + flush_shadow_used_ring_packed(dev, vq); + vhost_vring_call_packed(dev, vq); } return pkt_idx; @@ -591,6 +887,7 @@ virtio_dev_rx(struct virtio_net *dev, uint16_t queue_id, struct rte_mbuf **pkts, uint32_t count) { struct vhost_virtqueue *vq; + uint32_t nb_tx = 0; VHOST_LOG_DEBUG(VHOST_DATA, "(%d) %s\n", dev->vid, __func__); if (unlikely(!is_valid_virt_queue_idx(queue_id, 0, dev->nr_vring))) { @@ -617,7 +914,10 @@ virtio_dev_rx(struct virtio_net *dev, uint16_t queue_id, if (count == 0) goto out; - count = virtio_dev_rx_split(dev, vq, pkts, count); + if (vq_is_packed(dev)) + nb_tx = virtio_dev_rx_packed(dev, vq, pkts, count); + else + nb_tx = virtio_dev_rx_split(dev, vq, pkts, count); out: if (dev->features & (1ULL << VIRTIO_F_IOMMU_PLATFORM)) @@ -626,7 +926,7 @@ out: out_access_unlock: rte_spinlock_unlock(&vq->access_lock); - return count; + return nb_tx; } uint16_t @@ -665,38 +965,39 @@ virtio_net_with_host_offload(struct virtio_net *dev) static void parse_ethernet(struct rte_mbuf *m, uint16_t *l4_proto, void **l4_hdr) { - struct ipv4_hdr *ipv4_hdr; - struct ipv6_hdr *ipv6_hdr; + struct rte_ipv4_hdr *ipv4_hdr; + struct rte_ipv6_hdr *ipv6_hdr; void *l3_hdr = NULL; - struct ether_hdr *eth_hdr; + struct rte_ether_hdr *eth_hdr; uint16_t ethertype; - eth_hdr = rte_pktmbuf_mtod(m, struct ether_hdr *); + eth_hdr = rte_pktmbuf_mtod(m, struct rte_ether_hdr *); - m->l2_len = sizeof(struct ether_hdr); + m->l2_len = sizeof(struct rte_ether_hdr); ethertype = rte_be_to_cpu_16(eth_hdr->ether_type); - if (ethertype == ETHER_TYPE_VLAN) { - struct vlan_hdr *vlan_hdr = (struct vlan_hdr *)(eth_hdr + 1); + if (ethertype == RTE_ETHER_TYPE_VLAN) { + struct rte_vlan_hdr *vlan_hdr = + (struct rte_vlan_hdr *)(eth_hdr + 1); - m->l2_len += sizeof(struct vlan_hdr); + m->l2_len += sizeof(struct rte_vlan_hdr); ethertype = rte_be_to_cpu_16(vlan_hdr->eth_proto); } l3_hdr = (char *)eth_hdr + m->l2_len; switch (ethertype) { - case ETHER_TYPE_IPv4: + case RTE_ETHER_TYPE_IPV4: ipv4_hdr = l3_hdr; *l4_proto = ipv4_hdr->next_proto_id; m->l3_len = (ipv4_hdr->version_ihl & 0x0f) * 4; *l4_hdr = (char *)l3_hdr + m->l3_len; m->ol_flags |= PKT_TX_IPV4; break; - case ETHER_TYPE_IPv6: + case RTE_ETHER_TYPE_IPV6: ipv6_hdr = l3_hdr; *l4_proto = ipv6_hdr->proto; - m->l3_len = sizeof(struct ipv6_hdr); + m->l3_len = sizeof(struct rte_ipv6_hdr); *l4_hdr = (char *)l3_hdr + m->l3_len; m->ol_flags |= PKT_TX_IPV6; break; @@ -713,7 +1014,7 @@ vhost_dequeue_offload(struct virtio_net_hdr *hdr, struct rte_mbuf *m) { uint16_t l4_proto = 0; void *l4_hdr = NULL; - struct tcp_hdr *tcp_hdr = NULL; + struct rte_tcp_hdr *tcp_hdr = NULL; if (hdr->flags == 0 && hdr->gso_type == VIRTIO_NET_HDR_GSO_NONE) return; @@ -722,15 +1023,15 @@ vhost_dequeue_offload(struct virtio_net_hdr *hdr, struct rte_mbuf *m) if (hdr->flags == VIRTIO_NET_HDR_F_NEEDS_CSUM) { if (hdr->csum_start == (m->l2_len + m->l3_len)) { switch (hdr->csum_offset) { - case (offsetof(struct tcp_hdr, cksum)): + case (offsetof(struct rte_tcp_hdr, cksum)): if (l4_proto == IPPROTO_TCP) m->ol_flags |= PKT_TX_TCP_CKSUM; break; - case (offsetof(struct udp_hdr, dgram_cksum)): + case (offsetof(struct rte_udp_hdr, dgram_cksum)): if (l4_proto == IPPROTO_UDP) m->ol_flags |= PKT_TX_UDP_CKSUM; break; - case (offsetof(struct sctp_hdr, cksum)): + case (offsetof(struct rte_sctp_hdr, cksum)): if (l4_proto == IPPROTO_SCTP) m->ol_flags |= PKT_TX_SCTP_CKSUM; break; @@ -752,7 +1053,7 @@ vhost_dequeue_offload(struct virtio_net_hdr *hdr, struct rte_mbuf *m) case VIRTIO_NET_HDR_GSO_UDP: m->ol_flags |= PKT_TX_UDP_SEG; m->tso_segsz = hdr->gso_size; - m->l4_len = sizeof(struct udp_hdr); + m->l4_len = sizeof(struct rte_udp_hdr); break; default: RTE_LOG(WARNING, VHOST_DATA, @@ -762,10 +1063,25 @@ vhost_dequeue_offload(struct virtio_net_hdr *hdr, struct rte_mbuf *m) } } -static __rte_always_inline void -put_zmbuf(struct zcopy_mbuf *zmbuf) +static __rte_noinline void +copy_vnet_hdr_from_desc(struct virtio_net_hdr *hdr, + struct buf_vector *buf_vec) { - zmbuf->in_use = 0; + uint64_t len; + uint64_t remain = sizeof(struct virtio_net_hdr); + uint64_t src; + uint64_t dst = (uint64_t)(uintptr_t)hdr; + + while (remain) { + len = RTE_MIN(remain, buf_vec->buf_len); + src = buf_vec->buf_addr; + rte_memcpy((void *)(uintptr_t)dst, + (void *)(uintptr_t)src, len); + + remain -= len; + dst += len; + buf_vec++; + } } static __rte_always_inline int @@ -794,36 +1110,16 @@ copy_desc_to_mbuf(struct virtio_net *dev, struct vhost_virtqueue *vq, goto out; } - if (likely(nr_vec > 1)) - rte_prefetch0((void *)(uintptr_t)buf_vec[1].buf_addr); - if (virtio_net_with_host_offload(dev)) { if (unlikely(buf_len < sizeof(struct virtio_net_hdr))) { - uint64_t len; - uint64_t remain = sizeof(struct virtio_net_hdr); - uint64_t src; - uint64_t dst = (uint64_t)(uintptr_t)&tmp_hdr; - uint16_t hdr_vec_idx = 0; - /* * No luck, the virtio-net header doesn't fit * in a contiguous virtual area. */ - while (remain) { - len = remain; - src = buf_vec[hdr_vec_idx].buf_addr; - rte_memcpy((void *)(uintptr_t)dst, - (void *)(uintptr_t)src, len); - - remain -= len; - dst += len; - hdr_vec_idx++; - } - + copy_vnet_hdr_from_desc(&tmp_hdr, buf_vec); hdr = &tmp_hdr; } else { hdr = (struct virtio_net_hdr *)((uintptr_t)buf_addr); - rte_prefetch0(hdr); } } @@ -853,9 +1149,6 @@ copy_desc_to_mbuf(struct virtio_net *dev, struct vhost_virtqueue *vq, buf_avail = buf_vec[vec_idx].buf_len - dev->vhost_hlen; } - rte_prefetch0((void *)(uintptr_t) - (buf_addr + buf_offset)); - PRINT_PACKET(dev, (uintptr_t)(buf_addr + buf_offset), (uint32_t)buf_avail, 0); @@ -921,14 +1214,6 @@ copy_desc_to_mbuf(struct virtio_net *dev, struct vhost_virtqueue *vq, buf_iova = buf_vec[vec_idx].buf_iova; buf_len = buf_vec[vec_idx].buf_len; - /* - * Prefecth desc n + 1 buffer while - * desc n buffer is processed. - */ - if (vec_idx + 1 < nr_vec) - rte_prefetch0((void *)(uintptr_t) - buf_vec[vec_idx + 1].buf_addr); - buf_offset = 0; buf_avail = buf_len; @@ -1004,35 +1289,7 @@ again: return NULL; } -static __rte_always_inline bool -mbuf_is_consumed(struct rte_mbuf *m) -{ - while (m) { - if (rte_mbuf_refcnt_read(m) > 1) - return false; - m = m->next; - } - - return true; -} - -static __rte_always_inline void -restore_mbuf(struct rte_mbuf *m) -{ - uint32_t mbuf_size, priv_size; - - while (m) { - priv_size = rte_pktmbuf_priv_size(m->pool); - mbuf_size = sizeof(struct rte_mbuf) + priv_size; - /* start of buffer is after mbuf structure and priv data */ - - m->buf_addr = (char *)m + mbuf_size; - m->buf_iova = rte_mempool_virt2iova(m) + mbuf_size; - m = m->next; - } -} - -static __rte_always_inline uint16_t +static __rte_noinline uint16_t virtio_dev_tx_split(struct virtio_net *dev, struct vhost_virtqueue *vq, struct rte_mempool *mbuf_pool, struct rte_mbuf **pkts, uint16_t count) { @@ -1041,16 +1298,14 @@ virtio_dev_tx_split(struct virtio_net *dev, struct vhost_virtqueue *vq, if (unlikely(dev->dequeue_zero_copy)) { struct zcopy_mbuf *zmbuf, *next; - int nr_updated = 0; for (zmbuf = TAILQ_FIRST(&vq->zmbuf_list); zmbuf != NULL; zmbuf = next) { next = TAILQ_NEXT(zmbuf, next); if (mbuf_is_consumed(zmbuf->mbuf)) { - update_shadow_used_ring(vq, zmbuf->desc_idx, 0); - nr_updated += 1; - + update_shadow_used_ring_split(vq, + zmbuf->desc_idx, 0); TAILQ_REMOVE(&vq->zmbuf_list, zmbuf, next); restore_mbuf(zmbuf->mbuf); rte_pktmbuf_free(zmbuf->mbuf); @@ -1059,17 +1314,25 @@ virtio_dev_tx_split(struct virtio_net *dev, struct vhost_virtqueue *vq, } } - flush_shadow_used_ring(dev, vq); - vhost_vring_call(dev, vq); + if (likely(vq->shadow_used_idx)) { + flush_shadow_used_ring_split(dev, vq); + vhost_vring_call_split(dev, vq); + } } - rte_prefetch0(&vq->avail->ring[vq->last_avail_idx & (vq->size - 1)]); - free_entries = *((volatile uint16_t *)&vq->avail->idx) - vq->last_avail_idx; if (free_entries == 0) return 0; + /* + * The ordering between avail index and + * desc reads needs to be enforced. + */ + rte_smp_rmb(); + + rte_prefetch0(&vq->avail->ring[vq->last_avail_idx & (vq->size - 1)]); + VHOST_LOG_DEBUG(VHOST_DATA, "(%d) %s\n", dev->vid, __func__); count = RTE_MIN(count, MAX_PKT_BURST); @@ -1079,7 +1342,8 @@ virtio_dev_tx_split(struct virtio_net *dev, struct vhost_virtqueue *vq, for (i = 0; i < count; i++) { struct buf_vector buf_vec[BUF_VECTOR_MAX]; - uint16_t head_idx, dummy_len; + uint16_t head_idx; + uint32_t dummy_len; uint16_t nr_vec = 0; int err; @@ -1091,9 +1355,7 @@ virtio_dev_tx_split(struct virtio_net *dev, struct vhost_virtqueue *vq, break; if (likely(dev->dequeue_zero_copy == 0)) - update_shadow_used_ring(vq, head_idx, 0); - - rte_prefetch0((void *)(uintptr_t)buf_vec[0].buf_addr); + update_shadow_used_ring_split(vq, head_idx, 0); pkts[i] = rte_pktmbuf_alloc(mbuf_pool); if (unlikely(pkts[i] == NULL)) { @@ -1138,8 +1400,125 @@ virtio_dev_tx_split(struct virtio_net *dev, struct vhost_virtqueue *vq, do_data_copy_dequeue(vq); if (unlikely(i < count)) vq->shadow_used_idx = i; - flush_shadow_used_ring(dev, vq); - vhost_vring_call(dev, vq); + if (likely(vq->shadow_used_idx)) { + flush_shadow_used_ring_split(dev, vq); + vhost_vring_call_split(dev, vq); + } + } + + return i; +} + +static __rte_noinline uint16_t +virtio_dev_tx_packed(struct virtio_net *dev, struct vhost_virtqueue *vq, + struct rte_mempool *mbuf_pool, struct rte_mbuf **pkts, uint16_t count) +{ + uint16_t i; + + if (unlikely(dev->dequeue_zero_copy)) { + struct zcopy_mbuf *zmbuf, *next; + + for (zmbuf = TAILQ_FIRST(&vq->zmbuf_list); + zmbuf != NULL; zmbuf = next) { + next = TAILQ_NEXT(zmbuf, next); + + if (mbuf_is_consumed(zmbuf->mbuf)) { + update_shadow_used_ring_packed(vq, + zmbuf->desc_idx, + 0, + zmbuf->desc_count); + + TAILQ_REMOVE(&vq->zmbuf_list, zmbuf, next); + restore_mbuf(zmbuf->mbuf); + rte_pktmbuf_free(zmbuf->mbuf); + put_zmbuf(zmbuf); + vq->nr_zmbuf -= 1; + } + } + + if (likely(vq->shadow_used_idx)) { + flush_shadow_used_ring_packed(dev, vq); + vhost_vring_call_packed(dev, vq); + } + } + + VHOST_LOG_DEBUG(VHOST_DATA, "(%d) %s\n", dev->vid, __func__); + + count = RTE_MIN(count, MAX_PKT_BURST); + VHOST_LOG_DEBUG(VHOST_DATA, "(%d) about to dequeue %u buffers\n", + dev->vid, count); + + for (i = 0; i < count; i++) { + struct buf_vector buf_vec[BUF_VECTOR_MAX]; + uint16_t buf_id; + uint32_t dummy_len; + uint16_t desc_count, nr_vec = 0; + int err; + + if (unlikely(fill_vec_buf_packed(dev, vq, + vq->last_avail_idx, &desc_count, + buf_vec, &nr_vec, + &buf_id, &dummy_len, + VHOST_ACCESS_RO) < 0)) + break; + + if (likely(dev->dequeue_zero_copy == 0)) + update_shadow_used_ring_packed(vq, buf_id, 0, + desc_count); + + pkts[i] = rte_pktmbuf_alloc(mbuf_pool); + if (unlikely(pkts[i] == NULL)) { + RTE_LOG(ERR, VHOST_DATA, + "Failed to allocate memory for mbuf.\n"); + break; + } + + err = copy_desc_to_mbuf(dev, vq, buf_vec, nr_vec, pkts[i], + mbuf_pool); + if (unlikely(err)) { + rte_pktmbuf_free(pkts[i]); + break; + } + + if (unlikely(dev->dequeue_zero_copy)) { + struct zcopy_mbuf *zmbuf; + + zmbuf = get_zmbuf(vq); + if (!zmbuf) { + rte_pktmbuf_free(pkts[i]); + break; + } + zmbuf->mbuf = pkts[i]; + zmbuf->desc_idx = buf_id; + zmbuf->desc_count = desc_count; + + /* + * Pin lock the mbuf; we will check later to see + * whether the mbuf is freed (when we are the last + * user) or not. If that's the case, we then could + * update the used ring safely. + */ + rte_mbuf_refcnt_update(pkts[i], 1); + + vq->nr_zmbuf += 1; + TAILQ_INSERT_TAIL(&vq->zmbuf_list, zmbuf, next); + } + + vq->last_avail_idx += desc_count; + if (vq->last_avail_idx >= vq->size) { + vq->last_avail_idx -= vq->size; + vq->avail_wrap_counter ^= 1; + } + } + + if (likely(dev->dequeue_zero_copy == 0)) { + do_data_copy_dequeue(vq); + if (unlikely(i < count)) + vq->shadow_used_idx = i; + if (likely(vq->shadow_used_idx)) { + flush_shadow_used_ring_packed(dev, vq); + vhost_vring_call_packed(dev, vq); + } } return i; @@ -1175,15 +1554,19 @@ rte_vhost_dequeue_burst(int vid, uint16_t queue_id, if (unlikely(rte_spinlock_trylock(&vq->access_lock) == 0)) return 0; - if (unlikely(vq->enabled == 0)) + if (unlikely(vq->enabled == 0)) { + count = 0; goto out_access_unlock; + } if (dev->features & (1ULL << VIRTIO_F_IOMMU_PLATFORM)) vhost_user_iotlb_rd_lock(vq); if (unlikely(vq->access_ok == 0)) - if (unlikely(vring_translate(dev, vq) < 0)) + if (unlikely(vring_translate(dev, vq) < 0)) { + count = 0; goto out; + } /* * Construct a RARP broadcast packet, and inject it to the "pkts" @@ -1208,12 +1591,16 @@ rte_vhost_dequeue_burst(int vid, uint16_t queue_id, if (rarp_mbuf == NULL) { RTE_LOG(ERR, VHOST_DATA, "Failed to make RARP packet.\n"); - return 0; + count = 0; + goto out; } count -= 1; } - count = virtio_dev_tx_split(dev, vq, mbuf_pool, pkts, count); + if (vq_is_packed(dev)) + count = virtio_dev_tx_packed(dev, vq, mbuf_pool, pkts, count); + else + count = virtio_dev_tx_split(dev, vq, mbuf_pool, pkts, count); out: if (dev->features & (1ULL << VIRTIO_F_IOMMU_PLATFORM))