vhost: optimize cache access

This patch reorders the code to delay virtio header write to improve
cache access efficiency for cases where the mrg_rxbuf feature is turned
on. CPU pipeline stall cycles can be significantly reduced.

Virtio header write and mbuf data copy are all remote store operations
which takes a long time to finish. It's a good idea to put them together
to remove bubbles in between, to let as many remote store instructions
as possible go into store buffer at the same time to hide latency, and
to let the H/W prefetcher goes to work as early as possible.

On a Haswell machine, about 100 cycles can be saved per packet by this
patch alone. Taking 64B packets traffic for example, this means about 60%
efficiency improvement for the enqueue operation.

Signed-off-by: Zhihong Wang <zhihong.wang@intel.com>
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Jianbo Liu <jianbo.liu@linaro.org>
Tested-by: Maxime Coquelin <maxime.coquelin@redhat.com>

diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c
index 15ef0b0..27ebf3d 100644 (file)
--- a/lib/librte_vhost/virtio_net.c
+++ b/lib/librte_vhost/virtio_net.c
@@ -390,6 +390,8 @@ copy_mbuf_to_desc_mergeable(struct virtio_net *dev, struct vhost_virtqueue *vq,
        uint32_t desc_offset, desc_avail;
        uint32_t cpy_len;
        uint16_t desc_idx, used_idx;
+       uint64_t hdr_addr, hdr_phys_addr;
+       struct rte_mbuf *hdr_mbuf;
 
        if (unlikely(m == NULL))
                return 0;
@@ -401,17 +403,15 @@ copy_mbuf_to_desc_mergeable(struct virtio_net *dev, struct vhost_virtqueue *vq,
        if (buf_vec[vec_idx].buf_len < dev->vhost_hlen || !desc_addr)
                return 0;
 
-       rte_prefetch0((void *)(uintptr_t)desc_addr);
+       hdr_mbuf = m;
+       hdr_addr = desc_addr;
+       hdr_phys_addr = buf_vec[vec_idx].buf_addr;
+       rte_prefetch0((void *)(uintptr_t)hdr_addr);
 
        virtio_hdr.num_buffers = end_idx - start_idx;
        LOG_DEBUG(VHOST_DATA, "(%d) RX: num merge buffers %d\n",
                dev->vid, virtio_hdr.num_buffers);
 
-       virtio_enqueue_offload(m, &virtio_hdr.hdr);
-       copy_virtio_net_hdr(dev, desc_addr, virtio_hdr);
-       vhost_log_write(dev, buf_vec[vec_idx].buf_addr, dev->vhost_hlen);
-       PRINT_PACKET(dev, (uintptr_t)desc_addr, dev->vhost_hlen, 0);
-
        desc_avail  = buf_vec[vec_idx].buf_len - dev->vhost_hlen;
        desc_offset = dev->vhost_hlen;
        desc_chain_head = buf_vec[vec_idx].desc_idx;
@@ -456,6 +456,16 @@ copy_mbuf_to_desc_mergeable(struct virtio_net *dev, struct vhost_virtqueue *vq,
                        mbuf_avail  = rte_pktmbuf_data_len(m);
                }
 
+               if (hdr_addr) {
+                       virtio_enqueue_offload(hdr_mbuf, &virtio_hdr.hdr);
+                       copy_virtio_net_hdr(dev, hdr_addr, virtio_hdr);
+                       vhost_log_write(dev, hdr_phys_addr, dev->vhost_hlen);
+                       PRINT_PACKET(dev, (uintptr_t)hdr_addr,
+                                    dev->vhost_hlen, 0);
+
+                       hdr_addr = 0;
+               }
+
                cpy_len = RTE_MIN(desc_avail, mbuf_avail);
                rte_memcpy((void *)((uintptr_t)(desc_addr + desc_offset)),
                        rte_pktmbuf_mtod_offset(m, void *, mbuf_offset),