virtio: optimize ring layout

In DPDK based switching environment, mostly vhost runs on a dedicated core
while virtio processing in guest VMs runs on different cores.
Take RX for example, with generic implementation, for each guest buffer,
a) virtio driver allocates a descriptor from free descriptor list
b) modify the entry of avail ring to point to allocated descriptor
c) after packet is received, free the descriptor

When vhost fetches the avail ring, it need to fetch the modified L1 cache from
virtio core, which is a heavy cost in current CPU implementation.

This idea of this optimization is:
    allocate the fixed descriptor for each entry of avail ring, so avail ring will
always be the same during the run.
This removes L1M cache transfer from virtio core to vhost core for avail ring.
(Note we couldn't avoid the cache transfer for descriptors).
Besides, descriptor allocation and free operation is eliminated.
This also makes vector procesing possible to further accelerate the processing.

This is the layout for the avail ring(take 256 ring entries for example), with
each entry pointing to the descriptor with the same index.
                    avail
                    idx
                    +
                    |
+----+----+---+-------------+------+
| 0  | 1  | 2 | ... |  254  | 255  |  avail ring
+-+--+-+--+-+-+---------+---+--+---+
  |    |    |       |   |      |
  |    |    |       |   |      |
  v    v    v       |   v      v
+-+--+-+--+-+-+---------+---+--+---+
| 0  | 1  | 2 | ... |  254  | 255  |  desc ring
+----+----+---+-------------+------+
                    |
                    |
+----+----+---+-------------+------+
| 0  | 1  | 2 |     |  254  | 255  |  used ring
+----+----+---+-------------+------+
                    |
                    +

This is the ring layout for TX.
As we need one virtio header for each xmit packet, we have 128 slots available.

                         ++
                         ||
                         ||
+-----+-----+-----+--------------+------+------+------+
|  0  |  1  | ... |  127 || 128  | 129  | ...  | 255  |   avail ring
+--+--+--+--+-----+---+------+---+--+---+------+--+---+
   |     |            |  ||  |      |             |
   v     v            v  ||  v      v             v
+--+--+--+--+-----+---+------+---+--+---+------+--+---+
| 128 | 129 | ... |  255 || 128  | 129  | ...  | 255  |   desc ring for virtio_net_hdr
+--+--+--+--+-----+---+------+---+--+---+------+--+---+
   |     |            |  ||  |      |             |
   v     v            v  ||  v      v             v
+--+--+--+--+-----+---+------+---+--+---+------+--+---+
|  0  |  1  | ... |  127 ||  0   |  1   | ...  | 127  |   desc ring for tx dat
+-----+-----+-----+--------------+------+------+------+
                         ||
                         ||
                         ++

Signed-off-by: Huawei Xie <huawei.xie@intel.com>
Acked-by: Jianfeng Tan <jianfeng.tan@intel.com>

diff --git a/doc/guides/rel_notes/release_2_2.rst b/doc/guides/rel_notes/release_2_2.rst
index 8a20044..ba730f6 100644 (file)
--- a/doc/guides/rel_notes/release_2_2.rst
+++ b/doc/guides/rel_notes/release_2_2.rst
@@ -55,6 +55,10 @@ New Features
   Like mlx4, this PMD is only available for Linux and is disabled by default
   due to external dependencies (libibverbs and libmlx5).
 
+* **Enhanced support for virtio driver.**
+
+  * Virtio ring layout optimization (fixed avail ring)
+
 * **Added vhost-user multiple queue support.**
 
 * **Added port hotplug support to xenvirt.**

diff --git a/drivers/net/virtio/virtio_rxtx.c b/drivers/net/virtio/virtio_rxtx.c
index d35c5f9..68103f8 100644 (file)
--- a/drivers/net/virtio/virtio_rxtx.c
+++ b/drivers/net/virtio/virtio_rxtx.c
@@ -61,6 +61,8 @@
 #define  VIRTIO_DUMP_PACKET(m, len) do { } while (0)
 #endif
 
+static int use_simple_rxtx;
+
 static void
 vq_ring_free_chain(struct virtqueue *vq, uint16_t desc_idx)
 {
@@ -298,6 +300,13 @@ virtio_dev_vring_start(struct virtqueue *vq, int queue_type)
                /* Allocate blank mbufs for the each rx descriptor */
                nbufs = 0;
                error = ENOSPC;
+
+               if (use_simple_rxtx)
+                       for (i = 0; i < vq->vq_nentries; i++) {
+                               vq->vq_ring.avail->ring[i] = i;
+                               vq->vq_ring.desc[i].flags = VRING_DESC_F_WRITE;
+                       }
+
                while (!virtqueue_full(vq)) {
                        m = rte_rxmbuf_alloc(vq->mpool);
                        if (m == NULL)
@@ -324,6 +333,24 @@ virtio_dev_vring_start(struct virtqueue *vq, int queue_type)
                VIRTIO_WRITE_REG_4(vq->hw, VIRTIO_PCI_QUEUE_PFN,
                        vq->mz->phys_addr >> VIRTIO_PCI_QUEUE_ADDR_SHIFT);
        } else if (queue_type == VTNET_TQ) {
+               if (use_simple_rxtx) {
+                       int mid_idx  = vq->vq_nentries >> 1;
+                       for (i = 0; i < mid_idx; i++) {
+                               vq->vq_ring.avail->ring[i] = i + mid_idx;
+                               vq->vq_ring.desc[i + mid_idx].next = i;
+                               vq->vq_ring.desc[i + mid_idx].addr =
+                                       vq->virtio_net_hdr_mem +
+                                               mid_idx * vq->hw->vtnet_hdr_size;
+                               vq->vq_ring.desc[i + mid_idx].len =
+                                       vq->hw->vtnet_hdr_size;
+                               vq->vq_ring.desc[i + mid_idx].flags =
+                                       VRING_DESC_F_NEXT;
+                               vq->vq_ring.desc[i].flags = 0;
+                       }
+                       for (i = mid_idx; i < vq->vq_nentries; i++)
+                               vq->vq_ring.avail->ring[i] = i;
+               }
+
                VIRTIO_WRITE_REG_2(vq->hw, VIRTIO_PCI_QUEUE_SEL,
                        vq->vq_queue_index);
                VIRTIO_WRITE_REG_4(vq->hw, VIRTIO_PCI_QUEUE_PFN,