[dpdk-dev] [PATCH v1] vhost: support async copy free segmentations

[patrick . fu]

From: Patrick Fu <patrick...@intel.com>

Vhost async enqueue assumes that all async copies should break at packet
boundary. i.e. if a packet is splited into multiple copy segments, the
async engine should always report copy completion when entire packet is
finished. This patch removes the assumption.

Fixes: cd6760da1076 ("vhost: introduce async enqueue for split ring")

Signed-off-by: Patrick Fu <patrick...@intel.com>
---
 lib/librte_vhost/vhost.h      |  3 +++
 lib/librte_vhost/virtio_net.c | 12 ++++++++----
 2 files changed, 11 insertions(+), 4 deletions(-)

diff --git a/lib/librte_vhost/vhost.h b/lib/librte_vhost/vhost.h
index 8c01cee42..0f7212f88 100644
--- a/lib/librte_vhost/vhost.h
+++ b/lib/librte_vhost/vhost.h
@@ -46,6 +46,8 @@
 
 #define MAX_PKT_BURST 32
 
+#define ASYNC_MAX_POLL_SEG 255
+
 #define VHOST_MAX_ASYNC_IT (MAX_PKT_BURST * 2)
 #define VHOST_MAX_ASYNC_VEC (BUF_VECTOR_MAX * 2)
 
@@ -225,6 +227,7 @@ struct vhost_virtqueue {
        uint64_t        *async_pending_info;
        uint16_t        async_pkts_idx;
        uint16_t        async_pkts_inflight_n;
+       uint16_t        async_last_seg_n;
 
        /* vq async features */
        bool            async_inorder;
diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c
index 1d0be3dd4..c6fa33f37 100644
--- a/lib/librte_vhost/virtio_net.c
+++ b/lib/librte_vhost/virtio_net.c
@@ -1652,12 +1652,14 @@ uint16_t rte_vhost_poll_enqueue_completed(int vid, 
uint16_t queue_id,
        start_idx = virtio_dev_rx_async_get_info_idx(pkts_idx,
                vq_size, vq->async_pkts_inflight_n);
 
-       n_pkts_cpl =
-               vq->async_ops.check_completed_copies(vid, queue_id, 0, count);
+       n_pkts_cpl = vq->async_ops.check_completed_copies(vid, queue_id,
+               0, ASYNC_MAX_POLL_SEG - vq->async_last_seg_n) +
+               vq->async_last_seg_n;
 
        rte_smp_wmb();
 
-       while (likely(((start_idx + n_pkts_put) & (vq_size - 1)) != pkts_idx)) {
+       while (likely((n_pkts_put < count) &&
+               (((start_idx + n_pkts_put) & (vq_size - 1)) != pkts_idx))) {
                uint64_t info = async_pending_info[
                        (start_idx + n_pkts_put) & (vq_size - 1)];
                uint64_t n_segs;
@@ -1666,7 +1668,7 @@ uint16_t rte_vhost_poll_enqueue_completed(int vid, 
uint16_t queue_id,
                n_segs = info >> ASYNC_PENDING_INFO_N_SFT;
 
                if (n_segs) {
-                       if (!n_pkts_cpl || n_pkts_cpl < n_segs) {
+                       if (unlikely(n_pkts_cpl < n_segs)) {
                                n_pkts_put--;
                                n_descs -= info & ASYNC_PENDING_INFO_N_MSK;
                                if (n_pkts_cpl) {
@@ -1684,6 +1686,8 @@ uint16_t rte_vhost_poll_enqueue_completed(int vid, 
uint16_t queue_id,
                }
        }
 
+       vq->async_last_seg_n = n_pkts_cpl;
+
        if (n_pkts_put) {
                vq->async_pkts_inflight_n -= n_pkts_put;
                __atomic_add_fetch(&vq->used->idx, n_descs, __ATOMIC_RELEASE);
-- 
2.18.4