From: Luigi Leonardi <luigi.leona...@outlook.com>
Introduce an optimization in virtio_transport_send_pkt:
when the work queue (send_pkt_queue) is empty the packet is
put directly in the virtqueue increasing the throughput.
In the following benchmark (pingpong mode) the host sends
a payload to the guest and waits for the same payload back.
All vCPUs pinned individually to pCPUs.
vhost process pinned to a pCPU
fio process pinned both inside the host and the guest system.
Host CPU: Intel i7-10700KF CPU @ 3.80GHz
Tool: Fio version 3.37-56
Env: Phys host + L1 Guest
Runtime-per-test: 50s
Mode: pingpong (h-g-h)
Test runs: 50
Type: SOCK_STREAM
Before: Linux 6.9.7
Payload 512B:
1st perc. overall 99th perc.
Before 370 810.15 8656 ns
After 374 780.29 8741 ns
Payload 4K:
1st perc. overall 99th perc.
Before 460 1720.23 42752 ns
After 460 1520.84 36096 ns
The performance improvement is related to this optimization,
I used ebpf to check that each packet was sent directly to the
virtqueue.
Throughput: iperf-vsock
The size represents the buffer length (-l) to read/write
P represents the number parallel streams
P=1
4K 64K 128K
Before 6.87 29.3 29.5 Gb/s
After 10.5 39.4 39.9 Gb/s
P=2
4K 64K 128K
Before 10.5 32.8 33.2 Gb/s
After 17.8 47.7 48.5 Gb/s
P=4
4K 64K 128K
Before 12.7 33.6 34.2 Gb/s
After 16.9 48.1 50.5 Gb/s
Co-developed-by: Marco Pinna <marco.pin...@gmail.com>
Signed-off-by: Marco Pinna <marco.pin...@gmail.com>
Signed-off-by: Luigi Leonardi <luigi.leona...@outlook.com>
---
net/vmw_vsock/virtio_transport.c | 38 ++++++++++++++++++++++++++++++++++----
1 file changed, 34 insertions(+), 4 deletions(-)
diff --git a/net/vmw_vsock/virtio_transport.c b/net/vmw_vsock/virtio_transport.c
index c4205c22f40b..d75727fdc35f 100644
--- a/net/vmw_vsock/virtio_transport.c
+++ b/net/vmw_vsock/virtio_transport.c
@@ -208,6 +208,29 @@ virtio_transport_send_pkt_work(struct work_struct *work)
queue_work(virtio_vsock_workqueue, &vsock->rx_work);
}
+/* Caller need to hold RCU for vsock.
+ * Returns 0 if the packet is successfully put on the vq.
+ */
+static int virtio_transport_send_skb_fast_path(struct virtio_vsock *vsock,
struct sk_buff *skb)
+{
+ struct virtqueue *vq = vsock->vqs[VSOCK_VQ_TX];
+ int ret;
+
+ /* Inside RCU, can't sleep! */
+ ret = mutex_trylock(&vsock->tx_lock);
+ if (unlikely(ret == 0))
+ return -EBUSY;
+
+ ret = virtio_transport_send_skb(skb, vq, vsock);
+
+ mutex_unlock(&vsock->tx_lock);
+
+ /* Kick if virtio_transport_send_skb succeeded */
+ if (ret == 0)
+ virtqueue_kick(vq);
+ return ret;
+}
+
static int
virtio_transport_send_pkt(struct sk_buff *skb)
{
@@ -231,11 +254,18 @@ virtio_transport_send_pkt(struct sk_buff *skb)
goto out_rcu;
}
- if (virtio_vsock_skb_reply(skb))
- atomic_inc(&vsock->queued_replies);
+ /* If the workqueue (send_pkt_queue) is empty there is no need to
enqueue the packet.
+ * Just put it on the virtqueue using
virtio_transport_send_skb_fast_path.
+ */
- virtio_vsock_skb_queue_tail(&vsock->send_pkt_queue, skb);
- queue_work(virtio_vsock_workqueue, &vsock->send_pkt_work);
+ if (!skb_queue_empty_lockless(&vsock->send_pkt_queue) ||
+ virtio_transport_send_skb_fast_path(vsock, skb)) {
+ /* Packet must be queued */
+ if (virtio_vsock_skb_reply(skb))
+ atomic_inc(&vsock->queued_replies);
+ virtio_vsock_skb_queue_tail(&vsock->send_pkt_queue, skb);
+ queue_work(virtio_vsock_workqueue, &vsock->send_pkt_work);
+ }
out_rcu:
rcu_read_unlock();
--
2.45.2
