QEMU直接从tap/tun取数据
QEMU tap数据接收步骤：
- qemu从tun取数据包
- qemu将数据包放入virtio硬件网卡。
- qemu触发中断。
- 虚拟机收到中断，从virtio读取数据。
在qemu中步骤1（tap_read_packet）和步骤2（qemu_send_packet_async）都是在tap_send中完成的，其中步骤2是异步流程。
qemu/net/tap.c
static void tap_send(void *opaque)
{
TAPState *s = opaque;
int size;
int packets = 0;
while (true) {
uint8_t *buf = s->buf;
uint8_t min_pkt[ETH_ZLEN];
size_t min_pktsz = sizeof(min_pkt);
size = tap_read_packet(s->fd, s->buf, sizeof(s->buf));
if (size <= 0) {
break;
}
if (s->host_vnet_hdr_len && !s->using_vnet_hdr) {
buf += s->host_vnet_hdr_len;
size -= s->host_vnet_hdr_len;
}
if (net_peer_needs_padding(&s->nc)) {
if (eth_pad_short_frame(min_pkt, &min_pktsz, buf, size)) {
buf = min_pkt;
size = min_pktsz;
}
}
size = qemu_send_packet_async(&s->nc, buf, size, tap_send_completed);
if (size == 0) {
tap_read_poll(s, false);
break;
} else if (size < 0) {
break;
}
/*
* When the host keeps receiving more packets while tap_send() is
* running we can hog the QEMU global mutex. Limit the number of
* packets that are processed per tap_send() callback to prevent
* stalling the guest.
*/
packets++;
if (packets >= 50) {
break;
}
}
}
qemu通过qemu_net_queue_deliver将数据包发送到virtio_queue，在发送之前若delivering或!qemu_can_send_packet满足，则先将数据包加入packets队列，随后在qemu_net_queue_flush阶段将数据包发送到virtio_queue中，上图中virtio_net_receive就到达virtio虚拟硬件网卡了。
QEMU通过vhost-net从tap/tun取数据
vhost-net驱动加载时会生成/dev/vhost-net设备。
qemu-kvm启动时会open设备/dev/vhost-net，将调用vhost_net_open完成这个过程，vhost_net_open会进行handle_tx_net、handle_rx_net poll函数的初始化。
handle_tx_net、handle_rx_net最终会调用tun_recvmsg、tun_sendmsg进行数据收发。
/drivers/vhost/net.c:
static int vhost_net_open(struct inode *inode, struct file *f)
{
... ...
vhost_poll_init(n->poll + VHOST_NET_VQ_TX, handle_tx_net, EPOLLOUT, dev);
vhost_poll_init(n->poll + VHOST_NET_VQ_RX, handle_rx_net, EPOLLIN, dev);
... ...
}
static void handle_rx_net(struct vhost_work *work)
{
struct vhost_net *net = container_of(work, struct vhost_net,
poll[VHOST_NET_VQ_RX].work);
handle_rx(net);
}
handle_rx函数中recvmsg完成从tun取数据，通过copy_to_iter将msg放入virtio_queue，最后vhost_add_used_and_signal_n实现通知机制，qemu收到数据。
static void handle_rx(struct vhost_net *net)
{
... ...
err = sock->ops->recvmsg(sock, &msg,
sock_len, MSG_DONTWAIT | MSG_TRUNC);
... ...
num_buffers = cpu_to_vhost16(vq, headcount);
if (likely(mergeable) &&
copy_to_iter(&num_buffers, sizeof num_buffers,
&fixup) != sizeof num_buffers) {
vq_err(vq, "Failed num_buffers write");
vhost_discard_vq_desc(vq, headcount);
break;
}
vhost_add_used_and_signal_n(&net->dev, vq, vq->heads,
headcount);
... ...
}
vhost_net通过vhost_worker内核线程进行工作队列的调度用于完成poll，vhost_worker内核线程是qemu通过vhost_dev_ioctl VHOST_SET_OWNER时创建的。
drivers/vhost/vhost.c:
static int vhost_poll_wakeup(wait_queue_entry_t *wait, unsigned mode, int sync,
void *key)
{
struct vhost_poll *poll = container_of(wait, struct vhost_poll, wait);
if (!(key_to_poll(key) & poll->mask))
return 0;
vhost_poll_queue(poll);
return 0;
}
void vhost_work_init(struct vhost_work *work, vhost_work_fn_t fn)
{
clear_bit(VHOST_WORK_QUEUED, &a