diff --git a/devicemodel/core/main.c b/devicemodel/core/main.c
index a0ed4cc49..1ffa6f0b4 100644
--- a/devicemodel/core/main.c
+++ b/devicemodel/core/main.c
@@ -65,6 +65,7 @@
 #include "vmcfg_config.h"
 #include "vmcfg.h"
 #include "tpm.h"
+#include "virtio.h"
 
 #define GUEST_NIO_PORT		0x488	/* guest upcalls via i/o port */
 
@@ -703,6 +704,7 @@ enum {
 	CMD_OPT_VSBL = 1000,
 	CMD_OPT_PART_INFO,
 	CMD_OPT_TRUSTY_ENABLE,
+	CMD_OPT_VIRTIO_POLL_ENABLE,
 	CMD_OPT_PTDEV_NO_RESET,
 	CMD_OPT_DEBUGEXIT,
 	CMD_OPT_VMCFG,
@@ -739,6 +741,7 @@ static struct option long_options[] = {
 	{"part_info",		required_argument,	0, CMD_OPT_PART_INFO},
 	{"enable_trusty",	no_argument,		0,
 					CMD_OPT_TRUSTY_ENABLE},
+	{"virtio_poll",		required_argument,	0, CMD_OPT_VIRTIO_POLL_ENABLE},
 	{"ptdev_no_reset",	no_argument,		0,
 		CMD_OPT_PTDEV_NO_RESET},
 	{"debugexit",		no_argument,		0, CMD_OPT_DEBUGEXIT},
@@ -862,6 +865,14 @@ dm_run(int argc, char *argv[])
 		case CMD_OPT_TRUSTY_ENABLE:
 			trusty_enabled = 1;
 			break;
+		case CMD_OPT_VIRTIO_POLL_ENABLE:
+			if (acrn_parse_virtio_poll_interval(optarg) != 0) {
+				errx(EX_USAGE,
+					"invalid virtio poll interval %s",
+					optarg);
+				exit(1);
+			}
+			break;
 		case CMD_OPT_PTDEV_NO_RESET:
 			ptdev_no_reset(true);
 			break;
diff --git a/devicemodel/hw/pci/virtio/virtio.c b/devicemodel/hw/pci/virtio/virtio.c
index e4cc381ac..1d5dc3150 100644
--- a/devicemodel/hw/pci/virtio/virtio.c
+++ b/devicemodel/hw/pci/virtio/virtio.c
@@ -28,10 +28,13 @@
 #include <stdio.h>
 #include <stddef.h>
 #include <pthread.h>
+#include <string.h>
+#include <stdlib.h>
 
 #include "dm.h"
 #include "pci_core.h"
 #include "virtio.h"
+#include "timer.h"
 
 /*
  * Functions for dealing with generalized "virtual devices" as
@@ -45,6 +48,61 @@
  */
 #define DEV_STRUCT(vs) ((void *)(vs))
 
+static uint8_t virtio_poll_enabled;
+static size_t virtio_poll_interval;
+
+static void
+virtio_start_timer(struct acrn_timer *timer, time_t sec, time_t nsec)
+{
+	struct itimerspec ts;
+
+	/* setting the interval time */
+	ts.it_interval.tv_sec = 0;
+	ts.it_interval.tv_nsec = 0;
+	/* set the delay time it will be started when timer_setting */
+	ts.it_value.tv_sec = sec;
+	ts.it_value.tv_nsec = nsec;
+	assert(acrn_timer_settime(timer, &ts) == 0);
+}
+
+static void
+virtio_poll_timer(void *arg)
+{
+	struct virtio_base *base;
+	struct virtio_ops *vops;
+	struct virtio_vq_info *vq;
+	const char *name;
+	int i;
+
+	base = arg;
+	vops = base->vops;
+	name = vops->name;
+
+	if (base->mtx)
+		pthread_mutex_lock(base->mtx);
+
+	base->polling_in_progress = 1;
+
+	for (i = 0; i < base->vops->nvq; i++) {
+		vq = &base->queues[i];
+		vq->used->flags |= ACRN_VRING_USED_F_NO_NOTIFY;
+		/* TODO: call notify when necessary */
+		if (vq->notify)
+			(*vq->notify)(DEV_STRUCT(base), vq);
+		else if (vops->qnotify)
+			(*vops->qnotify)(DEV_STRUCT(base), vq);
+		else
+			fprintf(stderr,
+				"%s: qnotify queue %d: missing vq/vops notify\r\n",
+				name, i);
+	}
+
+	if (base->mtx)
+		pthread_mutex_unlock(base->mtx);
+
+	virtio_start_timer(&base->polling_timer, 0, virtio_poll_interval);
+}
+
 /**
  * @brief Link a virtio_base to its constants, the virtio device,
  * and the PCI emulation.
@@ -60,7 +118,8 @@
 void
 virtio_linkup(struct virtio_base *base, struct virtio_ops *vops,
 	      void *pci_virtio_dev, struct pci_vdev *dev,
-	      struct virtio_vq_info *queues)
+	      struct virtio_vq_info *queues,
+	      int backend_type)
 {
 	int i;
 
@@ -69,6 +128,7 @@ virtio_linkup(struct virtio_base *base, struct virtio_ops *vops,
 	base->vops = vops;
 	base->dev = dev;
 	dev->arg = base;
+	base->backend_type = backend_type;
 
 	base->queues = queues;
 	for (i = 0; i < vops->nvq; i++) {
@@ -100,6 +160,9 @@ virtio_reset_dev(struct virtio_base *base)
 /* if (base->mtx) */
 /* assert(pthread_mutex_isowned_np(base->mtx)); */
 
+	acrn_timer_deinit(&base->polling_timer);
+	base->polling_in_progress = 0;
+
 	nvq = base->vops->nvq;
 	for (vq = base->queues, i = 0; i < nvq; vq++, i++) {
 		vq->flags = 0;
@@ -570,6 +633,30 @@ vq_endchains(struct virtio_vq_info *vq, int used_all_avail)
 		vq_interrupt(base, vq);
 }
 
+/**
+ * @brief Helper function for clearing used ring flags.
+ *
+ * Driver should always use this helper function to clear used ring flags.
+ * For virtio poll mode, in order to avoid trap, we should never really
+ * clear used ring flags.
+ *
+ * @param base Pointer to struct virtio_base.
+ * @param vq Pointer to struct virtio_vq_info.
+ *
+ * @return None
+ */
+void vq_clear_used_ring_flags(struct virtio_base *base, struct virtio_vq_info *vq)
+{
+	int backend_type = base->backend_type;
+	int polling_in_progress = base->polling_in_progress;
+
+	/* we should never unmask notification in polling mode */
+	if (virtio_poll_enabled && backend_type == BACKEND_VBSU && polling_in_progress == 1)
+		return;
+
+	vq->used->flags &= ~ACRN_VRING_USED_F_NO_NOTIFY;
+}
+
 struct config_reg {
 	uint16_t	offset;	/* register offset */
 	uint8_t		size;	/* size (bytes) */
@@ -877,6 +964,17 @@ bad:
 			(*vops->set_status)(DEV_STRUCT(base), value);
 		if (value == 0)
 			(*vops->reset)(DEV_STRUCT(base));
+		if ((value & VIRTIO_CR_STATUS_DRIVER_OK) &&
+		     base->backend_type == BACKEND_VBSU &&
+		     virtio_poll_enabled) {
+			base->polling_timer.clockid = CLOCK_MONOTONIC;
+			acrn_timer_init(&base->polling_timer, virtio_poll_timer, base);
+			/* wait 5s to start virtio poll mode
+			 * skip vsbl and make sure device initialization completed
+			 * FIXME: Need optimization in the future
+			 */
+			virtio_start_timer(&base->polling_timer, 5, 0);
+		}
 		break;
 	case VIRTIO_CR_CFGVEC:
 		base->msix_cfg_idx = value;
@@ -1302,6 +1400,7 @@ virtio_common_cfg_write(struct pci_vdev *dev, uint64_t offset, int size,
 			(*vops->set_status)(DEV_STRUCT(base), value);
 		if (base->status == 0)
 			(*vops->reset)(DEV_STRUCT(base));
+		/* TODO: virtio poll mode for modern devices */
 		break;
 	case VIRTIO_COMMON_Q_SELECT:
 		/*
@@ -1878,3 +1977,26 @@ virtio_pci_modern_cfgwrite(struct vmctx *ctx, int vcpu, struct pci_vdev *dev,
 
 	return -1;
 }
+
+/**
+ * @brief Get the virtio poll parameters
+ *
+ * @param optarg Pointer to parameters string.
+ *
+ * @return fail -1 success 0
+ */
+int
+acrn_parse_virtio_poll_interval(const char *optarg)
+{
+	char *ptr;
+
+	virtio_poll_interval = strtoul(optarg, &ptr, 0);
+
+	/* poll interval is limited from 1us to 10ms */
+	if (virtio_poll_interval < 1 || virtio_poll_interval > 10000000)
+		return -1;
+
+	virtio_poll_enabled = 1;
+
+	return 0;
+}
\ No newline at end of file
diff --git a/devicemodel/hw/pci/virtio/virtio_audio.c b/devicemodel/hw/pci/virtio/virtio_audio.c
index 919fdb78a..26bee5c19 100644
--- a/devicemodel/hw/pci/virtio/virtio_audio.c
+++ b/devicemodel/hw/pci/virtio/virtio_audio.c
@@ -316,7 +316,8 @@ virtio_audio_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 		      &virtio_audio_ops_k,
 		      virt_audio,
 		      dev,
-		      virt_audio->vq);
+		      virt_audio->vq,
+		      BACKEND_VBSK);
 
 	rc = virtio_audio_kernel_init(virt_audio);
 	if (rc < 0) {
diff --git a/devicemodel/hw/pci/virtio/virtio_block.c b/devicemodel/hw/pci/virtio/virtio_block.c
index beb0c4162..4c405983b 100644
--- a/devicemodel/hw/pci/virtio/virtio_block.c
+++ b/devicemodel/hw/pci/virtio/virtio_block.c
@@ -384,7 +384,7 @@ virtio_blk_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 					"error %d!\n", rc));
 
 	/* init virtio struct and virtqueues */
-	virtio_linkup(&blk->base, &virtio_blk_ops, blk, dev, &blk->vq);
+	virtio_linkup(&blk->base, &virtio_blk_ops, blk, dev, &blk->vq, BACKEND_VBSU);
 	blk->base.mtx = &blk->mtx;
 
 	blk->vq.qsize = VIRTIO_BLK_RINGSZ;
diff --git a/devicemodel/hw/pci/virtio/virtio_console.c b/devicemodel/hw/pci/virtio/virtio_console.c
index 03376453f..e44db8fae 100644
--- a/devicemodel/hw/pci/virtio/virtio_console.c
+++ b/devicemodel/hw/pci/virtio/virtio_console.c
@@ -828,7 +828,7 @@ virtio_console_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 					"error %d!\n", rc));
 
 	virtio_linkup(&console->base, &virtio_console_ops, console, dev,
-		console->queues);
+		console->queues, BACKEND_VBSU);
 	console->base.mtx = &console->mtx;
 	console->base.device_caps = VIRTIO_CONSOLE_S_HOSTCAPS;
 
diff --git a/devicemodel/hw/pci/virtio/virtio_coreu.c b/devicemodel/hw/pci/virtio/virtio_coreu.c
index f30c2728e..05ebf1c57 100644
--- a/devicemodel/hw/pci/virtio/virtio_coreu.c
+++ b/devicemodel/hw/pci/virtio/virtio_coreu.c
@@ -281,7 +281,7 @@ virtio_coreu_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 
 	DPRINTF(("vcoreu init: using VBS-U...\n"));
 	virtio_linkup(&vcoreu->base, &virtio_coreu_ops,
-			vcoreu, dev, vcoreu->queues);
+			vcoreu, dev, vcoreu->queues, BACKEND_VBSU);
 	vcoreu->base.mtx = &vcoreu->mtx;
 
 	vcoreu->queues[0].qsize  = VIRTIO_COREU_RINGSZ;
diff --git a/devicemodel/hw/pci/virtio/virtio_hdcp.c b/devicemodel/hw/pci/virtio/virtio_hdcp.c
index e2cfa9e80..6ce00fc56 100644
--- a/devicemodel/hw/pci/virtio/virtio_hdcp.c
+++ b/devicemodel/hw/pci/virtio/virtio_hdcp.c
@@ -413,7 +413,7 @@ virtio_hdcp_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 
 	DPRINTF(("vhdcp init: using VBS-U...\n"));
 	virtio_linkup(&vhdcp->base, &virtio_hdcp_ops,
-			vhdcp, dev, vhdcp->queues);
+			vhdcp, dev, vhdcp->queues, BACKEND_VBSU);
 	vhdcp->base.mtx = &vhdcp->mtx;
 
 	vhdcp->queues[0].qsize  = VIRTIO_HDCP_RINGSZ;
diff --git a/devicemodel/hw/pci/virtio/virtio_hyper_dmabuf.c b/devicemodel/hw/pci/virtio/virtio_hyper_dmabuf.c
index 12fabb3b9..04294e272 100644
--- a/devicemodel/hw/pci/virtio/virtio_hyper_dmabuf.c
+++ b/devicemodel/hw/pci/virtio/virtio_hyper_dmabuf.c
@@ -299,7 +299,8 @@ virtio_hyper_dmabuf_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 		      &virtio_hyper_dmabuf_ops_k,
 		      hyper_dmabuf,
 		      dev,
-		      hyper_dmabuf->vq);
+		      hyper_dmabuf->vq,
+		      BACKEND_VBSK);
 
 	rc = virtio_hyper_dmabuf_k_init();
 	if (rc < 0) {
diff --git a/devicemodel/hw/pci/virtio/virtio_input.c b/devicemodel/hw/pci/virtio/virtio_input.c
index e93c4d4ba..fc284d86c 100644
--- a/devicemodel/hw/pci/virtio/virtio_input.c
+++ b/devicemodel/hw/pci/virtio/virtio_input.c
@@ -640,7 +640,7 @@ virtio_input_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 		goto fail;
 	}
 
-	virtio_linkup(&vi->base, &virtio_input_ops, vi, dev, vi->queues);
+	virtio_linkup(&vi->base, &virtio_input_ops, vi, dev, vi->queues, BACKEND_VBSU);
 	vi->base.mtx = &vi->mtx;
 	vi->base.device_caps = VIRTIO_INPUT_S_HOSTCAPS;
 
diff --git a/devicemodel/hw/pci/virtio/virtio_ipu.c b/devicemodel/hw/pci/virtio/virtio_ipu.c
index ef3214823..e4b1f7bf0 100644
--- a/devicemodel/hw/pci/virtio/virtio_ipu.c
+++ b/devicemodel/hw/pci/virtio/virtio_ipu.c
@@ -320,7 +320,8 @@ virtio_ipu_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 		      &virtio_ipu_ops_k,
 		      ipu,
 		      dev,
-		      ipu->vq);
+		      ipu->vq,
+		      BACKEND_VBSK);
 
 	rc = virtio_ipu_k_init(ipu);
 	if (rc < 0) {
diff --git a/devicemodel/hw/pci/virtio/virtio_mei.c b/devicemodel/hw/pci/virtio/virtio_mei.c
index f50c587a1..89eb0a1b4 100644
--- a/devicemodel/hw/pci/virtio/virtio_mei.c
+++ b/devicemodel/hw/pci/virtio/virtio_mei.c
@@ -1577,7 +1577,7 @@ vmei_notify_tx(void *data, struct virtio_vq_info *vq)
 
 	pthread_mutex_lock(&vmei->tx_mutex);
 	DPRINTF("TX: New OUT buffer available!\n");
-	vq->used->flags &= ~ACRN_VRING_USED_F_NO_NOTIFY;
+	vq_clear_used_ring_flags(&vmei->base, vq);
 	pthread_mutex_unlock(&vmei->tx_mutex);
 }
 
@@ -1870,7 +1870,7 @@ static void *vmei_rx_thread(void *param)
 	while (vmei->status != VMEI_STST_DEINIT) {
 		/* note - rx mutex is locked here */
 		while (vq_ring_ready(vq)) {
-			vq->used->flags &= ~ACRN_VRING_USED_F_NO_NOTIFY;
+			vq_clear_used_ring_flags(&vmei->base, vq);
 			mb();
 			if (vq_has_descs(vq) &&
 			    vmei->rx_need_sched &&
@@ -2255,7 +2255,7 @@ init:
 		goto fail;
 	}
 
-	virtio_linkup(&vmei->base, &virtio_mei_ops, vmei, dev, vmei->vqs);
+	virtio_linkup(&vmei->base, &virtio_mei_ops, vmei, dev, vmei->vqs, BACKEND_VBSU);
 	vmei->base.mtx = &vmei->mutex;
 
 	for (i = 0; i < VMEI_VQ_NUM; i++)
diff --git a/devicemodel/hw/pci/virtio/virtio_net.c b/devicemodel/hw/pci/virtio/virtio_net.c
index 32fd085ca..e33dc357f 100644
--- a/devicemodel/hw/pci/virtio/virtio_net.c
+++ b/devicemodel/hw/pci/virtio/virtio_net.c
@@ -555,7 +555,7 @@ virtio_net_tx_thread(void *param)
 	for (;;) {
 		/* note - tx mutex is locked here */
 		while (net->resetting || !vq_has_descs(vq)) {
-			vq->used->flags &= ~ACRN_VRING_USED_F_NO_NOTIFY;
+			vq_clear_used_ring_flags(&net->base, vq);
 			/* memory barrier */
 			mb();
 			if (!net->resetting && vq_has_descs(vq))
@@ -724,7 +724,7 @@ virtio_net_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 	char nstr[80];
 	char tname[MAXCOMLEN + 1];
 	struct virtio_net *net;
-	char *devname;
+	char *devname = NULL;
 	char *vtopts;
 	char *opt;
 	int mac_provided;
@@ -751,25 +751,10 @@ virtio_net_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 		DPRINTF(("virtio_net: pthread_mutex_init failed with "
 			"error %d!\n", rc));
 
-	virtio_linkup(&net->base, &virtio_net_ops, net, dev, net->queues);
-	net->base.mtx = &net->mtx;
-	net->base.device_caps = VIRTIO_NET_S_HOSTCAPS;
-
-	net->queues[VIRTIO_NET_RXQ].qsize = VIRTIO_NET_RINGSZ;
-	net->queues[VIRTIO_NET_RXQ].notify = virtio_net_ping_rxq;
-	net->queues[VIRTIO_NET_TXQ].qsize = VIRTIO_NET_RINGSZ;
-	net->queues[VIRTIO_NET_TXQ].notify = virtio_net_ping_txq;
-#ifdef notyet
-	net->queues[VIRTIO_NET_CTLQ].qsize = VIRTIO_NET_RINGSZ;
-	net->queues[VIRTIO_NET_CTLQ].notify = virtio_net_ping_ctlq;
-#endif
-
 	/*
-	 * Attempt to open the tap device and read the MAC address
-	 * if specified
+	 * Read the MAC address if specified
 	 */
 	mac_provided = 0;
-	net->tapfd = -1;
 	net->vhost_net = NULL;
 	if (opts != NULL) {
 		int err;
@@ -795,14 +780,38 @@ virtio_net_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 				mac_provided = 1;
 			}
 		}
-
-		if (strncmp(devname, "tap", 3) == 0 ||
-		    strncmp(devname, "vmnet", 5) == 0)
-			virtio_net_tap_setup(net, devname);
-
-		free(devname);
 	}
 
+	virtio_linkup(&net->base, &virtio_net_ops, net, dev, net->queues,
+		      net->use_vhost ? BACKEND_VHOST : BACKEND_VBSU);
+	net->base.mtx = &net->mtx;
+	net->base.device_caps = VIRTIO_NET_S_HOSTCAPS;
+
+	net->queues[VIRTIO_NET_RXQ].qsize = VIRTIO_NET_RINGSZ;
+	net->queues[VIRTIO_NET_RXQ].notify = virtio_net_ping_rxq;
+	net->queues[VIRTIO_NET_TXQ].qsize = VIRTIO_NET_RINGSZ;
+	net->queues[VIRTIO_NET_TXQ].notify = virtio_net_ping_txq;
+#ifdef notyet
+	net->queues[VIRTIO_NET_CTLQ].qsize = VIRTIO_NET_RINGSZ;
+	net->queues[VIRTIO_NET_CTLQ].notify = virtio_net_ping_ctlq;
+#endif
+
+	/*
+	 * Attempt to open the tap device
+	 */
+	net->tapfd = -1;
+
+	if (!devname) {
+		WPRINTF(("virtio_net: devname NULL\n"));
+		return -1;
+	}
+
+	if (strncmp(devname, "tap", 3) == 0 ||
+	    strncmp(devname, "vmnet", 5) == 0)
+		virtio_net_tap_setup(net, devname);
+
+	free(devname);
+
 	/*
 	 * The default MAC address is the standard NetApp OUI of 00-a0-98,
 	 * followed by an MD5 of the PCI slot/func number and dev name
diff --git a/devicemodel/hw/pci/virtio/virtio_rnd.c b/devicemodel/hw/pci/virtio/virtio_rnd.c
index cfddcd0c8..2b30f6ca8 100644
--- a/devicemodel/hw/pci/virtio/virtio_rnd.c
+++ b/devicemodel/hw/pci/virtio/virtio_rnd.c
@@ -404,7 +404,7 @@ virtio_rnd_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 	if (rnd->vbs_k.status == VIRTIO_DEV_PRE_INIT) {
 		DPRINTF(("%s: VBS-K option detected!\n", __func__));
 		virtio_linkup(&rnd->base, &virtio_rnd_ops_k,
-			      rnd, dev, &rnd->vq);
+			      rnd, dev, &rnd->vq, BACKEND_VBSK);
 		rc = virtio_rnd_kernel_init(rnd);
 		if (rc < 0) {
 			WPRINTF(("virtio_rnd: VBS-K init failed,error %d!\n",
@@ -417,7 +417,7 @@ virtio_rnd_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 	if (rnd->vbs_k.status == VIRTIO_DEV_INITIAL ||
 	    rnd->vbs_k.status != VIRTIO_DEV_INIT_SUCCESS) {
 		DPRINTF(("%s: fallback to VBS-U...\n", __func__));
-		virtio_linkup(&rnd->base, &virtio_rnd_ops, rnd, dev, &rnd->vq);
+		virtio_linkup(&rnd->base, &virtio_rnd_ops, rnd, dev, &rnd->vq, BACKEND_VBSU);
 	}
 
 	rnd->base.mtx = &rnd->mtx;
diff --git a/devicemodel/hw/pci/virtio/virtio_rpmb.c b/devicemodel/hw/pci/virtio/virtio_rpmb.c
index e72498479..34a7dca67 100644
--- a/devicemodel/hw/pci/virtio/virtio_rpmb.c
+++ b/devicemodel/hw/pci/virtio/virtio_rpmb.c
@@ -512,7 +512,7 @@ virtio_rpmb_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
 		goto out;
 	}
 
-	virtio_linkup(&rpmb->base, &virtio_rpmb_ops, rpmb, dev, &rpmb->vq);
+	virtio_linkup(&rpmb->base, &virtio_rpmb_ops, rpmb, dev, &rpmb->vq, BACKEND_VBSU);
 
 	rpmb->base.mtx = &rpmb->mtx;
 	rpmb->vq.qsize = VIRTIO_RPMB_RINGSZ;
diff --git a/devicemodel/include/virtio.h b/devicemodel/include/virtio.h
index 3a07fe545..bcce4fafc 100644
--- a/devicemodel/include/virtio.h
+++ b/devicemodel/include/virtio.h
@@ -127,6 +127,7 @@
  */
 
 #include "types.h"
+#include "timer.h"
 
 /**
  * @brief virtio API
@@ -171,6 +172,14 @@ struct virtio_vring_used {
 /*	uint16_t	avail_event;	-- after N ring entries */
 } __attribute__((packed));
 
+enum {
+	BACKEND_UNKNOWN = 0,
+	BACKEND_VBSU,
+	BACKEND_VBSK,
+	BACKEND_VHOST,
+	BACKEND_MAX
+};
+
 /*
  * The address of any given virtual queue is determined by a single
  * Page Frame Number register.  The guest writes the PFN into the
@@ -520,6 +529,9 @@ struct virtio_base {
 	uint32_t device_feature_select;	/**< current selected device feature */
 	uint32_t driver_feature_select;	/**< current selected guest feature */
 	int cfg_coff;			/**< PCI cfg access capability offset */
+	int backend_type;               /**< VBSU, VBSK or VHOST */
+	struct acrn_timer polling_timer; /**< timer for polling mode */
+	int polling_in_progress;        /**< The polling status */
 };
 
 #define	VIRTIO_BASE_LOCK(vb)					\
@@ -703,12 +715,23 @@ struct iovec;
  * @param pci_virtio_dev Pointer to instance of certain virtio device.
  * @param dev Pointer to struct pci_vdev which emulates a PCI device.
  * @param queues Pointer to struct virtio_vq_info, normally an array.
+ * @param backend_type can be VBSU, VBSK or VHOST
  *
  * @return None
  */
 void virtio_linkup(struct virtio_base *base, struct virtio_ops *vops,
 		   void *pci_virtio_dev, struct pci_vdev *dev,
-		   struct virtio_vq_info *queues);
+		   struct virtio_vq_info *queues,
+		   int backend_type);
+
+/**
+ * @brief Get the virtio poll parameters
+ *
+ * @param optarg Pointer to parameters string.
+ *
+ * @return fail -1 success 0
+ */
+int acrn_parse_virtio_poll_interval(const char *optarg);
 
 /**
  * @brief Initialize MSI-X vector capabilities if we're to use MSI-X,
@@ -816,6 +839,20 @@ void vq_relchain(struct virtio_vq_info *vq, uint16_t idx, uint32_t iolen);
  */
 void vq_endchains(struct virtio_vq_info *vq, int used_all_avail);
 
+/**
+ * @brief Helper function for clearing used ring flags.
+ *
+ * Driver should always use this helper function to clear used ring flags.
+ * For virtio poll mode, in order to avoid trap, we should never really
+ * clear used ring flags.
+ *
+ * @param base Pointer to struct virtio_base.
+ * @param vq Pointer to struct virtio_vq_info.
+ *
+ * @return None
+ */
+void vq_clear_used_ring_flags(struct virtio_base *base, struct virtio_vq_info *vq);
+
 /**
  * @brief Handle PCI configuration space reads.
  *