[dpdk-dev] [PATCH v3 0/2] support inflight share memory protocol feature in vhost

[dpdk-dev] [PATCH v3 0/2] support inflight share memory protocol feature in vhost

[JinYu]

This patch introduces two new messages VHOST_USER_GET_INFLIGHT_FD
and VHOST_USER_SET_INFLIGHT_FD to support transferring a shared
buffer between qemu and backend. And the example code shows how
these APIs work.

JinYu (2):
  vhost: support inflight share memory protocol feature
  vhost: Add vhost-user-blk example which support inflight

 examples/vhost_blk/Makefile            |  67 +++
 examples/vhost_blk/blk.c               | 117 +++++
 examples/vhost_blk/blk_spec.h          |  95 ++++
 examples/vhost_blk/meson.build         |  20 +
 examples/vhost_blk/vhost_blk.c         | 589 +++++++++++++++++++++++++
 examples/vhost_blk/vhost_blk.h         |  96 ++++
 examples/vhost_blk/vhost_blk_compat.c  | 193 ++++++++
 lib/librte_vhost/rte_vhost.h           |  89 ++++
 lib/librte_vhost/rte_vhost_version.map |   4 +
 lib/librte_vhost/vhost.c               | 163 ++++++-
 lib/librte_vhost/vhost.h               |  32 ++
 lib/librte_vhost/vhost_user.c          | 316 +++++++++++++
 lib/librte_vhost/vhost_user.h          |  13 +-
 13 files changed, 1792 insertions(+), 2 deletions(-)
 create mode 100644 examples/vhost_blk/Makefile
 create mode 100644 examples/vhost_blk/blk.c
 create mode 100644 examples/vhost_blk/blk_spec.h
 create mode 100644 examples/vhost_blk/meson.build
 create mode 100644 examples/vhost_blk/vhost_blk.c
 create mode 100644 examples/vhost_blk/vhost_blk.h
 create mode 100644 examples/vhost_blk/vhost_blk_compat.c

-- 
2.17.2

[dpdk-dev] [PATCH v3 1/2] vhost: support inflight share memory protocol feature

[JinYu]

This patch introduces two new messages VHOST_USER_GET_INFLIGHT_FD
and VHOST_USER_SET_INFLIGHT_FD to support transferring a shared
buffer between qemu and backend.

Firstly, qemu uses VHOST_USER_GET_INFLIGHT_FD to get the
shared buffer from backend. Then qemu should send it back
through VHOST_USER_SET_INFLIGHT_FD each time we start vhost-user.

This shared buffer is used to process inflight I/O when backend
reconnect.

Signed-off-by: Lin Li <lilin24@baidu.com>
Signed-off-by: Xun Ni <nixun@baidu.com>
Signed-off-by: Yu Zhang <zhangyu31@baidu.com>
Signed-off-by: Jin Yu <jin.yu@intel.com>
---
v1 - specify the APIs are split-ring only
v2 - fix APIs and judge split or packed
v3 - Add rte_vhost_ prefix and fix one issue.
---
 lib/librte_vhost/rte_vhost.h           |  89 +++++++
 lib/librte_vhost/rte_vhost_version.map |   4 +
 lib/librte_vhost/vhost.c               | 163 ++++++++++++-
 lib/librte_vhost/vhost.h               |  32 +++
 lib/librte_vhost/vhost_user.c          | 316 +++++++++++++++++++++++++
 lib/librte_vhost/vhost_user.h          |  13 +-
 6 files changed, 615 insertions(+), 2 deletions(-)

diff --git a/lib/librte_vhost/rte_vhost.h b/lib/librte_vhost/rte_vhost.h
index 0226b3eff..01a875abd 100644
--- a/lib/librte_vhost/rte_vhost.h
+++ b/lib/librte_vhost/rte_vhost.h
@@ -71,6 +71,10 @@ extern "C" {
 #define VHOST_USER_PROTOCOL_F_HOST_NOTIFIER 11
 #endif
 
+#ifndef VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD
+#define VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD 12
+#endif
+
 /** Indicate whether protocol features negotiation is supported. */
 #ifndef VHOST_USER_F_PROTOCOL_FEATURES
 #define VHOST_USER_F_PROTOCOL_FEATURES	30
@@ -98,6 +102,25 @@ struct rte_vhost_memory {
 	struct rte_vhost_mem_region regions[];
 };
 
+struct rte_vhost_resubmit_desc {
+	uint16_t index;
+	uint64_t counter;
+};
+
+struct rte_vhost_resubmit_info {
+	struct rte_vhost_resubmit_desc	*resubmit_list;
+	uint16_t		resubmit_num;
+};
+
+struct rte_vhost_ring_inflight_split {
+	union {
+		struct inflight_info_split *inflight_split;
+		/* TODO */
+	};
+
+	struct rte_vhost_resubmit_info *resubmit_inflight_split;
+};
+
 struct rte_vhost_vring {
 	struct vring_desc	*desc;
 	struct vring_avail	*avail;
@@ -603,6 +626,22 @@ uint16_t rte_vhost_dequeue_burst(int vid, uint16_t queue_id,
  */
 int rte_vhost_get_mem_table(int vid, struct rte_vhost_memory **mem);
 
+/**
+ * Get guest inflight vring info, including inflight ring and resubmit list.
+ *
+ * @param vid
+ *  vhost device ID
+ * @param vring_idx
+ *  vring index
+ * @param vring
+ *  the structure to hold the requested inflight vring info
+ * @return
+ *  0 on success, -1 on failure
+ */
+int __rte_experimental
+rte_vhost_get_vhost_ring_inflight_split(int vid, uint16_t vring_idx,
+			      struct rte_vhost_ring_inflight_split *vring);
+
 /**
  * Get guest vring info, including the vring address, vring size, etc.
  *
@@ -631,6 +670,56 @@ int rte_vhost_get_vhost_vring(int vid, uint16_t vring_idx,
  */
 int rte_vhost_vring_call(int vid, uint16_t vring_idx);
 
+/**
+ * set inflight flag for a desc.
+ *
+ * @param vid
+ *  vhost device ID
+ * @param vring_idx
+ *  vring index
+ * @param idx
+ *  inflight entry index
+ * @return
+ *  0 on success, -1 on failure
+ */
+int __rte_experimental
+rte_vhost_set_inflight_desc_split(int vid, uint16_t vring_idx,
+		uint16_t idx);
+
+/**
+ * clear inflight flag for a desc.
+ *
+ * @param vid
+ *  vhost device ID
+ * @param vring_idx
+ *  vring index
+ * @param last_used_idx
+ *  next free used_idx
+ * @param idx
+ *  inflight entry index
+ * @return
+ *  0 on success, -1 on failure
+ */
+int __rte_experimental
+rte_vhost_clr_inflight_desc_split(int vid, uint16_t vring_idx,
+		uint16_t last_used_idx, uint16_t idx);
+
+/**
+ * set last inflight io index.
+ *
+ * @param vid
+ *  vhost device ID
+ * @param vring_idx
+ *  vring index
+ * @param idx
+ *  inflight entry index
+ * @return
+ *  0 on success, -1 on failure
+ */
+int __rte_experimental
+rte_vhost_set_last_inflight_io_split(int vid, uint16_t vring_idx,
+		uint16_t idx);
+
 /**
  * Get vhost RX queue avail count.
  *
diff --git a/lib/librte_vhost/rte_vhost_version.map b/lib/librte_vhost/rte_vhost_version.map
index 5f1d4a75c..57a166f4f 100644
--- a/lib/librte_vhost/rte_vhost_version.map
+++ b/lib/librte_vhost/rte_vhost_version.map
@@ -87,4 +87,8 @@ EXPERIMENTAL {
 	rte_vdpa_relay_vring_used;
 	rte_vhost_extern_callback_register;
 	rte_vhost_driver_set_protocol_features;
+	rte_vhost_set_inflight_desc_split;
+	rte_vhost_clr_inflight_desc_split;
+	rte_vhost_set_last_inflight_io_split;
+	rte_vhost_get_vhost_ring_inflight_split;
 };
diff --git a/lib/librte_vhost/vhost.c b/lib/librte_vhost/vhost.c
index 981837b5d..dac641dc0 100644
--- a/lib/librte_vhost/vhost.c
+++ b/lib/librte_vhost/vhost.c
@@ -242,6 +242,30 @@ cleanup_vq(struct vhost_virtqueue *vq, int destroy)
 		close(vq->kickfd);
 }
 
+void
+cleanup_vq_inflight_split(struct virtio_net *dev, struct vhost_virtqueue *vq)
+{
+	if (!(dev->protocol_features &
+		(1ULL << VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD)))
+		return;
+
+	if (vq_is_packed(dev)) {
+		/* TODO */
+		RTE_LOG(ERR, VHOST_CONFIG, "function only be used for split inflight.\n");
+	} else {
+		if (vq->inflight_split)
+			vq->inflight_split = NULL;
+		if (vq->resubmit_inflight_split) {
+			if (vq->resubmit_inflight_split->resubmit_list) {
+				free(vq->resubmit_inflight_split->resubmit_list);
+				vq->resubmit_inflight_split->resubmit_list = NULL;
+			}
+			free(vq->resubmit_inflight_split);
+			vq->resubmit_inflight_split = NULL;
+		}
+	}
+}
+
 /*
  * Unmap any memory, close any file descriptors and
  * free any memory owned by a device.
@@ -253,8 +277,10 @@ cleanup_device(struct virtio_net *dev, int destroy)
 
 	vhost_backend_cleanup(dev);
 
-	for (i = 0; i < dev->nr_vring; i++)
+	for (i = 0; i < dev->nr_vring; i++) {
 		cleanup_vq(dev->virtqueue[i], destroy);
+		cleanup_vq_inflight_split(dev, dev->virtqueue[i]);
+	}
 }
 
 void
@@ -726,6 +752,40 @@ rte_vhost_get_mem_table(int vid, struct rte_vhost_memory **mem)
 	return 0;
 }
 
+int
+rte_vhost_get_vhost_ring_inflight_split(int vid, uint16_t vring_idx,
+			  struct rte_vhost_ring_inflight_split *vring)
+{
+	struct virtio_net *dev;
+	struct vhost_virtqueue *vq;
+	struct inflight_info_split *inflight_vring;
+
+	dev = get_device(vid);
+	if (dev == NULL || vring == NULL)
+		return -1;
+
+	if (vring_idx >= VHOST_MAX_VRING)
+		return -1;
+
+	vq = dev->virtqueue[vring_idx];
+	if (!vq)
+		return -1;
+
+	inflight_vring = vq->inflight_split;
+	if (!inflight_vring)
+		return -1;
+
+	if (vq_is_packed(dev)) {
+		/* TODO */
+		RTE_LOG(ERR, VHOST_CONFIG, "Can't be packed inflight ring.\n");
+	} else {
+		vring->inflight_split = inflight_vring;
+		vring->resubmit_inflight_split = vq->resubmit_inflight_split;
+	}
+
+	return 0;
+}
+
 int
 rte_vhost_get_vhost_vring(int vid, uint16_t vring_idx,
 			  struct rte_vhost_vring *vring)
@@ -781,6 +841,107 @@ rte_vhost_vring_call(int vid, uint16_t vring_idx)
 	return 0;
 }
 
+int
+rte_vhost_set_inflight_desc_split(int vid, uint16_t vring_idx, uint16_t idx)
+{
+	struct virtio_net *dev;
+	struct vhost_virtqueue *vq;
+
+	dev = get_device(vid);
+	if (unlikely(!dev))
+		return -1;
+
+	if (unlikely(!(dev->protocol_features &
+		(1ULL << VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD))))
+		return 0;
+
+	if (unlikely(vq_is_packed(dev)))
+		return -1;
+
+	if (unlikely(vring_idx >= VHOST_MAX_VRING))
+		return -1;
+
+	vq = dev->virtqueue[vring_idx];
+	if (unlikely(!vq))
+		return -1;
+
+	if (unlikely(!vq->inflight_split))
+		return -1;
+
+	vq->inflight_split->desc[idx].counter = vq->global_counter++;
+	vq->inflight_split->desc[idx].inflight = 1;
+	return 0;
+}
+
+int
+rte_vhost_clr_inflight_desc_split(int vid, uint16_t vring_idx,
+	uint16_t last_used_idx, uint16_t idx)
+{
+	struct virtio_net *dev;
+	struct vhost_virtqueue *vq;
+
+	dev = get_device(vid);
+	if (unlikely(!dev))
+		return -1;
+
+	if (unlikely(!(dev->protocol_features &
+		(1ULL << VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD))))
+		return 0;
+
+	if (unlikely(vq_is_packed(dev)))
+		return -1;
+
+	if (unlikely(vring_idx >= VHOST_MAX_VRING))
+		return -1;
+
+	vq = dev->virtqueue[vring_idx];
+	if (unlikely(!vq))
+		return -1;
+
+	if (unlikely(!vq->inflight_split))
+		return -1;
+
+	rte_compiler_barrier();
+
+	vq->inflight_split->desc[idx].inflight = 0;
+
+	rte_compiler_barrier();
+
+	vq->inflight_split->used_idx = last_used_idx;
+	return 0;
+}
+
+int
+rte_vhost_set_last_inflight_io_split(int vid, uint16_t vring_idx, uint16_t idx)
+{
+	struct virtio_net *dev;
+	struct vhost_virtqueue *vq;
+
+	dev = get_device(vid);
+	if (unlikely(!dev))
+		return -1;
+
+	if (unlikely(!(dev->protocol_features &
+		(1ULL << VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD))))
+		return 0;
+
+	if (unlikely(vq_is_packed(dev)))
+		return -1;
+
+	if (unlikely(vring_idx >= VHOST_MAX_VRING))
+		return -1;
+
+	vq = dev->virtqueue[vring_idx];
+	if (!vq)
+		return -1;
+
+	if (unlikely(!vq->inflight_split))
+		return -1;
+
+	vq->inflight_split->last_inflight_io = idx;
+	return 0;
+}
+
 uint16_t
 rte_vhost_avail_entries(int vid, uint16_t queue_id)
 {
diff --git a/lib/librte_vhost/vhost.h b/lib/librte_vhost/vhost.h
index 884befa85..1770754eb 100644
--- a/lib/librte_vhost/vhost.h
+++ b/lib/librte_vhost/vhost.h
@@ -88,6 +88,22 @@ struct vring_used_elem_packed {
 	uint32_t count;
 };
 
+struct inflight_desc_split {
+	uint8_t		inflight;
+	uint8_t		padding[5];
+	uint16_t	next;
+	uint64_t	counter;
+};
+
+struct inflight_info_split {
+	uint64_t		features;
+	uint16_t		version;
+	uint16_t		desc_num;
+	uint16_t		last_inflight_io;
+	uint16_t		used_idx;
+	struct inflight_desc_split desc[0];
+};
+
 /**
  * Structure contains variables relevant to RX/TX virtqueues.
  */
@@ -128,6 +144,14 @@ struct vhost_virtqueue {
 	/* Physical address of used ring, for logging */
 	uint64_t		log_guest_addr;
 
+	/* inflight share memory info */
+	union {
+		struct inflight_info_split *inflight_split;
+		/* TODO */
+	};
+	struct rte_vhost_resubmit_info *resubmit_inflight_split;
+	uint64_t		global_counter;
+
 	uint16_t		nr_zmbuf;
 	uint16_t		zmbuf_size;
 	uint16_t		last_zmbuf_idx;
@@ -286,6 +310,12 @@ struct guest_page {
 	uint64_t size;
 };
 
+struct inflight_mem_info {
+	int		fd;
+	void		*addr;
+	uint64_t	size;
+};
+
 /**
  * Device structure contains all configuration information relating
  * to the device.
@@ -303,6 +333,7 @@ struct virtio_net {
 	uint32_t		nr_vring;
 	int			dequeue_zero_copy;
 	struct vhost_virtqueue	*virtqueue[VHOST_MAX_QUEUE_PAIRS * 2];
+	struct inflight_mem_info inflight_info;
 #define IF_NAME_SZ (PATH_MAX > IFNAMSIZ ? PATH_MAX : IFNAMSIZ)
 	char			ifname[IF_NAME_SZ];
 	uint64_t		log_size;
@@ -467,6 +498,7 @@ void vhost_destroy_device(int);
 void vhost_destroy_device_notify(struct virtio_net *dev);
 
 void cleanup_vq(struct vhost_virtqueue *vq, int destroy);
+void cleanup_vq_inflight_split(struct virtio_net *dev, struct vhost_virtqueue *vq);
 void free_vq(struct virtio_net *dev, struct vhost_virtqueue *vq);
 
 int alloc_vring_queue(struct virtio_net *dev, uint32_t vring_idx);
diff --git a/lib/librte_vhost/vhost_user.c b/lib/librte_vhost/vhost_user.c
index c9e29ece8..c76df099a 100644
--- a/lib/librte_vhost/vhost_user.c
+++ b/lib/librte_vhost/vhost_user.c
@@ -31,6 +31,8 @@
 #include <sys/stat.h>
 #include <sys/syscall.h>
 #include <assert.h>
+#include <sys/syscall.h>
+#include <asm/unistd.h>
 #ifdef RTE_LIBRTE_VHOST_NUMA
 #include <numaif.h>
 #endif
@@ -49,6 +51,15 @@
 #define VIRTIO_MIN_MTU 68
 #define VIRTIO_MAX_MTU 65535
 
+#define INFLIGHT_ALIGNMENT	64
+#define INFLIGHT_VERSION	0xabcd
+#define VIRTQUEUE_MAX_SIZE	1024
+
+#define CLOEXEC	0x0001U
+
+#define ALIGN_DOWN(n, m) ((n) / (m) * (m))
+#define ALIGN_UP(n, m) ALIGN_DOWN((n) + (m) - 1, (m))
+
 static const char *vhost_message_str[VHOST_USER_MAX] = {
 	[VHOST_USER_NONE] = "VHOST_USER_NONE",
 	[VHOST_USER_GET_FEATURES] = "VHOST_USER_GET_FEATURES",
@@ -78,6 +89,8 @@ static const char *vhost_message_str[VHOST_USER_MAX] = {
 	[VHOST_USER_POSTCOPY_ADVISE]  = "VHOST_USER_POSTCOPY_ADVISE",
 	[VHOST_USER_POSTCOPY_LISTEN]  = "VHOST_USER_POSTCOPY_LISTEN",
 	[VHOST_USER_POSTCOPY_END]  = "VHOST_USER_POSTCOPY_END",
+	[VHOST_USER_GET_INFLIGHT_FD] = "VHOST_USER_GET_INFLIGHT_FD",
+	[VHOST_USER_SET_INFLIGHT_FD] = "VHOST_USER_SET_INFLIGHT_FD",
 };
 
 static int send_vhost_reply(int sockfd, struct VhostUserMsg *msg);
@@ -160,6 +173,16 @@ vhost_backend_cleanup(struct virtio_net *dev)
 		dev->log_addr = 0;
 	}
 
+	if (dev->inflight_info.addr) {
+		munmap(dev->inflight_info.addr, dev->inflight_info.size);
+		dev->inflight_info.addr = NULL;
+	}
+
+	if (dev->inflight_info.fd > 0) {
+		close(dev->inflight_info.fd);
+		dev->inflight_info.fd = -1;
+	}
+
 	if (dev->slave_req_fd >= 0) {
 		close(dev->slave_req_fd);
 		dev->slave_req_fd = -1;
@@ -306,6 +329,7 @@ vhost_user_set_features(struct virtio_net **pdev, struct VhostUserMsg *msg,
 
 			dev->virtqueue[dev->nr_vring] = NULL;
 			cleanup_vq(vq, 1);
+			cleanup_vq_inflight_split(dev, vq);
 			free_vq(dev, vq);
 		}
 	}
@@ -1165,6 +1189,202 @@ virtio_is_ready(struct virtio_net *dev)
 	return 1;
 }
 
+static int mem_create(const char *name, unsigned int flags)
+{
+#ifdef __NR_memfd_create
+	return syscall(__NR_memfd_create, name, flags);
+#else
+	return -1;
+#endif
+}
+
+void *inflight_mem_alloc(const char *name, size_t size, int *fd)
+{
+	void *ptr;
+	int mfd = -1;
+	char fname[20] = "/tmp/memfd-XXXXXX";
+
+	*fd = -1;
+	mfd = mem_create(name, CLOEXEC);
+	if (mfd != -1) {
+		if (ftruncate(mfd, size) == -1) {
+			RTE_LOG(ERR, VHOST_CONFIG,
+					"ftruncate fail for alloc inflight buffer\n");
+			close(mfd);
+			return NULL;
+		}
+	} else {
+		mfd = mkstemp(fname);
+		unlink(fname);
+
+		if (mfd == -1) {
+			RTE_LOG(ERR, VHOST_CONFIG,
+					"mkstemp fail for alloc inflight buffer\n");
+			return NULL;
+		}
+
+		if (ftruncate(mfd, size) == -1) {
+			RTE_LOG(ERR, VHOST_CONFIG,
+					"ftruncate fail for alloc inflight buffer\n");
+			close(mfd);
+			return NULL;
+		}
+	}
+
+	ptr = mmap(0, size, PROT_READ | PROT_WRITE, MAP_SHARED, mfd, 0);
+	if (ptr == MAP_FAILED) {
+		RTE_LOG(ERR, VHOST_CONFIG,
+				"mmap fail for alloc inflight buffer\n");
+		close(mfd);
+		return NULL;
+	}
+
+	*fd = mfd;
+	return ptr;
+}
+
+static uint32_t get_pervq_shm_size_split(uint16_t queue_size)
+{
+	return ALIGN_UP(sizeof(struct inflight_desc_split) * queue_size +
+		sizeof(uint64_t) * 1 + sizeof(uint16_t) * 4, INFLIGHT_ALIGNMENT);
+}
+
+static int
+vhost_user_get_inflight_fd(struct virtio_net **pdev, VhostUserMsg *msg,
+		int main_fd __rte_unused)
+{
+	int fd;
+	uint64_t mmap_size;
+	void *addr;
+	uint16_t num_queues, queue_size;
+	struct virtio_net *dev = *pdev;
+
+	if (msg->size != sizeof(msg->payload.inflight)) {
+		RTE_LOG(ERR, VHOST_CONFIG,
+			"Invalid get_inflight_fd message size is %d",
+			msg->size);
+		return RTE_VHOST_MSG_RESULT_ERR;
+	}
+
+	num_queues = msg->payload.inflight.num_queues;
+	queue_size = msg->payload.inflight.queue_size;
+
+	RTE_LOG(INFO, VHOST_CONFIG, "get_inflight_fd num_queues: %u\n",
+			msg->payload.inflight.num_queues);
+	RTE_LOG(INFO, VHOST_CONFIG, "get_inflight_fd queue_size: %u\n",
+			msg->payload.inflight.queue_size);
+
+	if (vq_is_packed(dev)) {
+		/* TODO */
+		RTE_LOG(ERR, VHOST_CONFIG,
+			"Don't support the packed ring\n");
+		return RTE_VHOST_MSG_RESULT_ERR;
+	}
+	mmap_size = num_queues * get_pervq_shm_size_split(queue_size);
+
+	addr = inflight_mem_alloc("vhost-inflight", mmap_size, &fd);
+	if (!addr) {
+		RTE_LOG(ERR, VHOST_CONFIG, "Failed to alloc vhost inflight area");
+			msg->payload.inflight.mmap_size = 0;
+		return RTE_VHOST_MSG_RESULT_ERR;
+	}
+	memset(addr, 0, mmap_size);
+
+	dev->inflight_info.addr = addr;
+	dev->inflight_info.size = msg->payload.inflight.mmap_size = mmap_size;
+	dev->inflight_info.fd = msg->fds[0] = fd;
+	msg->payload.inflight.mmap_offset = 0;
+	msg->fd_num = 1;
+
+	RTE_LOG(INFO, VHOST_CONFIG,
+			"send inflight mmap_size: %lu\n",
+			msg->payload.inflight.mmap_size);
+	RTE_LOG(INFO, VHOST_CONFIG,
+			"send inflight mmap_offset: %lu\n",
+			msg->payload.inflight.mmap_offset);
+	RTE_LOG(INFO, VHOST_CONFIG,
+			"send inflight fd: %d\n", msg->fds[0]);
+
+	return RTE_VHOST_MSG_RESULT_REPLY;
+}
+
+static int
+vhost_user_set_inflight_fd(struct virtio_net **pdev, VhostUserMsg *msg,
+		int main_fd __rte_unused)
+{
+	int fd, i;
+	uint64_t mmap_size, mmap_offset;
+	uint16_t num_queues, queue_size;
+	uint32_t pervq_inflight_size;
+	void *addr;
+	struct vhost_virtqueue *vq;
+	struct virtio_net *dev = *pdev;
+
+	fd = msg->fds[0];
+	if (msg->size != sizeof(msg->payload.inflight) || fd < 0) {
+		RTE_LOG(ERR, VHOST_CONFIG, "Invalid set_inflight_fd message size is %d,fd is %d\n",
+			msg->size, fd);
+		return RTE_VHOST_MSG_RESULT_ERR;
+	}
+
+	mmap_size = msg->payload.inflight.mmap_size;
+	mmap_offset = msg->payload.inflight.mmap_offset;
+	num_queues = msg->payload.inflight.num_queues;
+	queue_size = msg->payload.inflight.queue_size;
+
+	if (vq_is_packed(dev)) {
+		/* TODO */
+		RTE_LOG(ERR, VHOST_CONFIG,
+			"Don't support the packed ring\n");
+		return RTE_VHOST_MSG_RESULT_ERR;
+	}
+	pervq_inflight_size = get_pervq_shm_size_split(queue_size);
+
+	RTE_LOG(INFO, VHOST_CONFIG,
+		"set_inflight_fd mmap_size: %lu\n", mmap_size);
+	RTE_LOG(INFO, VHOST_CONFIG,
+		"set_inflight_fd mmap_offset: %lu\n", mmap_offset);
+	RTE_LOG(INFO, VHOST_CONFIG,
+		"set_inflight_fd num_queues: %u\n", num_queues);
+	RTE_LOG(INFO, VHOST_CONFIG,
+		"set_inflight_fd queue_size: %u\n", queue_size);
+	RTE_LOG(INFO, VHOST_CONFIG,
+		"set_inflight_fd fd: %d\n", fd);
+	RTE_LOG(INFO, VHOST_CONFIG,
+		"set_inflight_fd pervq_inflight_size: %d\n",
+		pervq_inflight_size);
+
+	if (dev->inflight_info.addr)
+		munmap(dev->inflight_info.addr, dev->inflight_info.size);
+
+	addr = mmap(0, mmap_size, PROT_READ | PROT_WRITE, MAP_SHARED,
+			fd, mmap_offset);
+	if (addr == MAP_FAILED) {
+		RTE_LOG(ERR, VHOST_CONFIG, "failed to mmap share memory.\n");
+		return RTE_VHOST_MSG_RESULT_ERR;
+	}
+
+	if (dev->inflight_info.fd)
+		close(dev->inflight_info.fd);
+
+	dev->inflight_info.fd = fd;
+	dev->inflight_info.addr = addr;
+	dev->inflight_info.size = mmap_size;
+
+	for (i = 0; i < num_queues; i++) {
+		vq = dev->virtqueue[i];
+		if (vq_is_packed(dev)) {
+			/* TODO */
+		} else {
+			vq->inflight_split = (struct inflight_info_split *)addr;
+			vq->inflight_split->desc_num = queue_size;
+		}
+		addr = (void *)((char *)addr + pervq_inflight_size);
+	}
+
+	return RTE_VHOST_MSG_RESULT_OK;
+}
+
 static int
 vhost_user_set_vring_call(struct virtio_net **pdev, struct VhostUserMsg *msg,
 			int main_fd __rte_unused)
@@ -1201,6 +1421,94 @@ static int vhost_user_set_vring_err(struct virtio_net **pdev __rte_unused,
 	return RTE_VHOST_MSG_RESULT_OK;
 }
 
+static int
+resubmit_desc_compare(const void *a, const void *b)
+{
+	const struct rte_vhost_resubmit_desc *desc0 = 
+		(const struct rte_vhost_resubmit_desc *)a;
+	const struct rte_vhost_resubmit_desc *desc1 = 
+		(const struct rte_vhost_resubmit_desc *)b;
+
+	if (desc1->counter > desc0->counter &&
+		(desc1->counter - desc0->counter) < VIRTQUEUE_MAX_SIZE * 2)
+		return 1;
+
+	return -1;
+}
+
+static int
+vhost_check_queue_inflights_split(struct virtio_net *dev, struct vhost_virtqueue *vq)
+{
+	struct vring_used *used = vq->used;
+	uint16_t i = 0;
+	uint16_t resubmit_num = 0;
+	struct rte_vhost_resubmit_info *resubmit = NULL;
+	struct inflight_info_split *inflight_split;
+
+	if (!(dev->protocol_features &
+		(1ULL << VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD)))
+		return RTE_VHOST_MSG_RESULT_OK;
+
+	if ((!vq->inflight_split))
+		return RTE_VHOST_MSG_RESULT_ERR;
+
+	if (!vq->inflight_split->version) {
+		vq->inflight_split->version = INFLIGHT_VERSION;
+		return RTE_VHOST_MSG_RESULT_OK;
+	}
+
+	inflight_split = vq->inflight_split;
+	vq->resubmit_inflight_split = NULL;
+	vq->global_counter = 0;
+
+	if (inflight_split->used_idx != used->idx) {
+		inflight_split->desc[inflight_split->last_inflight_io].inflight = 0;
+		rte_compiler_barrier();
+		inflight_split->used_idx = used->idx;
+	}
+
+	for (i = 0; i < inflight_split->desc_num; i++) {
+		if (inflight_split->desc[i].inflight == 1)
+			resubmit_num++;
+	}
+
+	vq->last_avail_idx += resubmit_num;
+
+	if (resubmit_num) {
+		resubmit  = calloc(1, sizeof(struct rte_vhost_resubmit_info));
+		if (!resubmit) {
+			RTE_LOG(ERR, VHOST_CONFIG, "Failed to allocate memory for resubmit info.\n");
+			return RTE_VHOST_MSG_RESULT_ERR;
+		}
+
+		resubmit->resubmit_list = calloc(resubmit_num,
+			sizeof(struct rte_vhost_resubmit_desc));
+		if (!resubmit->resubmit_list) {
+			RTE_LOG(ERR, VHOST_CONFIG, "Failed to allocate memory for inflight desc.\n");
+			return RTE_VHOST_MSG_RESULT_ERR;
+		}
+
+		for (i = 0; i < vq->inflight_split->desc_num; i++) {
+			if (vq->inflight_split->desc[i].inflight == 1) {
+				resubmit->resubmit_list[resubmit->resubmit_num].index = i;
+				resubmit->resubmit_list[resubmit->resubmit_num].counter =
+					inflight_split->desc[i].counter;
+				resubmit->resubmit_num++;
+			}
+		}
+
+		if (resubmit->resubmit_num > 1)
+			qsort(resubmit->resubmit_list, resubmit->resubmit_num,
+				sizeof(struct rte_vhost_resubmit_desc),
+				resubmit_desc_compare);
+
+		vq->global_counter = resubmit->resubmit_list[0].counter + 1;
+		vq->resubmit_inflight_split = resubmit;
+	}
+
+	return RTE_VHOST_MSG_RESULT_OK;
+}
+
 static int
 vhost_user_set_vring_kick(struct virtio_net **pdev, struct VhostUserMsg *msg,
 			int main_fd __rte_unused)
@@ -1242,6 +1550,12 @@ vhost_user_set_vring_kick(struct virtio_net **pdev, struct VhostUserMsg *msg,
 		close(vq->kickfd);
 	vq->kickfd = file.fd;
 
+	if (vhost_check_queue_inflights_split(dev, vq)) {
+		RTE_LOG(ERR, VHOST_CONFIG,
+			"Failed to inflights for vq: %d\n", file.index);
+		return RTE_VHOST_MSG_RESULT_ERR;
+	}
+
 	return RTE_VHOST_MSG_RESULT_OK;
 }
 
@@ -1762,6 +2076,8 @@ static vhost_message_handler_t vhost_message_handlers[VHOST_USER_MAX] = {
 	[VHOST_USER_POSTCOPY_ADVISE] = vhost_user_set_postcopy_advise,
 	[VHOST_USER_POSTCOPY_LISTEN] = vhost_user_set_postcopy_listen,
 	[VHOST_USER_POSTCOPY_END] = vhost_user_postcopy_end,
+	[VHOST_USER_GET_INFLIGHT_FD] = vhost_user_get_inflight_fd,
+	[VHOST_USER_SET_INFLIGHT_FD] = vhost_user_set_inflight_fd,
 };
 
 
diff --git a/lib/librte_vhost/vhost_user.h b/lib/librte_vhost/vhost_user.h
index 2a650fe4b..99a773910 100644
--- a/lib/librte_vhost/vhost_user.h
+++ b/lib/librte_vhost/vhost_user.h
@@ -54,7 +54,9 @@ typedef enum VhostUserRequest {
 	VHOST_USER_POSTCOPY_ADVISE = 28,
 	VHOST_USER_POSTCOPY_LISTEN = 29,
 	VHOST_USER_POSTCOPY_END = 30,
-	VHOST_USER_MAX = 31
+	VHOST_USER_GET_INFLIGHT_FD = 31,
+	VHOST_USER_SET_INFLIGHT_FD = 32,
+	VHOST_USER_MAX = 33
 } VhostUserRequest;
 
 typedef enum VhostUserSlaveRequest {
@@ -112,6 +114,13 @@ typedef struct VhostUserVringArea {
 	uint64_t offset;
 } VhostUserVringArea;
 
+typedef struct VhostUserInflight {
+	uint64_t mmap_size;
+	uint64_t mmap_offset;
+	uint16_t num_queues;
+	uint16_t queue_size;
+} VhostUserInflight;
+
 typedef struct VhostUserMsg {
 	union {
 		uint32_t master; /* a VhostUserRequest value */
@@ -131,6 +140,7 @@ typedef struct VhostUserMsg {
 		struct vhost_vring_addr addr;
 		VhostUserMemory memory;
 		VhostUserLog    log;
+		VhostUserInflight inflight;
 		struct vhost_iotlb_msg iotlb;
 		VhostUserCryptoSessionParam crypto_session;
 		VhostUserVringArea area;
@@ -148,6 +158,7 @@ typedef struct VhostUserMsg {
 /* vhost_user.c */
 int vhost_user_msg_handler(int vid, int fd);
 int vhost_user_iotlb_miss(struct virtio_net *dev, uint64_t iova, uint8_t perm);
+void *inflight_mem_alloc(const char *name, size_t size, int *fd);
 
 /* socket.c */
 int read_fd_message(int sockfd, char *buf, int buflen, int *fds, int max_fds,
-- 
2.17.2

[dpdk-dev] [PATCH v3 2/2] vhost: Add vhost-user-blk example which support inflight

[JinYu]

A vhost-user-blk example that support inflight feature. It uses the
new APIs that introduced in the first patch, so It can show how there
APIs work to support inflight feature.

Signed-off-by: JinYu <jin.yu@intel.com>
---
V1 - add the case.
V2 - add the rte_vhost prefix.
---
 examples/vhost_blk/Makefile           |  67 +++
 examples/vhost_blk/blk.c              | 117 +++++
 examples/vhost_blk/blk_spec.h         |  95 +++++
 examples/vhost_blk/meson.build        |  20 +
 examples/vhost_blk/vhost_blk.c        | 589 ++++++++++++++++++++++++++
 examples/vhost_blk/vhost_blk.h        |  96 +++++
 examples/vhost_blk/vhost_blk_compat.c | 193 +++++++++
 7 files changed, 1177 insertions(+)
 create mode 100644 examples/vhost_blk/Makefile
 create mode 100644 examples/vhost_blk/blk.c
 create mode 100644 examples/vhost_blk/blk_spec.h
 create mode 100644 examples/vhost_blk/meson.build
 create mode 100644 examples/vhost_blk/vhost_blk.c
 create mode 100644 examples/vhost_blk/vhost_blk.h
 create mode 100644 examples/vhost_blk/vhost_blk_compat.c

diff --git a/examples/vhost_blk/Makefile b/examples/vhost_blk/Makefile
new file mode 100644
index 000000000..52e9befd8
--- /dev/null
+++ b/examples/vhost_blk/Makefile
@@ -0,0 +1,67 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2010-2017 Intel Corporation
+
+# binary name
+APP = vhost-blk
+
+# all source are stored in SRCS-y
+SRCS-y := blk.c vhost_blk.c vhost_blk_compat.c
+
+# Build using pkg-config variables if possible
+$(shell pkg-config --exists libdpdk)
+ifeq ($(.SHELLSTATUS),0)
+
+all: shared
+.PHONY: shared static
+shared: build/$(APP)-shared
+	ln -sf $(APP)-shared build/$(APP)
+static: build/$(APP)-static
+	ln -sf $(APP)-static build/$(APP)
+
+CFLAGS += -D_FILE_OFFSET_BITS=64
+LDFLAGS += -pthread
+
+PC_FILE := $(shell pkg-config --path libdpdk)
+CFLAGS += -O3 $(shell pkg-config --cflags libdpdk)
+LDFLAGS_SHARED = $(shell pkg-config --libs libdpdk)
+LDFLAGS_STATIC = -Wl,-Bstatic $(shell pkg-config --static --libs libdpdk)
+
+build/$(APP)-shared: $(SRCS-y) Makefile $(PC_FILE) | build
+	$(CC) $(CFLAGS) $(SRCS-y) -o $@ $(LDFLAGS) $(LDFLAGS_SHARED)
+
+build/$(APP)-static: $(SRCS-y) Makefile $(PC_FILE) | build
+	$(CC) $(CFLAGS) $(SRCS-y) -o $@ $(LDFLAGS) $(LDFLAGS_STATIC)
+
+build:
+	@mkdir -p $@
+
+.PHONY: clean
+clean:
+	rm -f build/$(APP) build/$(APP)-static build/$(APP)-shared
+	test -d build && rmdir -p build || true
+
+else # Build using legacy build system
+
+ifeq ($(RTE_SDK),)
+$(error "Please define RTE_SDK environment variable")
+endif
+
+# Default target, detect a build directory, by looking for a path with a .config
+RTE_TARGET ?= $(notdir $(abspath $(dir $(firstword $(wildcard $(RTE_SDK)/*/.config)))))
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+ifneq ($(CONFIG_RTE_EXEC_ENV_LINUX),y)
+$(info This application can only operate in a linux environment, \
+please change the definition of the RTE_TARGET environment variable)
+all:
+else
+
+CFLAGS += -D_FILE_OFFSET_BITS=64
+CFLAGS += -O2
+#CFLAGS += $(WERROR_FLAGS)
+
+include $(RTE_SDK)/mk/rte.extapp.mk
+
+endif
+endif
diff --git a/examples/vhost_blk/blk.c b/examples/vhost_blk/blk.c
new file mode 100644
index 000000000..b60550b26
--- /dev/null
+++ b/examples/vhost_blk/blk.c
@@ -0,0 +1,117 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2010-2019 Intel Corporation
+ */
+
+/**
+ * This work is largely based on the "vhost-user-blk" implementation by
+ * SPDK(https://github.com/spdk/spdk).
+ */
+
+#include <stdio.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <assert.h>
+#include <ctype.h>
+#include <string.h>
+#include <stddef.h>
+
+#include <rte_atomic.h>
+#include <rte_cycles.h>
+#include <rte_log.h>
+#include <rte_malloc.h>
+#include <rte_byteorder.h>
+#include <rte_string_fns.h>
+
+#include "vhost_blk.h"
+#include "blk_spec.h"
+
+static void
+vhost_strcpy_pad(void *dst, const char *src, size_t size, int pad)
+{
+	size_t len;
+
+	len = strlen(src);
+	if (len < size) {
+		memcpy(dst, src, len);
+		memset((char *)dst + len, pad, size - len);
+	} else {
+		memcpy(dst, src, size);
+	}
+}
+
+static int
+vhost_bdev_blk_readwrite(struct vhost_block_dev *bdev,
+			  struct vhost_blk_task *task,
+			  uint64_t lba_512, __rte_unused uint32_t xfer_len)
+{
+	uint32_t i;
+	uint64_t offset;
+	uint32_t nbytes = 0;
+
+	offset = lba_512 * 512;
+
+	for (i = 0; i < task->iovs_cnt; i++) {
+		if (task->dxfer_dir == BLK_DIR_TO_DEV)
+			memcpy(bdev->data + offset, task->iovs[i].iov_base,
+			       task->iovs[i].iov_len);
+		else
+			memcpy(task->iovs[i].iov_base, bdev->data + offset,
+			       task->iovs[i].iov_len);
+		offset += task->iovs[i].iov_len;
+		nbytes += task->iovs[i].iov_len;
+	}
+
+	return nbytes;
+}
+
+int
+vhost_bdev_process_blk_commands(struct vhost_block_dev *bdev,
+				 struct vhost_blk_task *task)
+{
+	int used_len;
+
+	if (unlikely(task->data_len > (bdev->blockcnt * bdev->blocklen))) {
+		fprintf(stderr, "read or write beyond capacity\n");
+		return VIRTIO_BLK_S_UNSUPP;
+	}
+
+	switch (task->req->type) {
+	case VIRTIO_BLK_T_IN:
+		if (unlikely(task->data_len == 0 || (task->data_len & (512 - 1)) != 0)) {
+			fprintf(stderr,
+				"%s - passed IO buffer is not multiple of 512b (req_idx = %"PRIu16").\n",
+				task->req->type ? "WRITE" : "READ", task->req_idx);
+			return VIRTIO_BLK_S_UNSUPP;
+		}
+
+		task->dxfer_dir = BLK_DIR_FROM_DEV;
+		vhost_bdev_blk_readwrite(bdev, task,
+					 task->req->sector, task->data_len);
+		break;
+	case VIRTIO_BLK_T_OUT:
+		if (unlikely(task->data_len == 0 || (task->data_len & (512 - 1)) != 0)) {
+			fprintf(stderr,
+				"%s - passed IO buffer is not multiple of 512b (req_idx = %"PRIu16").\n",
+				task->req->type ? "WRITE" : "READ", task->req_idx);
+			return VIRTIO_BLK_S_UNSUPP;
+		}
+
+		if (task->readtype)
+			return VIRTIO_BLK_S_IOERR;
+		task->dxfer_dir = BLK_DIR_TO_DEV;
+		vhost_bdev_blk_readwrite(bdev, task,
+					 task->req->sector, task->data_len);
+		break;
+	case VIRTIO_BLK_T_GET_ID:
+		if (!task->iovs_cnt || task->data_len)
+			return VIRTIO_BLK_S_UNSUPP;
+		used_len = min((size_t)VIRTIO_BLK_ID_BYTES, task->data_len);
+		vhost_strcpy_pad(task->iovs[0].iov_base,
+				 bdev->product_name, used_len, ' ');
+		break;
+	default:
+		return VIRTIO_BLK_S_UNSUPP;
+	}
+
+	return VIRTIO_BLK_S_OK;
+}
diff --git a/examples/vhost_blk/blk_spec.h b/examples/vhost_blk/blk_spec.h
new file mode 100644
index 000000000..5875e2f86
--- /dev/null
+++ b/examples/vhost_blk/blk_spec.h
@@ -0,0 +1,95 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2019 Intel Corporation
+ */
+
+#ifndef _BLK_SPEC_H
+#define _BLK_SPEC_H
+
+#include <stdint.h>
+
+#ifndef VHOST_USER_MEMORY_MAX_NREGIONS
+#define VHOST_USER_MEMORY_MAX_NREGIONS		8
+#endif
+
+#ifndef VHOST_USER_MAX_CONFIG_SIZE
+#define VHOST_USER_MAX_CONFIG_SIZE		256
+#endif
+
+#ifndef VHOST_USER_PROTOCOL_F_CONFIG
+#define VHOST_USER_PROTOCOL_F_CONFIG		9
+#endif
+
+#ifndef VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD
+#define VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD	12
+#endif
+
+#define VIRTIO_BLK_ID_BYTES	20	/* ID string length */
+
+#define VIRTIO_BLK_T_IN			0
+#define VIRTIO_BLK_T_OUT		1
+#define VIRTIO_BLK_T_FLUSH		4
+#define VIRTIO_BLK_T_GET_ID		8
+#define VIRTIO_BLK_T_DISCARD		11
+#define VIRTIO_BLK_T_WRITE_ZEROES	13
+
+#define VIRTIO_BLK_S_OK			0
+#define VIRTIO_BLK_S_IOERR		1
+#define VIRTIO_BLK_S_UNSUPP		2
+
+enum vhost_user_request {
+	VHOST_USER_NONE = 0,
+	VHOST_USER_GET_FEATURES = 1,
+	VHOST_USER_SET_FEATURES = 2,
+	VHOST_USER_SET_OWNER = 3,
+	VHOST_USER_RESET_OWNER = 4,
+	VHOST_USER_SET_MEM_TABLE = 5,
+	VHOST_USER_SET_LOG_BASE = 6,
+	VHOST_USER_SET_LOG_FD = 7,
+	VHOST_USER_SET_VRING_NUM = 8,
+	VHOST_USER_SET_VRING_ADDR = 9,
+	VHOST_USER_SET_VRING_BASE = 10,
+	VHOST_USER_GET_VRING_BASE = 11,
+	VHOST_USER_SET_VRING_KICK = 12,
+	VHOST_USER_SET_VRING_CALL = 13,
+	VHOST_USER_SET_VRING_ERR = 14,
+	VHOST_USER_GET_PROTOCOL_FEATURES = 15,
+	VHOST_USER_SET_PROTOCOL_FEATURES = 16,
+	VHOST_USER_GET_QUEUE_NUM = 17,
+	VHOST_USER_SET_VRING_ENABLE = 18,
+	VHOST_USER_MAX
+};
+
+/** Get/set config msg payload */
+struct vhost_user_config {
+	uint32_t offset;
+	uint32_t size;
+	uint32_t flags;
+	uint8_t region[VHOST_USER_MAX_CONFIG_SIZE];
+};
+
+/** Fixed-size vhost_memory struct */
+struct vhost_memory_padded {
+	uint32_t nregions;
+	uint32_t padding;
+	struct vhost_memory_region regions[VHOST_USER_MEMORY_MAX_NREGIONS];
+};
+
+struct vhost_user_msg {
+	enum vhost_user_request request;
+
+#define VHOST_USER_VERSION_MASK     0x3
+#define VHOST_USER_REPLY_MASK       (0x1 << 2)
+	uint32_t flags;
+	uint32_t size; /**< the following payload size */
+	union {
+#define VHOST_USER_VRING_IDX_MASK   0xff
+#define VHOST_USER_VRING_NOFD_MASK  (0x1 << 8)
+		uint64_t u64;
+		struct vhost_vring_state state;
+		struct vhost_vring_addr addr;
+		struct vhost_memory_padded memory;
+		struct vhost_user_config cfg;
+	} payload;
+} __attribute((packed));
+
+#endif
diff --git a/examples/vhost_blk/meson.build b/examples/vhost_blk/meson.build
new file mode 100644
index 000000000..028aa4f62
--- /dev/null
+++ b/examples/vhost_blk/meson.build
@@ -0,0 +1,20 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2017 Intel Corporation
+
+# meson file, for building this example as part of a main DPDK build.
+#
+# To build this example as a standalone application with an already-installed
+# DPDK instance, use 'make'
+
+if not is_linux
+	build = false
+endif
+
+if not cc.has_header('linux/virtio_blk.h')
+	build = false
+endif
+
+deps += 'vhost'
+sources = files(
+	'blk.c', 'vhost_blk.c', 'vhost_blk_compat.c'
+)
diff --git a/examples/vhost_blk/vhost_blk.c b/examples/vhost_blk/vhost_blk.c
new file mode 100644
index 000000000..f1a1f4f03
--- /dev/null
+++ b/examples/vhost_blk/vhost_blk.c
@@ -0,0 +1,589 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2010-2017 Intel Corporation
+ */
+
+#include <stdint.h>
+#include <unistd.h>
+#include <stdbool.h>
+#include <signal.h>
+#include <assert.h>
+#include <semaphore.h>
+#include <linux/virtio_blk.h>
+#include <linux/virtio_ring.h>
+
+#include <rte_atomic.h>
+#include <rte_cycles.h>
+#include <rte_log.h>
+#include <rte_malloc.h>
+#include <rte_vhost.h>
+
+#include "vhost_blk.h"
+#include "blk_spec.h"
+
+#define VHOST_BLK_FEATURES ((1ULL << VHOST_USER_F_PROTOCOL_FEATURES) | \
+			    (1ULL << VIRTIO_F_NOTIFY_ON_EMPTY) | \
+			    (1ULL << VIRTIO_F_VERSION_1))
+
+/* Path to folder where character device will be created. Can be set by user. */
+static char dev_pathname[PATH_MAX] = "";
+static sem_t exit_sem;
+
+struct vhost_blk_ctrlr *
+vhost_blk_ctrlr_find(const char *ctrlr_name)
+{
+	/* currently we only support 1 socket file fd */
+	return g_vhost_ctrlr;
+}
+
+static uint64_t gpa_to_vva(int vid, uint64_t gpa, uint64_t *len)
+{
+	char path[PATH_MAX];
+	struct vhost_blk_ctrlr *ctrlr;
+	int ret = 0;
+
+	ret = rte_vhost_get_ifname(vid, path, PATH_MAX);
+	if (ret) {
+		fprintf(stderr, "Cannot get socket name\n");
+		assert(ret != 0);
+	}
+
+	ctrlr = vhost_blk_ctrlr_find(path);
+	if (!ctrlr) {
+		fprintf(stderr, "Controller is not ready\n");
+		assert(ctrlr != NULL);
+	}
+
+	assert(ctrlr->mem != NULL);
+
+	return rte_vhost_va_from_guest_pa(ctrlr->mem, gpa, len);
+}
+
+static struct vring_desc *
+descriptor_get_next(struct vring_desc *vq_desc, struct vring_desc *cur_desc)
+{
+	return &vq_desc[cur_desc->next];
+}
+
+static bool
+descriptor_has_next(struct vring_desc *cur_desc)
+{
+	return !!(cur_desc->flags & VRING_DESC_F_NEXT);
+}
+
+static bool
+descriptor_is_wr(struct vring_desc *cur_desc)
+{
+	return !!(cur_desc->flags & VRING_DESC_F_WRITE);
+}
+
+static void
+submit_completion(struct vhost_blk_task *task, uint32_t vid, uint32_t q_idx)
+{
+	struct rte_vhost_vring *vq;
+	struct vring_used *used;
+
+	vq = task->vq;
+	used = vq->used;
+
+	rte_vhost_set_last_inflight_io_split(vid, q_idx, task->req_idx);
+
+	/* Fill out the next entry in the "used" ring.  id = the
+	 * index of the descriptor that contained the blk request.
+	 * len = the total amount of data transferred for the blk
+	 * request. We must report the correct len, for variable
+	 * length blk CDBs, where we may return less data than
+	 * allocated by the guest VM.
+	 */
+	used->ring[used->idx & (vq->size - 1)].id = task->req_idx;
+	used->ring[used->idx & (vq->size - 1)].len = task->data_len;
+	used->idx++;
+
+	rte_vhost_clr_inflight_desc_split(vid, q_idx, used->idx, task->req_idx);
+
+	/* Send an interrupt back to the guest VM so that it knows
+	 * a completion is ready to be processed.
+	 */
+	rte_vhost_vring_call(task->bdev->vid, q_idx);
+}
+
+static void
+vhost_process_payload_chain(struct vhost_blk_task *task)
+{
+	void *data;
+	uint64_t chunck_len;
+
+	task->iovs_cnt = 0;
+
+	do {
+		chunck_len = task->desc->len;
+		data = (void *)(uintptr_t)gpa_to_vva(task->bdev->vid,
+						     task->desc->addr,
+							 &chunck_len);
+		if (!data || chunck_len != task->desc->len) {
+			fprintf(stderr, "failed to translate desc address.\n");
+			return;
+		}
+
+		task->iovs[task->iovs_cnt].iov_base = data;
+		task->iovs[task->iovs_cnt].iov_len = task->desc->len;
+		task->data_len += task->desc->len;
+		task->iovs_cnt++;
+		task->desc = descriptor_get_next(task->vq->desc, task->desc);
+	} while (descriptor_has_next(task->desc));
+
+	chunck_len = task->desc->len;
+	task->status = (void *)(uintptr_t)gpa_to_vva(task->bdev->vid,
+						   task->desc->addr,
+						   &chunck_len);
+	if (!task->status || chunck_len != task->desc->len)
+		fprintf(stderr, "failed to translate desc address.\n");
+}
+
+static struct vhost_block_dev *
+vhost_blk_bdev_construct(const char *bdev_name, const char *bdev_serial,
+			  uint32_t blk_size, uint64_t blk_cnt,
+			  bool wce_enable)
+{
+	struct vhost_block_dev *bdev;
+
+	bdev = rte_zmalloc(NULL, sizeof(*bdev), RTE_CACHE_LINE_SIZE);
+	if (!bdev)
+		return NULL;
+
+	strncpy(bdev->name, bdev_name, sizeof(bdev->name));
+	strncpy(bdev->product_name, bdev_serial, sizeof(bdev->product_name));
+	bdev->blocklen = blk_size;
+	bdev->blockcnt = blk_cnt;
+	bdev->write_cache = wce_enable;
+
+	fprintf(stdout, "blocklen=%d, blockcnt=%d\n", bdev->blocklen, bdev->blockcnt);
+
+	/* use memory as disk storage space */
+	bdev->data = rte_zmalloc(NULL, blk_cnt * blk_size, 0);
+	if (!bdev->data) {
+		fprintf(stderr, "no enough reseverd huge memory for disk\n");
+		free(bdev);
+		return NULL;
+	}
+
+	return bdev;
+}
+
+static void
+submit_inflight_vq(struct vhost_blk_ctrlr *ctrlr, uint32_t q_idx)
+{
+	struct vhost_blk_queue *blk_vq;
+	struct rte_vhost_ring_inflight_split *inflight_vq;
+	struct rte_vhost_resubmit_info *resubmit_inflight;
+	struct rte_vhost_resubmit_desc *resubmit_list;
+	int i, req_idx;
+
+	blk_vq = &ctrlr->bdev->queues[q_idx];
+	inflight_vq = &blk_vq->inflight_vq;
+
+	resubmit_inflight = inflight_vq->resubmit_inflight_split;
+	resubmit_list = resubmit_inflight->resubmit_list;
+
+	while (resubmit_list && resubmit_inflight->resubmit_num) {
+		struct vhost_blk_task *task;
+		uint64_t chunck_len;
+		int ret;
+
+		i = (--resubmit_inflight->resubmit_num);
+		req_idx = resubmit_list[i].index;
+
+		task = rte_zmalloc(NULL, sizeof(*task), 0);
+		assert(task != NULL);
+
+		task->ctrlr = ctrlr;
+		task->bdev = ctrlr->bdev;
+		task->vq = &blk_vq->vq;
+		task->req_idx = req_idx;
+		task->desc = &task->vq->desc[task->req_idx];
+
+		/* does not support indirect descriptors */
+		assert((task->desc->flags & VRING_DESC_F_INDIRECT) == 0);
+
+		chunck_len = task->desc->len;
+		task->req = (void *)(uintptr_t)gpa_to_vva(task->bdev->vid,
+							  task->desc->addr,
+							  &chunck_len);
+		if (!task->req || chunck_len != task->desc->len) {
+			fprintf(stderr, "failed to translate desc address.\n");
+			rte_free(task);
+			return;
+		}
+
+		task->desc = descriptor_get_next(task->vq->desc, task->desc);
+		if (!descriptor_has_next(task->desc)) {
+			task->dxfer_dir = BLK_DIR_NONE;
+			chunck_len = task->desc->len;
+			task->status = (void *)(uintptr_t)
+					      gpa_to_vva(task->bdev->vid,
+							 task->desc->addr,
+							 &chunck_len);
+			if (!task->status || chunck_len != task->desc->len) {
+				fprintf(stderr, "failed to translate desc address.\n");
+				rte_free(task);
+				return;
+			}
+		} else {
+			task->readtype = descriptor_is_wr(task->desc);
+			vhost_process_payload_chain(task);
+		}
+
+		ret = vhost_bdev_process_blk_commands(ctrlr->bdev, task);
+		if (ret) {
+			/* invalid response */
+			*task->status = VIRTIO_BLK_S_IOERR;
+		} else {
+			/* successfully */
+			*task->status = VIRTIO_BLK_S_OK;
+		}
+		submit_completion(task, ctrlr->bdev->vid, q_idx);
+		rte_free(task);
+	}
+}
+
+static void
+process_requestq(struct vhost_blk_ctrlr *ctrlr, uint32_t q_idx)
+{
+	int ret;
+	struct vhost_blk_queue *blk_vq;
+	struct rte_vhost_vring *vq;
+
+	blk_vq = &ctrlr->bdev->queues[q_idx];
+	vq = &blk_vq->vq;
+
+	while (vq->avail->idx != blk_vq->last_avail_idx) {
+		int req_idx;
+		uint16_t last_idx;
+		struct vhost_blk_task *task;
+		uint64_t chunck_len;
+
+		last_idx = blk_vq->last_avail_idx & (vq->size - 1);
+		req_idx = vq->avail->ring[last_idx];
+
+		task = rte_zmalloc(NULL, sizeof(*task), 0);
+		assert(task != NULL);
+
+		task->ctrlr = ctrlr;
+		task->bdev = ctrlr->bdev;
+		task->vq = vq;
+		task->req_idx = req_idx;
+		task->desc = &task->vq->desc[task->req_idx];
+
+		rte_vhost_set_inflight_desc_split(ctrlr->bdev->vid, q_idx, last_idx);
+
+		/* does not support indirect descriptors */
+		assert((task->desc->flags & VRING_DESC_F_INDIRECT) == 0);
+		blk_vq->last_avail_idx++;
+
+		chunck_len = task->desc->len;
+		task->req = (void *)(uintptr_t)gpa_to_vva(task->bdev->vid,
+							  task->desc->addr,
+							  &chunck_len);
+		if (!task->req || chunck_len != task->desc->len) {
+			fprintf(stderr, "failed to translate desc address.\n");
+			rte_free(task);
+			return;
+		}
+
+		task->desc = descriptor_get_next(task->vq->desc, task->desc);
+		if (!descriptor_has_next(task->desc)) {
+			task->dxfer_dir = BLK_DIR_NONE;
+			chunck_len = task->desc->len;
+			task->status = (void *)(uintptr_t)
+					      gpa_to_vva(task->bdev->vid,
+							 task->desc->addr,
+							 &chunck_len);
+			if (!task->status || chunck_len != task->desc->len) {
+				fprintf(stderr, "failed to translate desc address.\n");
+				rte_free(task);
+				return;
+			}
+		} else {
+			task->readtype = descriptor_is_wr(task->desc);
+			vhost_process_payload_chain(task);
+		}
+
+		ret = vhost_bdev_process_blk_commands(ctrlr->bdev, task);
+		if (ret) {
+			/* invalid response */
+			*task->status = VIRTIO_BLK_S_IOERR;
+		} else {
+			/* successfully */
+			*task->status = VIRTIO_BLK_S_OK;
+		}
+
+		submit_completion(task, ctrlr->bdev->vid, q_idx);
+		rte_free(task);
+	}
+}
+
+/* Main framework for processing IOs */
+static void *
+ctrlr_worker(void *arg)
+{
+	struct vhost_blk_ctrlr *ctrlr = (struct vhost_blk_ctrlr *)arg;
+	struct vhost_blk_queue *blk_vq;
+	struct rte_vhost_ring_inflight_split *inflight_vq;
+	struct rte_vhost_resubmit_info *resubmit_inflight;
+	cpu_set_t cpuset;
+	pthread_t thread;
+	int i, ret;
+
+	fprintf(stdout, "Ctrlr Worker Thread start\n");
+
+	if (ctrlr == NULL || ctrlr->bdev == NULL) {
+		fprintf(stderr, "%s: Error, invalid argument passed to worker thread\n",
+				__func__);
+		exit(0);
+	}
+
+	thread = pthread_self();
+	CPU_ZERO(&cpuset);
+	CPU_SET(0, &cpuset);
+	pthread_setaffinity_np(thread, sizeof(cpu_set_t), &cpuset);
+
+	for (i = 0; i < NUM_OF_BLK_QUEUES; i++) {
+		blk_vq = &ctrlr->bdev->queues[i];
+		inflight_vq = &blk_vq->inflight_vq;
+		resubmit_inflight = inflight_vq->resubmit_inflight_split;
+		if (resubmit_inflight && resubmit_inflight->resubmit_num)
+			submit_inflight_vq(ctrlr, i);
+	}
+
+	while (!g_should_stop && ctrlr->bdev != NULL) {
+		for (i = 0; i < NUM_OF_BLK_QUEUES; i++)
+			process_requestq(ctrlr, i);
+	}
+
+	fprintf(stdout, "Ctrlr Worker Thread Exiting\n");
+	sem_post(&exit_sem);
+	return NULL;
+}
+
+static int
+new_device(int vid)
+{
+	char path[PATH_MAX];
+	struct vhost_blk_ctrlr *ctrlr;
+	struct vhost_blk_queue *blk_vq;
+	struct rte_vhost_vring *vq;
+	struct rte_vhost_ring_inflight_split *inflight_vq;
+	pthread_t tid;
+	int i, ret;
+
+	ctrlr = vhost_blk_ctrlr_find(path);
+	if (!ctrlr) {
+		fprintf(stderr, "Controller is not ready\n");
+		return -1;
+	}
+
+	if (ctrlr->started)
+		return 0;
+
+	ctrlr->bdev->vid = vid;
+
+	ret = rte_vhost_get_mem_table(vid, &ctrlr->mem);
+	if (ret)
+		fprintf(stderr, "Get Controller memory region failed\n");
+	assert(ctrlr->mem != NULL);
+
+	/* Disable Notifications and init last idx */
+	for (i = 0; i < NUM_OF_BLK_QUEUES; i++) {
+		rte_vhost_enable_guest_notification(vid, i, 0);
+
+		blk_vq = &ctrlr->bdev->queues[i];
+		vq = &blk_vq->vq;
+		inflight_vq = &blk_vq->inflight_vq;
+		ret = rte_vhost_get_vring_base(ctrlr->bdev->vid, i,
+					       &blk_vq->last_avail_idx,
+					       &blk_vq->last_used_idx);
+		assert(ret == 0);
+		ret = rte_vhost_get_vhost_vring(ctrlr->bdev->vid, i, vq);
+		assert(ret == 0);
+		ret = rte_vhost_get_vhost_ring_inflight_split(ctrlr->bdev->vid, i,
+							      inflight_vq);
+		assert(ret == 0);
+	}
+
+	/* start polling vring */
+	g_should_stop = 0;
+	fprintf(stdout, "New Device %s, Device ID %d\n", path, vid);
+	if (pthread_create(&tid, NULL, &ctrlr_worker, ctrlr) < 0) {
+		fprintf(stderr, "Worker Thread Started Failed\n");
+		return -1;
+	}
+
+	/* device has been started */
+	ctrlr->started = 1;
+	pthread_detach(tid);
+	return 0;
+}
+
+static void
+destroy_device(int vid)
+{
+	char path[PATH_MAX];
+	struct vhost_blk_ctrlr *ctrlr;
+	struct vhost_blk_queue *blk_vq;
+	struct rte_vhost_vring *vq;
+	int i, ret;
+
+	ret = rte_vhost_get_ifname(vid, path, PATH_MAX);
+	if (ret) {
+		fprintf(stderr, "Destroy Ctrlr Failed\n");
+		return;
+	}
+	fprintf(stdout, "Destroy %s Device ID %d\n", path, vid);
+	ctrlr = vhost_blk_ctrlr_find(path);
+	if (!ctrlr) {
+		fprintf(stderr, "Destroy Ctrlr Failed\n");
+		return;
+	}
+
+	if (!ctrlr->started)
+		return;
+
+	g_should_stop = 1;
+
+	for (i = 0; i < NUM_OF_BLK_QUEUES; i++) {
+		blk_vq = &ctrlr->bdev->queues[i];
+		rte_vhost_set_vring_base(ctrlr->bdev->vid, i,
+					 blk_vq->last_avail_idx, blk_vq->last_used_idx);
+	}
+
+	free(ctrlr->mem);
+
+	ctrlr->started = 0;
+	sem_wait(&exit_sem);
+}
+
+static int
+new_connection(int vid)
+{
+	/* extend the proper features for block device */
+	vhost_session_install_rte_compat_hooks(vid);
+}
+
+struct vhost_device_ops vhost_blk_device_ops = {
+	.new_device =  new_device,
+	.destroy_device = destroy_device,
+	.new_connection = new_connection,
+};
+
+static struct vhost_blk_ctrlr *
+vhost_blk_ctrlr_construct(const char *ctrlr_name)
+{
+	int ret;
+	struct vhost_blk_ctrlr *ctrlr;
+	char *path;
+	char cwd[PATH_MAX];
+
+	/* always use current directory */
+	path = getcwd(cwd, PATH_MAX);
+	if (!path) {
+		fprintf(stderr, "Cannot get current working directory\n");
+		return NULL;
+	}
+	snprintf(dev_pathname, sizeof(dev_pathname), "%s/%s", path, ctrlr_name);
+
+	if (access(dev_pathname, F_OK) != -1) {
+		if (unlink(dev_pathname) != 0)
+			rte_exit(EXIT_FAILURE, "Cannot remove %s.\n",
+				 dev_pathname);
+	}
+
+	if (rte_vhost_driver_register(dev_pathname, 0) != 0) {
+		fprintf(stderr, "socket %s already exists\n", dev_pathname);
+		return NULL;
+	}
+
+	ret = rte_vhost_driver_set_features(dev_pathname, VHOST_BLK_FEATURES);
+	if (ret != 0) {
+		fprintf(stderr, "Set vhost driver features failed\n");
+		rte_vhost_driver_unregister(dev_pathname);
+		return NULL;
+	}
+
+	/* set proper features */
+	vhost_dev_install_rte_compat_hooks(dev_pathname);
+
+	ctrlr = rte_zmalloc(NULL, sizeof(*ctrlr), RTE_CACHE_LINE_SIZE);
+	if (!ctrlr) {
+		rte_vhost_driver_unregister(dev_pathname);
+		return NULL;
+	}
+
+	/* hardcoded block device information with 128MiB */
+	ctrlr->bdev = vhost_blk_bdev_construct("malloc0", "vhost_blk_malloc0",
+						4096, 32768, 0);
+	if (!ctrlr->bdev) {
+		rte_free(ctrlr);
+		rte_vhost_driver_unregister(dev_pathname);
+		return NULL;
+	}
+
+	rte_vhost_driver_callback_register(dev_pathname,
+					   &vhost_blk_device_ops);
+
+	return ctrlr;
+}
+
+static void
+signal_handler(__rte_unused int signum)
+{
+	struct vhost_blk_ctrlr *ctrlr;
+
+	if (access(dev_pathname, F_OK) == 0)
+		unlink(dev_pathname);
+
+	g_should_stop = 1;
+
+	ctrlr = vhost_blk_ctrlr_find(NULL);
+	if (ctrlr != NULL) {
+		fprintf(stderr, "never come in\n");
+		if (ctrlr->bdev != NULL) {
+			rte_free(ctrlr->bdev->data);
+			rte_free(ctrlr->bdev);
+		}
+		rte_free(ctrlr);
+	}
+
+	rte_vhost_driver_unregister(dev_pathname);
+	exit(0);
+}
+
+int main(int argc, char *argv[])
+{
+	int ret;
+
+	signal(SIGINT, signal_handler);
+
+	/* init EAL */
+	ret = rte_eal_init(argc, argv);
+	if (ret < 0)
+		rte_exit(EXIT_FAILURE, "Error with EAL initialization\n");
+
+	g_vhost_ctrlr = vhost_blk_ctrlr_construct("vhost.socket");
+	if (g_vhost_ctrlr == NULL) {
+		fprintf(stderr, "Construct vhost blk controller failed\n");
+		return 0;
+	}
+
+	if (sem_init(&exit_sem, 0, 0) < 0) {
+		fprintf(stderr, "Error init exit_sem\n");
+		return -1;
+	}
+
+	rte_vhost_driver_start(dev_pathname);
+
+	/* loop for exit the application */
+	while (1)
+		sleep(1);
+
+	return 0;
+}
+
diff --git a/examples/vhost_blk/vhost_blk.h b/examples/vhost_blk/vhost_blk.h
new file mode 100644
index 000000000..95350adba
--- /dev/null
+++ b/examples/vhost_blk/vhost_blk.h
@@ -0,0 +1,96 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2010-2017 Intel Corporation
+ */
+
+#ifndef _VHOST_BLK_H_
+#define _VHOST_BLK_H_
+
+#include <stdio.h>
+#include <sys/uio.h>
+#include <stdint.h>
+#include <linux/virtio_blk.h>
+#include <linux/virtio_ring.h>
+
+#include <rte_vhost.h>
+
+struct vhost_blk_queue {
+	struct rte_vhost_vring vq;
+	struct rte_vhost_ring_inflight_split inflight_vq;
+	uint16_t last_avail_idx;
+	uint16_t last_used_idx;
+};
+
+#define NUM_OF_BLK_QUEUES 1
+
+#define min(a, b) (((a) < (b)) ? (a) : (b))
+
+struct vhost_block_dev {
+	/** ID for vhost library. */
+	int vid;
+	/** Queues for the block device */
+	struct vhost_blk_queue queues[NUM_OF_BLK_QUEUES];
+	/** Unique name for this block device. */
+	char name[64];
+
+	/** Unique product name for this kind of block device. */
+	char product_name[256];
+
+	/** Size in bytes of a logical block for the backend */
+	uint32_t blocklen;
+
+	/** Number of blocks */
+	uint64_t blockcnt;
+
+	/** write cache enabled, not used at the moment */
+	int write_cache;
+
+	/** use memory as disk storage space */
+	uint8_t *data;
+};
+
+struct vhost_blk_ctrlr {
+	uint8_t started;
+	uint8_t need_restart;
+	/** Only support 1 LUN for the example */
+	struct vhost_block_dev *bdev;
+	/** VM memory region */
+	struct rte_vhost_memory *mem;
+} __rte_cache_aligned;
+
+#define VHOST_BLK_MAX_IOVS 128
+
+enum blk_data_dir {
+	BLK_DIR_NONE = 0,
+	BLK_DIR_TO_DEV = 1,
+	BLK_DIR_FROM_DEV = 2,
+};
+
+struct vhost_blk_task {
+	uint8_t readtype;
+	int req_idx;
+	uint32_t dxfer_dir;
+	uint32_t data_len;
+	struct virtio_blk_outhdr *req;
+	volatile uint8_t *status;
+	struct iovec iovs[VHOST_BLK_MAX_IOVS];
+	uint32_t iovs_cnt;
+	struct vring_desc *desc;
+	struct rte_vhost_vring *vq;
+	struct vhost_block_dev *bdev;
+	struct vhost_blk_ctrlr *ctrlr;
+};
+
+struct vhost_blk_ctrlr *g_vhost_ctrlr;
+struct vhost_device_ops vhost_blk_device_ops;
+int g_should_stop;
+
+int vhost_bdev_process_blk_commands(struct vhost_block_dev *bdev,
+				     struct vhost_blk_task *task);
+
+void vhost_session_install_rte_compat_hooks(uint32_t vid);
+
+void vhost_dev_install_rte_compat_hooks(const char *path);
+
+struct vhost_blk_ctrlr *vhost_blk_ctrlr_find(const char *ctrlr_name);
+
+#endif /* _VHOST_blk_H_ */
diff --git a/examples/vhost_blk/vhost_blk_compat.c b/examples/vhost_blk/vhost_blk_compat.c
new file mode 100644
index 000000000..405b091d7
--- /dev/null
+++ b/examples/vhost_blk/vhost_blk_compat.c
@@ -0,0 +1,193 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2010-2017 Intel Corporation
+ */
+
+#ifndef _VHOST_BLK_COMPAT_H_
+#define _VHOST_BLK_COMPAT_H_
+
+#include <sys/uio.h>
+#include <stdint.h>
+#include <linux/virtio_blk.h>
+#include <linux/virtio_ring.h>
+
+#include <rte_vhost.h>
+#include "vhost_blk.h"
+#include "blk_spec.h"
+
+#define VHOST_MAX_VQUEUES	256
+#define SPDK_VHOST_MAX_VQ_SIZE	1024
+
+#define VHOST_USER_GET_CONFIG 24
+#define VHOST_USER_SET_CONFIG 25
+
+static int
+vhost_blk_get_config(struct vhost_block_dev *bdev, uint8_t *config,
+			  uint32_t len)
+{
+	struct virtio_blk_config blkcfg;
+	uint32_t blk_size;
+	uint64_t blkcnt;
+
+	if (bdev == NULL) {
+		/* We can't just return -1 here as this GET_CONFIG message might
+		 * be caused by a QEMU VM reboot. Returning -1 will indicate an
+		 * error to QEMU, who might then decide to terminate itself.
+		 * We don't want that. A simple reboot shouldn't break the system.
+		 *
+		 * Presenting a block device with block size 0 and block count 0
+		 * doesn't cause any problems on QEMU side and the virtio-pci
+		 * device is even still available inside the VM, but there will
+		 * be no block device created for it - the kernel drivers will
+		 * silently reject it.
+		 */
+		blk_size = 0;
+		blkcnt = 0;
+	} else {
+		blk_size = bdev->blocklen;
+		blkcnt = bdev->blockcnt;
+	}
+
+	memset(&blkcfg, 0, sizeof(blkcfg));
+	blkcfg.blk_size = blk_size;
+	/* minimum I/O size in blocks */
+	blkcfg.min_io_size = 1;
+	/* expressed in 512 Bytes sectors */
+	blkcfg.capacity = (blkcnt * blk_size) / 512;
+	/* QEMU can overwrite this value when started */
+	blkcfg.num_queues = VHOST_MAX_VQUEUES;
+
+	fprintf(stdout, "block device:blk_size = %d, blkcnt = %d\n", blk_size, blkcnt);
+
+	memcpy(config, &blkcfg, min(len, sizeof(blkcfg)));
+
+	return 0;
+}
+
+static enum rte_vhost_msg_result
+extern_vhost_pre_msg_handler(int vid, void *_msg)
+{
+	char path[PATH_MAX];
+	struct vhost_blk_ctrlr *ctrlr;
+	struct vhost_user_msg *msg = _msg;
+	int ret;
+
+	ret = rte_vhost_get_ifname(vid, path, PATH_MAX);
+	if (ret) {
+		fprintf(stderr, "Cannot get socket name\n");
+		return -1;
+	}
+
+	ctrlr = vhost_blk_ctrlr_find(path);
+	if (!ctrlr) {
+		fprintf(stderr, "Controller is not ready\n");
+		return -1;
+	}
+
+	switch (msg->request) {
+	case VHOST_USER_GET_VRING_BASE:
+		if (!g_should_stop && ctrlr->started)
+			vhost_blk_device_ops.destroy_device(vid);
+		break;
+	case VHOST_USER_SET_VRING_BASE:
+	case VHOST_USER_SET_VRING_ADDR:
+	case VHOST_USER_SET_VRING_NUM:
+	case VHOST_USER_SET_VRING_KICK:
+		if (!g_should_stop && ctrlr->started)
+			vhost_blk_device_ops.destroy_device(vid);
+		break;
+	case VHOST_USER_SET_VRING_CALL:
+	case VHOST_USER_SET_MEM_TABLE:
+		if (!g_should_stop && ctrlr->started) {
+			vhost_blk_device_ops.destroy_device(vid);
+			ctrlr->need_restart = 1;
+		}
+		break;
+	case VHOST_USER_GET_CONFIG: {
+		int rc = 0;
+
+		rc = vhost_blk_get_config(ctrlr->bdev,
+					  msg->payload.cfg.region, msg->payload.cfg.size);
+		if (rc != 0)
+			msg->size = 0;
+
+		return RTE_VHOST_MSG_RESULT_REPLY;
+	}
+	case VHOST_USER_SET_CONFIG:
+	default:
+		break;
+	}
+
+	return RTE_VHOST_MSG_RESULT_NOT_HANDLED;
+}
+
+static enum rte_vhost_msg_result
+extern_vhost_post_msg_handler(int vid, void *_msg)
+{
+	char path[PATH_MAX];
+	struct vhost_blk_ctrlr *ctrlr;
+	struct vhost_user_msg *msg = _msg;
+	int ret;
+
+	ret = rte_vhost_get_ifname(vid, path, PATH_MAX);
+	if (ret) {
+		fprintf(stderr, "Cannot get socket name\n");
+		return -1;
+	}
+
+	ctrlr = vhost_blk_ctrlr_find(path);
+	if (!ctrlr) {
+		fprintf(stderr, "Controller is not ready\n");
+		return -1;
+	}
+
+	if (ctrlr->need_restart) {
+		vhost_blk_device_ops.new_device(vid);
+		ctrlr->need_restart = 0;
+	}
+
+	switch (msg->request) {
+	case VHOST_USER_SET_FEATURES:
+		break;
+	case VHOST_USER_SET_VRING_KICK:
+		/* vhost-user spec tells us to start polling a queue after receiving
+		 * its SET_VRING_KICK message. Let's do it!
+		 */
+		if (g_should_stop && !ctrlr->started)
+			vhost_blk_device_ops.new_device(vid);
+		break;
+	default:
+		break;
+	}
+
+	return RTE_VHOST_MSG_RESULT_NOT_HANDLED;
+}
+
+struct rte_vhost_user_extern_ops g_extern_vhost_ops = {
+	.pre_msg_handle = extern_vhost_pre_msg_handler,
+	.post_msg_handle = extern_vhost_post_msg_handler,
+};
+
+void
+vhost_session_install_rte_compat_hooks(uint32_t vid)
+{
+	int rc;
+
+	rc = rte_vhost_extern_callback_register(vid, &g_extern_vhost_ops, NULL);
+	if (rc != 0)
+		fprintf(stderr,
+			"rte_vhost_extern_callback_register() failed for vid = %d\n",
+			vid);
+}
+
+void
+vhost_dev_install_rte_compat_hooks(const char *path)
+{
+	uint64_t protocol_features = 0;
+
+	rte_vhost_driver_get_protocol_features(path, &protocol_features);
+	protocol_features |= (1ULL << VHOST_USER_PROTOCOL_F_CONFIG);
+	protocol_features |= (1ULL << VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD);
+	rte_vhost_driver_set_protocol_features(path, protocol_features);
+}
+
+#endif
-- 
2.17.2