1 files changed, 318 insertions, 52 deletions

diff --git a/lib/librte_vhost/vhost_user.c b/lib/librte_vhost/vhost_user.c
index ad2e8d38..f4c7ce46 100644
--- a/lib/librte_vhost/vhost_user.c
+++ b/lib/librte_vhost/vhost_user.c
@@ -48,6 +48,7 @@
 #include <rte_malloc.h>
 #include <rte_log.h>
 
+#include "iotlb.h"
 #include "vhost.h"
 #include "vhost_user.h"
 
@@ -76,6 +77,8 @@ static const char *vhost_message_str[VHOST_USER_MAX] = {
 	[VHOST_USER_SET_VRING_ENABLE]  = "VHOST_USER_SET_VRING_ENABLE",
 	[VHOST_USER_SEND_RARP]  = "VHOST_USER_SEND_RARP",
 	[VHOST_USER_NET_SET_MTU]  = "VHOST_USER_NET_SET_MTU",
+	[VHOST_USER_SET_SLAVE_REQ_FD]  = "VHOST_USER_SET_SLAVE_REQ_FD",
+	[VHOST_USER_IOTLB_MSG]  = "VHOST_USER_IOTLB_MSG",
 };
 
 static uint64_t
@@ -122,6 +125,11 @@ vhost_backend_cleanup(struct virtio_net *dev)
 		munmap((void *)(uintptr_t)dev->log_addr, dev->log_size);
 		dev->log_addr = 0;
 	}
+
+	if (dev->slave_req_fd >= 0) {
+		close(dev->slave_req_fd);
+		dev->slave_req_fd = -1;
+	}
 }
 
 /*
@@ -230,6 +238,15 @@ vhost_user_set_vring_num(struct virtio_net *dev,
 		return -1;
 	}
 
+	vq->batch_copy_elems = rte_malloc(NULL,
+				vq->size * sizeof(struct batch_copy_elem),
+				RTE_CACHE_LINE_SIZE);
+	if (!vq->batch_copy_elems) {
+		RTE_LOG(ERR, VHOST_CONFIG,
+			"failed to allocate memory for batching copy.\n");
+		return -1;
+	}
+
 	return 0;
 }
 
@@ -297,6 +314,9 @@ out:
 	dev->virtqueue[index] = vq;
 	vhost_devices[dev->vid] = dev;
 
+	if (old_vq != vq)
+		vhost_user_iotlb_init(dev, index);
+
 	return dev;
 }
 #else
@@ -307,10 +327,7 @@ numa_realloc(struct virtio_net *dev, int index __rte_unused)
 }
 #endif
 
-/*
- * Converts QEMU virtual address to Vhost virtual address. This function is
- * used to convert the ring addresses to our address space.
- */
+/* Converts QEMU virtual address to Vhost virtual address. */
 static uint64_t
 qva_to_vva(struct virtio_net *dev, uint64_t qva)
 {
@@ -331,50 +348,69 @@ qva_to_vva(struct virtio_net *dev, uint64_t qva)
 	return 0;
 }
 
+
 /*
- * The virtio device sends us the desc, used and avail ring addresses.
- * This function then converts these to our address space.
+ * Converts ring address to Vhost virtual address.
+ * If IOMMU is enabled, the ring address is a guest IO virtual address,
+ * else it is a QEMU virtual address.
  */
-static int
-vhost_user_set_vring_addr(struct virtio_net *dev, VhostUserMsg *msg)
+static uint64_t
+ring_addr_to_vva(struct virtio_net *dev, struct vhost_virtqueue *vq,
+		uint64_t ra, uint64_t size)
 {
-	struct vhost_virtqueue *vq;
+	if (dev->features & (1ULL << VIRTIO_F_IOMMU_PLATFORM)) {
+		uint64_t vva;
 
-	if (dev->mem == NULL)
-		return -1;
+		vva = vhost_user_iotlb_cache_find(vq, ra,
+					&size, VHOST_ACCESS_RW);
+		if (!vva)
+			vhost_user_iotlb_miss(dev, ra, VHOST_ACCESS_RW);
 
-	/* addr->index refers to the queue index. The txq 1, rxq is 0. */
-	vq = dev->virtqueue[msg->payload.addr.index];
+		return vva;
+	}
+
+	return qva_to_vva(dev, ra);
+}
+
+static struct virtio_net *
+translate_ring_addresses(struct virtio_net *dev, int vq_index)
+{
+	struct vhost_virtqueue *vq = dev->virtqueue[vq_index];
+	struct vhost_vring_addr *addr = &vq->ring_addrs;
 
 	/* The addresses are converted from QEMU virtual to Vhost virtual. */
-	vq->desc = (struct vring_desc *)(uintptr_t)qva_to_vva(dev,
-			msg->payload.addr.desc_user_addr);
+	if (vq->desc && vq->avail && vq->used)
+		return dev;
+
+	vq->desc = (struct vring_desc *)(uintptr_t)ring_addr_to_vva(dev,
+			vq, addr->desc_user_addr, sizeof(struct vring_desc));
 	if (vq->desc == 0) {
-		RTE_LOG(ERR, VHOST_CONFIG,
+		RTE_LOG(DEBUG, VHOST_CONFIG,
 			"(%d) failed to find desc ring address.\n",
 			dev->vid);
-		return -1;
+		return dev;
 	}
 
-	dev = numa_realloc(dev, msg->payload.addr.index);
-	vq = dev->virtqueue[msg->payload.addr.index];
+	dev = numa_realloc(dev, vq_index);
+	vq = dev->virtqueue[vq_index];
+	addr = &vq->ring_addrs;
 
-	vq->avail = (struct vring_avail *)(uintptr_t)qva_to_vva(dev,
-			msg->payload.addr.avail_user_addr);
+	vq->avail = (struct vring_avail *)(uintptr_t)ring_addr_to_vva(dev,
+			vq, addr->avail_user_addr, sizeof(struct vring_avail));
 	if (vq->avail == 0) {
-		RTE_LOG(ERR, VHOST_CONFIG,
+		RTE_LOG(DEBUG, VHOST_CONFIG,
 			"(%d) failed to find avail ring address.\n",
 			dev->vid);
-		return -1;
+		return dev;
 	}
 
-	vq->used = (struct vring_used *)(uintptr_t)qva_to_vva(dev,
-			msg->payload.addr.used_user_addr);
+	vq->used = (struct vring_used *)(uintptr_t)ring_addr_to_vva(dev,
+			vq, addr->used_user_addr, sizeof(struct vring_used));
 	if (vq->used == 0) {
-		RTE_LOG(ERR, VHOST_CONFIG,
+		RTE_LOG(DEBUG, VHOST_CONFIG,
 			"(%d) failed to find used ring address.\n",
 			dev->vid);
-		return -1;
+		return dev;
 	}
 
 	if (vq->last_used_idx != vq->used->idx) {
@@ -386,7 +422,7 @@ vhost_user_set_vring_addr(struct virtio_net *dev, VhostUserMsg *msg)
 		vq->last_avail_idx = vq->used->idx;
 	}
 
-	vq->log_guest_addr = msg->payload.addr.log_guest_addr;
+	vq->log_guest_addr = addr->log_guest_addr;
 
 	LOG_DEBUG(VHOST_CONFIG, "(%d) mapped address desc: %p\n",
 			dev->vid, vq->desc);
@@ -397,6 +433,43 @@ vhost_user_set_vring_addr(struct virtio_net *dev, VhostUserMsg *msg)
 	LOG_DEBUG(VHOST_CONFIG, "(%d) log_guest_addr: %" PRIx64 "\n",
 			dev->vid, vq->log_guest_addr);
 
+	return dev;
+}
+
+/*
+ * The virtio device sends us the desc, used and avail ring addresses.
+ * This function then converts these to our address space.
+ */
+static int
+vhost_user_set_vring_addr(struct virtio_net **pdev, VhostUserMsg *msg)
+{
+	struct vhost_virtqueue *vq;
+	struct vhost_vring_addr *addr = &msg->payload.addr;
+	struct virtio_net *dev = *pdev;
+
+	if (dev->mem == NULL)
+		return -1;
+
+	/* addr->index refers to the queue index. The txq 1, rxq is 0. */
+	vq = dev->virtqueue[msg->payload.addr.index];
+
+	/*
+	 * Rings addresses should not be interpreted as long as the ring is not
+	 * started and enabled
+	 */
+	memcpy(&vq->ring_addrs, addr, sizeof(*addr));
+
+	vring_invalidate(dev, vq);
+
+	if (vq->enabled && (dev->features &
+				(1ULL << VHOST_USER_F_PROTOCOL_FEATURES))) {
+		dev = translate_ring_addresses(dev, msg->payload.state.index);
+		if (!dev)
+			return -1;
+
+		*pdev = dev;
+	}
+
 	return 0;
 }
 
@@ -453,7 +526,7 @@ add_guest_pages(struct virtio_net *dev, struct rte_vhost_mem_region *reg,
 	uint64_t host_phys_addr;
 	uint64_t size;
 
-	host_phys_addr = rte_mem_virt2phy((void *)(uintptr_t)host_user_addr);
+	host_phys_addr = rte_mem_virt2iova((void *)(uintptr_t)host_user_addr);
 	size = page_size - (guest_phys_addr & (page_size - 1));
 	size = RTE_MIN(size, reg_size);
 
@@ -464,7 +537,7 @@ add_guest_pages(struct virtio_net *dev, struct rte_vhost_mem_region *reg,
 
 	while (reg_size > 0) {
 		size = RTE_MIN(reg_size, page_size);
-		host_phys_addr = rte_mem_virt2phy((void *)(uintptr_t)
+		host_phys_addr = rte_mem_virt2iova((void *)(uintptr_t)
 						  host_user_addr);
 		add_one_guest_page(dev, guest_phys_addr, host_phys_addr, size);
 
@@ -620,7 +693,7 @@ err_mmap:
 static int
 vq_is_ready(struct vhost_virtqueue *vq)
 {
-	return vq && vq->desc   &&
+	return vq && vq->desc && vq->avail && vq->used &&
 	       vq->kickfd != VIRTIO_UNINITIALIZED_EVENTFD &&
 	       vq->callfd != VIRTIO_UNINITIALIZED_EVENTFD;
 }
@@ -668,10 +741,11 @@ vhost_user_set_vring_call(struct virtio_net *dev, struct VhostUserMsg *pmsg)
 }
 
 static void
-vhost_user_set_vring_kick(struct virtio_net *dev, struct VhostUserMsg *pmsg)
+vhost_user_set_vring_kick(struct virtio_net **pdev, struct VhostUserMsg *pmsg)
 {
 	struct vhost_vring_file file;
 	struct vhost_virtqueue *vq;
+	struct virtio_net *dev = *pdev;
 
 	file.index = pmsg->payload.u64 & VHOST_USER_VRING_IDX_MASK;
 	if (pmsg->payload.u64 & VHOST_USER_VRING_NOFD_MASK)
@@ -681,7 +755,23 @@ vhost_user_set_vring_kick(struct virtio_net *dev, struct VhostUserMsg *pmsg)
 	RTE_LOG(INFO, VHOST_CONFIG,
 		"vring kick idx:%d file:%d\n", file.index, file.fd);
 
+	/* Interpret ring addresses only when ring is started. */
+	dev = translate_ring_addresses(dev, file.index);
+	if (!dev)
+		return;
+
+	*pdev = dev;
+
 	vq = dev->virtqueue[file.index];
+
+	/*
+	 * When VHOST_USER_F_PROTOCOL_FEATURES is not negotiated,
+	 * the ring starts already enabled. Otherwise, it is enabled via
+	 * the SET_VRING_ENABLE message.
+	 */
+	if (!(dev->features & (1ULL << VHOST_USER_F_PROTOCOL_FEATURES)))
+		vq->enabled = 1;
+
 	if (vq->kickfd >= 0)
 		close(vq->kickfd);
 	vq->kickfd = file.fd;
@@ -741,6 +831,9 @@ vhost_user_get_vring_base(struct virtio_net *dev,
 	rte_free(vq->shadow_used_ring);
 	vq->shadow_used_ring = NULL;
 
+	rte_free(vq->batch_copy_elems);
+	vq->batch_copy_elems = NULL;
+
 	return 0;
 }
 
@@ -768,6 +861,27 @@ vhost_user_set_vring_enable(struct virtio_net *dev,
 }
 
 static void
+vhost_user_get_protocol_features(struct virtio_net *dev,
+				 struct VhostUserMsg *msg)
+{
+	uint64_t features, protocol_features = VHOST_USER_PROTOCOL_FEATURES;
+
+	rte_vhost_driver_get_features(dev->ifname, &features);
+
+	/*
+	 * REPLY_ACK protocol feature is only mandatory for now
+	 * for IOMMU feature. If IOMMU is explicitly disabled by the
+	 * application, disable also REPLY_ACK feature for older buggy
+	 * Qemu versions (from v2.7.0 to v2.9.0).
+	 */
+	if (!(features & (1ULL << VIRTIO_F_IOMMU_PLATFORM)))
+		protocol_features &= ~(1ULL << VHOST_USER_PROTOCOL_F_REPLY_ACK);
+
+	msg->payload.u64 = protocol_features;
+	msg->size = sizeof(msg->payload.u64);
+}
+
+static void
 vhost_user_set_protocol_features(struct virtio_net *dev,
 				 uint64_t protocol_features)
 {
@@ -874,6 +988,116 @@ vhost_user_net_set_mtu(struct virtio_net *dev, struct VhostUserMsg *msg)
 	return 0;
 }
 
+static int
+vhost_user_set_req_fd(struct virtio_net *dev, struct VhostUserMsg *msg)
+{
+	int fd = msg->fds[0];
+
+	if (fd < 0) {
+		RTE_LOG(ERR, VHOST_CONFIG,
+				"Invalid file descriptor for slave channel (%d)\n",
+				fd);
+		return -1;
+	}
+
+	dev->slave_req_fd = fd;
+
+	return 0;
+}
+
+static int
+is_vring_iotlb_update(struct vhost_virtqueue *vq, struct vhost_iotlb_msg *imsg)
+{
+	struct vhost_vring_addr *ra;
+	uint64_t start, end;
+
+	start = imsg->iova;
+	end = start + imsg->size;
+
+	ra = &vq->ring_addrs;
+	if (ra->desc_user_addr >= start && ra->desc_user_addr < end)
+		return 1;
+	if (ra->avail_user_addr >= start && ra->avail_user_addr < end)
+		return 1;
+	if (ra->used_user_addr >= start && ra->used_user_addr < end)
+		return 1;
+
+	return 0;
+}
+
+static int
+is_vring_iotlb_invalidate(struct vhost_virtqueue *vq,
+				struct vhost_iotlb_msg *imsg)
+{
+	uint64_t istart, iend, vstart, vend;
+
+	istart = imsg->iova;
+	iend = istart + imsg->size - 1;
+
+	vstart = (uintptr_t)vq->desc;
+	vend = vstart + sizeof(struct vring_desc) * vq->size - 1;
+	if (vstart <= iend && istart <= vend)
+		return 1;
+
+	vstart = (uintptr_t)vq->avail;
+	vend = vstart + sizeof(struct vring_avail);
+	vend += sizeof(uint16_t) * vq->size - 1;
+	if (vstart <= iend && istart <= vend)
+		return 1;
+
+	vstart = (uintptr_t)vq->used;
+	vend = vstart + sizeof(struct vring_used);
+	vend += sizeof(struct vring_used_elem) * vq->size - 1;
+	if (vstart <= iend && istart <= vend)
+		return 1;
+
+	return 0;
+}
+
+static int
+vhost_user_iotlb_msg(struct virtio_net **pdev, struct VhostUserMsg *msg)
+{
+	struct virtio_net *dev = *pdev;
+	struct vhost_iotlb_msg *imsg = &msg->payload.iotlb;
+	uint16_t i;
+	uint64_t vva;
+
+	switch (imsg->type) {
+	case VHOST_IOTLB_UPDATE:
+		vva = qva_to_vva(dev, imsg->uaddr);
+		if (!vva)
+			return -1;
+
+		for (i = 0; i < dev->nr_vring; i++) {
+			struct vhost_virtqueue *vq = dev->virtqueue[i];
+
+			vhost_user_iotlb_cache_insert(vq, imsg->iova, vva,
+					imsg->size, imsg->perm);
+
+			if (is_vring_iotlb_update(vq, imsg))
+				*pdev = dev = translate_ring_addresses(dev, i);
+		}
+		break;
+	case VHOST_IOTLB_INVALIDATE:
+		for (i = 0; i < dev->nr_vring; i++) {
+			struct vhost_virtqueue *vq = dev->virtqueue[i];
+
+			vhost_user_iotlb_cache_remove(vq, imsg->iova,
+					imsg->size);
+
+			if (is_vring_iotlb_invalidate(vq, imsg))
+				vring_invalidate(dev, vq);
+		}
+		break;
+	default:
+		RTE_LOG(ERR, VHOST_CONFIG, "Invalid IOTLB message type (%d)\n",
+				imsg->type);
+		return -1;
+	}
+
+	return 0;
+}
+
 /* return bytes# of read on success or negative val on failure. */
 static int
 read_vhost_message(int sockfd, struct VhostUserMsg *msg)
@@ -907,8 +1131,16 @@ read_vhost_message(int sockfd, struct VhostUserMsg *msg)
 static int
 send_vhost_message(int sockfd, struct VhostUserMsg *msg)
 {
-	int ret;
+	if (!msg)
+		return 0;
+
+	return send_fd_message(sockfd, (char *)msg,
+		VHOST_USER_HDR_SIZE + msg->size, NULL, 0);
+}
 
+static int
+send_vhost_reply(int sockfd, struct VhostUserMsg *msg)
+{
 	if (!msg)
 		return 0;
 
@@ -917,10 +1149,7 @@ send_vhost_message(int sockfd, struct VhostUserMsg *msg)
 	msg->flags |= VHOST_USER_VERSION;
 	msg->flags |= VHOST_USER_REPLY_MASK;
 
-	ret = send_fd_message(sockfd, (char *)msg,
-		VHOST_USER_HDR_SIZE + msg->size, NULL, 0);
-
-	return ret;
+	return send_vhost_message(sockfd, msg);
 }
 
 /*
@@ -931,7 +1160,7 @@ vhost_user_check_and_alloc_queue_pair(struct virtio_net *dev, VhostUserMsg *msg)
 {
 	uint16_t vring_idx;
 
-	switch (msg->request) {
+	switch (msg->request.master) {
 	case VHOST_USER_SET_VRING_KICK:
 	case VHOST_USER_SET_VRING_CALL:
 	case VHOST_USER_SET_VRING_ERR:
@@ -983,7 +1212,7 @@ vhost_user_msg_handler(int vid, int fd)
 	}
 
 	ret = read_vhost_message(fd, &msg);
-	if (ret <= 0 || msg.request >= VHOST_USER_MAX) {
+	if (ret <= 0 || msg.request.master >= VHOST_USER_MAX) {
 		if (ret < 0)
 			RTE_LOG(ERR, VHOST_CONFIG,
 				"vhost read message failed\n");
@@ -998,8 +1227,12 @@ vhost_user_msg_handler(int vid, int fd)
 	}
 
 	ret = 0;
-	RTE_LOG(INFO, VHOST_CONFIG, "read message %s\n",
-		vhost_message_str[msg.request]);
+	if (msg.request.master != VHOST_USER_IOTLB_MSG)
+		RTE_LOG(INFO, VHOST_CONFIG, "read message %s\n",
+			vhost_message_str[msg.request.master]);
+	else
+		RTE_LOG(DEBUG, VHOST_CONFIG, "read message %s\n",
+			vhost_message_str[msg.request.master]);
 
 	ret = vhost_user_check_and_alloc_queue_pair(dev, &msg);
 	if (ret < 0) {
@@ -1008,20 +1241,19 @@ vhost_user_msg_handler(int vid, int fd)
 		return -1;
 	}
 
-	switch (msg.request) {
+	switch (msg.request.master) {
 	case VHOST_USER_GET_FEATURES:
 		msg.payload.u64 = vhost_user_get_features(dev);
 		msg.size = sizeof(msg.payload.u64);
-		send_vhost_message(fd, &msg);
+		send_vhost_reply(fd, &msg);
 		break;
 	case VHOST_USER_SET_FEATURES:
 		vhost_user_set_features(dev, msg.payload.u64);
 		break;
 
 	case VHOST_USER_GET_PROTOCOL_FEATURES:
-		msg.payload.u64 = VHOST_USER_PROTOCOL_FEATURES;
-		msg.size = sizeof(msg.payload.u64);
-		send_vhost_message(fd, &msg);
+		vhost_user_get_protocol_features(dev, &msg);
+		send_vhost_reply(fd, &msg);
 		break;
 	case VHOST_USER_SET_PROTOCOL_FEATURES:
 		vhost_user_set_protocol_features(dev, msg.payload.u64);
@@ -1043,7 +1275,7 @@ vhost_user_msg_handler(int vid, int fd)
 
 		/* it needs a reply */
 		msg.size = sizeof(msg.payload.u64);
-		send_vhost_message(fd, &msg);
+		send_vhost_reply(fd, &msg);
 		break;
 	case VHOST_USER_SET_LOG_FD:
 		close(msg.fds[0]);
@@ -1054,7 +1286,7 @@ vhost_user_msg_handler(int vid, int fd)
 		vhost_user_set_vring_num(dev, &msg);
 		break;
 	case VHOST_USER_SET_VRING_ADDR:
-		vhost_user_set_vring_addr(dev, &msg);
+		vhost_user_set_vring_addr(&dev, &msg);
 		break;
 	case VHOST_USER_SET_VRING_BASE:
 		vhost_user_set_vring_base(dev, &msg);
@@ -1063,11 +1295,11 @@ vhost_user_msg_handler(int vid, int fd)
 	case VHOST_USER_GET_VRING_BASE:
 		vhost_user_get_vring_base(dev, &msg);
 		msg.size = sizeof(msg.payload.state);
-		send_vhost_message(fd, &msg);
+		send_vhost_reply(fd, &msg);
 		break;
 
 	case VHOST_USER_SET_VRING_KICK:
-		vhost_user_set_vring_kick(dev, &msg);
+		vhost_user_set_vring_kick(&dev, &msg);
 		break;
 	case VHOST_USER_SET_VRING_CALL:
 		vhost_user_set_vring_call(dev, &msg);
@@ -1082,7 +1314,7 @@ vhost_user_msg_handler(int vid, int fd)
 	case VHOST_USER_GET_QUEUE_NUM:
 		msg.payload.u64 = VHOST_MAX_QUEUE_PAIRS;
 		msg.size = sizeof(msg.payload.u64);
-		send_vhost_message(fd, &msg);
+		send_vhost_reply(fd, &msg);
 		break;
 
 	case VHOST_USER_SET_VRING_ENABLE:
@@ -1096,6 +1328,14 @@ vhost_user_msg_handler(int vid, int fd)
 		ret = vhost_user_net_set_mtu(dev, &msg);
 		break;
 
+	case VHOST_USER_SET_SLAVE_REQ_FD:
+		ret = vhost_user_set_req_fd(dev, &msg);
+		break;
+
+	case VHOST_USER_IOTLB_MSG:
+		ret = vhost_user_iotlb_msg(&dev, &msg);
+		break;
+
 	default:
 		ret = -1;
 		break;
@@ -1105,7 +1345,7 @@ vhost_user_msg_handler(int vid, int fd)
 	if (msg.flags & VHOST_USER_NEED_REPLY) {
 		msg.payload.u64 = !!ret;
 		msg.size = sizeof(msg.payload.u64);
-		send_vhost_message(fd, &msg);
+		send_vhost_reply(fd, &msg);
 	}
 
 	if (!(dev->flags & VIRTIO_DEV_RUNNING) && virtio_is_ready(dev)) {
@@ -1124,3 +1364,29 @@ vhost_user_msg_handler(int vid, int fd)
 
 	return 0;
 }
+
+int
+vhost_user_iotlb_miss(struct virtio_net *dev, uint64_t iova, uint8_t perm)
+{
+	int ret;
+	struct VhostUserMsg msg = {
+		.request.slave = VHOST_USER_SLAVE_IOTLB_MSG,
+		.flags = VHOST_USER_VERSION,
+		.size = sizeof(msg.payload.iotlb),
+		.payload.iotlb = {
+			.iova = iova,
+			.perm = perm,
+			.type = VHOST_IOTLB_MISS,
+		},
+	};
+
+	ret = send_vhost_message(dev->slave_req_fd, &msg);
+	if (ret < 0) {
+		RTE_LOG(ERR, VHOST_CONFIG,
+				"Failed to send IOTLB miss message (%d)\n",
+				ret);
+		return ret;
+	}
+
+	return 0;
+}