vhost: improve device readiness notifications
[dpdk.git] / lib / librte_vhost / vhost_user.c
index 6dbba04..4ca8728 100644 (file)
@@ -206,7 +206,7 @@ vhost_backend_cleanup(struct virtio_net *dev)
                        dev->inflight_info->addr = NULL;
                }
 
-               if (dev->inflight_info->fd > 0) {
+               if (dev->inflight_info->fd >= 0) {
                        close(dev->inflight_info->fd);
                        dev->inflight_info->fd = -1;
                }
@@ -228,6 +228,20 @@ vhost_backend_cleanup(struct virtio_net *dev)
        dev->postcopy_listening = 0;
 }
 
+static void
+vhost_user_notify_queue_state(struct virtio_net *dev, uint16_t index,
+                             int enable)
+{
+       struct rte_vdpa_device *vdpa_dev = dev->vdpa_dev;
+
+       if (vdpa_dev && vdpa_dev->ops->set_vring_state)
+               vdpa_dev->ops->set_vring_state(dev->vid, index, enable);
+
+       if (dev->notify_ops->vring_state_changed)
+               dev->notify_ops->vring_state_changed(dev->vid,
+                               index, enable);
+}
+
 /*
  * This function just returns success at the moment unless
  * the device hasn't been initialised.
@@ -315,7 +329,6 @@ vhost_user_set_features(struct virtio_net **pdev, struct VhostUserMsg *msg,
        uint64_t features = msg->payload.u64;
        uint64_t vhost_features = 0;
        struct rte_vdpa_device *vdpa_dev;
-       int did = -1;
 
        if (validate_msg_fds(msg, 0) != 0)
                return RTE_VHOST_MSG_RESULT_ERR;
@@ -384,8 +397,7 @@ vhost_user_set_features(struct virtio_net **pdev, struct VhostUserMsg *msg,
                }
        }
 
-       did = dev->vdpa_dev_id;
-       vdpa_dev = rte_vdpa_get_device(did);
+       vdpa_dev = dev->vdpa_dev;
        if (vdpa_dev && vdpa_dev->ops->set_features)
                vdpa_dev->ops->set_features(dev->vid);
 
@@ -1306,27 +1318,31 @@ vq_is_ready(struct virtio_net *dev, struct vhost_virtqueue *vq)
 
        return rings_ok &&
               vq->kickfd != VIRTIO_UNINITIALIZED_EVENTFD &&
-              vq->callfd != VIRTIO_UNINITIALIZED_EVENTFD;
+              vq->callfd != VIRTIO_UNINITIALIZED_EVENTFD &&
+              vq->enabled;
 }
 
+#define VIRTIO_DEV_NUM_VQS_TO_BE_READY 2u
+
 static int
 virtio_is_ready(struct virtio_net *dev)
 {
        struct vhost_virtqueue *vq;
        uint32_t i;
 
-       if (dev->nr_vring == 0)
+       if (dev->nr_vring < VIRTIO_DEV_NUM_VQS_TO_BE_READY)
                return 0;
 
-       for (i = 0; i < dev->nr_vring; i++) {
+       for (i = 0; i < VIRTIO_DEV_NUM_VQS_TO_BE_READY; i++) {
                vq = dev->virtqueue[i];
 
                if (!vq_is_ready(dev, vq))
                        return 0;
        }
 
-       VHOST_LOG_CONFIG(INFO,
-               "virtio is now ready for processing.\n");
+       if (!(dev->flags & VIRTIO_DEV_RUNNING))
+               VHOST_LOG_CONFIG(INFO,
+                       "virtio is now ready for processing.\n");
        return 1;
 }
 
@@ -1417,6 +1433,7 @@ vhost_user_get_inflight_fd(struct virtio_net **pdev,
                                "failed to alloc dev inflight area\n");
                        return RTE_VHOST_MSG_RESULT_ERR;
                }
+               dev->inflight_info->fd = -1;
        }
 
        num_queues = msg->payload.inflight.num_queues;
@@ -1442,6 +1459,16 @@ vhost_user_get_inflight_fd(struct virtio_net **pdev,
        }
        memset(addr, 0, mmap_size);
 
+       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;
+       }
+
        dev->inflight_info->addr = addr;
        dev->inflight_info->size = msg->payload.inflight.mmap_size = mmap_size;
        dev->inflight_info->fd = msg->fds[0] = fd;
@@ -1524,10 +1551,13 @@ vhost_user_set_inflight_fd(struct virtio_net **pdev, VhostUserMsg *msg,
                                "failed to alloc dev inflight area\n");
                        return RTE_VHOST_MSG_RESULT_ERR;
                }
+               dev->inflight_info->fd = -1;
        }
 
-       if (dev->inflight_info->addr)
+       if (dev->inflight_info->addr) {
                munmap(dev->inflight_info->addr, dev->inflight_info->size);
+               dev->inflight_info->addr = NULL;
+       }
 
        addr = mmap(0, mmap_size, PROT_READ | PROT_WRITE, MAP_SHARED,
                    fd, mmap_offset);
@@ -1536,8 +1566,10 @@ vhost_user_set_inflight_fd(struct virtio_net **pdev, VhostUserMsg *msg,
                return RTE_VHOST_MSG_RESULT_ERR;
        }
 
-       if (dev->inflight_info->fd)
+       if (dev->inflight_info->fd >= 0) {
                close(dev->inflight_info->fd);
+               dev->inflight_info->fd = -1;
+       }
 
        dev->inflight_info->fd = fd;
        dev->inflight_info->addr = addr;
@@ -1954,8 +1986,6 @@ vhost_user_set_vring_enable(struct virtio_net **pdev,
        struct virtio_net *dev = *pdev;
        int enable = (int)msg->payload.state.num;
        int index = (int)msg->payload.state.index;
-       struct rte_vdpa_device *vdpa_dev;
-       int did = -1;
 
        if (validate_msg_fds(msg, 0) != 0)
                return RTE_VHOST_MSG_RESULT_ERR;
@@ -1964,15 +1994,6 @@ vhost_user_set_vring_enable(struct virtio_net **pdev,
                "set queue enable: %d to qp idx: %d\n",
                enable, index);
 
-       did = dev->vdpa_dev_id;
-       vdpa_dev = rte_vdpa_get_device(did);
-       if (vdpa_dev && vdpa_dev->ops->set_vring_state)
-               vdpa_dev->ops->set_vring_state(dev->vid, index, enable);
-
-       if (dev->notify_ops->vring_state_changed)
-               dev->notify_ops->vring_state_changed(dev->vid,
-                               index, enable);
-
        /* On disable, rings have to be stopped being processed. */
        if (!enable && dev->dequeue_zero_copy)
                drain_zmbuf_list(dev->virtqueue[index]);
@@ -1996,15 +2017,6 @@ vhost_user_get_protocol_features(struct virtio_net **pdev,
        rte_vhost_driver_get_features(dev->ifname, &features);
        rte_vhost_driver_get_protocol_features(dev->ifname, &protocol_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);
        msg->fd_num = 0;
@@ -2140,7 +2152,6 @@ vhost_user_send_rarp(struct virtio_net **pdev, struct VhostUserMsg *msg,
        struct virtio_net *dev = *pdev;
        uint8_t *mac = (uint8_t *)&msg->payload.u64;
        struct rte_vdpa_device *vdpa_dev;
-       int did = -1;
 
        if (validate_msg_fds(msg, 0) != 0)
                return RTE_VHOST_MSG_RESULT_ERR;
@@ -2158,8 +2169,7 @@ vhost_user_send_rarp(struct virtio_net **pdev, struct VhostUserMsg *msg,
         * copied before the flag is set.
         */
        __atomic_store_n(&dev->broadcast_rarp, 1, __ATOMIC_RELEASE);
-       did = dev->vdpa_dev_id;
-       vdpa_dev = rte_vdpa_get_device(did);
+       vdpa_dev = dev->vdpa_dev;
        if (vdpa_dev && vdpa_dev->ops->migration_done)
                vdpa_dev->ops->migration_done(dev->vid);
 
@@ -2606,11 +2616,11 @@ vhost_user_msg_handler(int vid, int fd)
        struct virtio_net *dev;
        struct VhostUserMsg msg;
        struct rte_vdpa_device *vdpa_dev;
-       int did = -1;
        int ret;
        int unlock_required = 0;
        bool handled;
        int request;
+       uint32_t i;
 
        dev = get_device(vid);
        if (dev == NULL)
@@ -2683,8 +2693,10 @@ vhost_user_msg_handler(int vid, int fd)
        case VHOST_USER_SEND_RARP:
        case VHOST_USER_NET_SET_MTU:
        case VHOST_USER_SET_SLAVE_REQ_FD:
-               vhost_user_lock_all_queue_pairs(dev);
-               unlock_required = 1;
+               if (!(dev->flags & VIRTIO_DEV_VDPA_CONFIGURED)) {
+                       vhost_user_lock_all_queue_pairs(dev);
+                       unlock_required = 1;
+               }
                break;
        default:
                break;
@@ -2784,6 +2796,17 @@ skip_to_post_handle:
                return -1;
        }
 
+       for (i = 0; i < dev->nr_vring; i++) {
+               struct vhost_virtqueue *vq = dev->virtqueue[i];
+               bool cur_ready = vq_is_ready(dev, vq);
+
+               if (cur_ready != (vq && vq->ready)) {
+                       vhost_user_notify_queue_state(dev, i, cur_ready);
+                       vq->ready = cur_ready;
+               }
+       }
+
+
        if (!(dev->flags & VIRTIO_DEV_RUNNING) && virtio_is_ready(dev)) {
                dev->flags |= VIRTIO_DEV_READY;
 
@@ -2798,11 +2821,9 @@ skip_to_post_handle:
                }
        }
 
-       did = dev->vdpa_dev_id;
-       vdpa_dev = rte_vdpa_get_device(did);
+       vdpa_dev = dev->vdpa_dev;
        if (vdpa_dev && virtio_is_ready(dev) &&
-                       !(dev->flags & VIRTIO_DEV_VDPA_CONFIGURED) &&
-                       msg.request.master == VHOST_USER_SET_VRING_CALL) {
+           !(dev->flags & VIRTIO_DEV_VDPA_CONFIGURED)) {
                if (vdpa_dev->ops->dev_conf)
                        vdpa_dev->ops->dev_conf(dev->vid);
                dev->flags |= VIRTIO_DEV_VDPA_CONFIGURED;
@@ -2944,21 +2965,21 @@ static int vhost_user_slave_set_vring_host_notifier(struct virtio_net *dev,
        return process_slave_message_reply(dev, &msg);
 }
 
-int rte_vhost_host_notifier_ctrl(int vid, bool enable)
+int rte_vhost_host_notifier_ctrl(int vid, uint16_t qid, bool enable)
 {
        struct virtio_net *dev;
        struct rte_vdpa_device *vdpa_dev;
-       int vfio_device_fd, did, ret = 0;
+       int vfio_device_fd, ret = 0;
        uint64_t offset, size;
-       unsigned int i;
+       unsigned int i, q_start, q_last;
 
        dev = get_device(vid);
        if (!dev)
                return -ENODEV;
 
-       did = dev->vdpa_dev_id;
-       if (did < 0)
-               return -EINVAL;
+       vdpa_dev = dev->vdpa_dev;
+       if (vdpa_dev == NULL)
+               return -ENODEV;
 
        if (!(dev->features & (1ULL << VIRTIO_F_VERSION_1)) ||
            !(dev->features & (1ULL << VHOST_USER_F_PROTOCOL_FEATURES)) ||
@@ -2970,9 +2991,15 @@ int rte_vhost_host_notifier_ctrl(int vid, bool enable)
                        (1ULL << VHOST_USER_PROTOCOL_F_HOST_NOTIFIER)))
                return -ENOTSUP;
 
-       vdpa_dev = rte_vdpa_get_device(did);
-       if (!vdpa_dev)
-               return -ENODEV;
+       if (qid == RTE_VHOST_QUEUE_ALL) {
+               q_start = 0;
+               q_last = dev->nr_vring - 1;
+       } else {
+               if (qid >= dev->nr_vring)
+                       return -EINVAL;
+               q_start = qid;
+               q_last = qid;
+       }
 
        RTE_FUNC_PTR_OR_ERR_RET(vdpa_dev->ops->get_vfio_device_fd, -ENOTSUP);
        RTE_FUNC_PTR_OR_ERR_RET(vdpa_dev->ops->get_notify_area, -ENOTSUP);
@@ -2982,7 +3009,7 @@ int rte_vhost_host_notifier_ctrl(int vid, bool enable)
                return -ENOTSUP;
 
        if (enable) {
-               for (i = 0; i < dev->nr_vring; i++) {
+               for (i = q_start; i <= q_last; i++) {
                        if (vdpa_dev->ops->get_notify_area(vid, i, &offset,
                                        &size) < 0) {
                                ret = -ENOTSUP;
@@ -2997,7 +3024,7 @@ int rte_vhost_host_notifier_ctrl(int vid, bool enable)
                }
        } else {
 disable:
-               for (i = 0; i < dev->nr_vring; i++) {
+               for (i = q_start; i <= q_last; i++) {
                        vhost_user_slave_set_vring_host_notifier(dev, i, -1,
                                        0, 0);
                }