vhost: fix guest notification setting
[dpdk.git] / lib / librte_vhost / vhost.c
index 1cbe948..8f20a08 100644 (file)
@@ -57,7 +57,7 @@ __vhost_iova_to_vva(struct virtio_net *dev, struct vhost_virtqueue *vq,
 
                vhost_user_iotlb_pending_insert(vq, iova, perm);
                if (vhost_user_iotlb_miss(dev, iova, perm)) {
-                       RTE_LOG(ERR, VHOST_CONFIG,
+                       VHOST_LOG_CONFIG(ERR,
                                "IOTLB miss req failed for IOVA 0x%" PRIx64 "\n",
                                iova);
                        vhost_user_iotlb_pending_remove(vq, iova, 1, perm);
@@ -124,7 +124,7 @@ __vhost_log_write_iova(struct virtio_net *dev, struct vhost_virtqueue *vq,
 
        hva = __vhost_iova_to_vva(dev, vq, iova, &map_len, VHOST_ACCESS_RW);
        if (map_len != len) {
-               RTE_LOG(ERR, VHOST_CONFIG,
+               VHOST_LOG_DATA(ERR,
                        "Failed to write log for IOVA 0x%" PRIx64 ". No IOTLB entry found\n",
                        iova);
                return;
@@ -229,7 +229,7 @@ __vhost_log_cache_write_iova(struct virtio_net *dev, struct vhost_virtqueue *vq,
 
        hva = __vhost_iova_to_vva(dev, vq, iova, &map_len, VHOST_ACCESS_RW);
        if (map_len != len) {
-               RTE_LOG(ERR, VHOST_CONFIG,
+               VHOST_LOG_DATA(ERR,
                        "Failed to write log for IOVA 0x%" PRIx64 ". No IOTLB entry found\n",
                        iova);
                return;
@@ -329,8 +329,13 @@ free_vq(struct virtio_net *dev, struct vhost_virtqueue *vq)
 {
        if (vq_is_packed(dev))
                rte_free(vq->shadow_used_packed);
-       else
+       else {
                rte_free(vq->shadow_used_split);
+               if (vq->async_pkts_pending)
+                       rte_free(vq->async_pkts_pending);
+               if (vq->async_pending_info)
+                       rte_free(vq->async_pending_info);
+       }
        rte_free(vq->batch_copy_elems);
        rte_mempool_free(vq->iotlb_pool);
        rte_free(vq);
@@ -350,6 +355,57 @@ free_device(struct virtio_net *dev)
        rte_free(dev);
 }
 
+static __rte_always_inline int
+log_translate(struct virtio_net *dev, struct vhost_virtqueue *vq)
+{
+       if (likely(!(vq->ring_addrs.flags & (1 << VHOST_VRING_F_LOG))))
+               return 0;
+
+       vq->log_guest_addr = translate_log_addr(dev, vq,
+                                               vq->ring_addrs.log_guest_addr);
+       if (vq->log_guest_addr == 0)
+               return -1;
+
+       return 0;
+}
+
+/*
+ * Converts vring log address to GPA
+ * If IOMMU is enabled, the log address is IOVA
+ * If IOMMU not enabled, the log address is already GPA
+ *
+ * Caller should have iotlb_lock read-locked
+ */
+uint64_t
+translate_log_addr(struct virtio_net *dev, struct vhost_virtqueue *vq,
+               uint64_t log_addr)
+{
+       if (dev->features & (1ULL << VIRTIO_F_IOMMU_PLATFORM)) {
+               const uint64_t exp_size = sizeof(uint64_t);
+               uint64_t hva, gpa;
+               uint64_t size = exp_size;
+
+               hva = vhost_iova_to_vva(dev, vq, log_addr,
+                                       &size, VHOST_ACCESS_RW);
+
+               if (size != exp_size)
+                       return 0;
+
+               gpa = hva_to_gpa(dev, hva, exp_size);
+               if (!gpa) {
+                       VHOST_LOG_CONFIG(ERR,
+                               "VQ: Failed to find GPA for log_addr: 0x%"
+                               PRIx64 " hva: 0x%" PRIx64 "\n",
+                               log_addr, hva);
+                       return 0;
+               }
+               return gpa;
+
+       } else
+               return log_addr;
+}
+
+/* Caller should have iotlb_lock read-locked */
 static int
 vring_translate_split(struct virtio_net *dev, struct vhost_virtqueue *vq)
 {
@@ -388,6 +444,7 @@ vring_translate_split(struct virtio_net *dev, struct vhost_virtqueue *vq)
        return 0;
 }
 
+/* Caller should have iotlb_lock read-locked */
 static int
 vring_translate_packed(struct virtio_net *dev, struct vhost_virtqueue *vq)
 {
@@ -434,6 +491,10 @@ vring_translate(struct virtio_net *dev, struct vhost_virtqueue *vq)
                if (vring_translate_split(dev, vq) < 0)
                        return -1;
        }
+
+       if (log_translate(dev, vq) < 0)
+               return -1;
+
        vq->access_ok = 1;
 
        return 0;
@@ -461,7 +522,7 @@ init_vring_queue(struct virtio_net *dev, uint32_t vring_idx)
        struct vhost_virtqueue *vq;
 
        if (vring_idx >= VHOST_MAX_VRING) {
-               RTE_LOG(ERR, VHOST_CONFIG,
+               VHOST_LOG_CONFIG(ERR,
                                "Failed not init vring, out of bound (%d)\n",
                                vring_idx);
                return;
@@ -473,6 +534,7 @@ init_vring_queue(struct virtio_net *dev, uint32_t vring_idx)
 
        vq->kickfd = VIRTIO_UNINITIALIZED_EVENTFD;
        vq->callfd = VIRTIO_UNINITIALIZED_EVENTFD;
+       vq->notif_enable = VIRTIO_UNINITIALIZED_NOTIF;
 
        vhost_user_iotlb_init(dev, vring_idx);
        /* Backends are set to -1 indicating an inactive device. */
@@ -488,7 +550,7 @@ reset_vring_queue(struct virtio_net *dev, uint32_t vring_idx)
        int callfd;
 
        if (vring_idx >= VHOST_MAX_VRING) {
-               RTE_LOG(ERR, VHOST_CONFIG,
+               VHOST_LOG_CONFIG(ERR,
                                "Failed not init vring, out of bound (%d)\n",
                                vring_idx);
                return;
@@ -507,7 +569,7 @@ alloc_vring_queue(struct virtio_net *dev, uint32_t vring_idx)
 
        vq = rte_malloc(NULL, sizeof(struct vhost_virtqueue), 0);
        if (vq == NULL) {
-               RTE_LOG(ERR, VHOST_CONFIG,
+               VHOST_LOG_CONFIG(ERR,
                        "Failed to allocate memory for vring:%u.\n", vring_idx);
                return -1;
        }
@@ -558,14 +620,14 @@ vhost_new_device(void)
        }
 
        if (i == MAX_VHOST_DEVICE) {
-               RTE_LOG(ERR, VHOST_CONFIG,
+               VHOST_LOG_CONFIG(ERR,
                        "Failed to find a free slot for new device.\n");
                return -1;
        }
 
        dev = rte_zmalloc(NULL, sizeof(struct virtio_net), 0);
        if (dev == NULL) {
-               RTE_LOG(ERR, VHOST_CONFIG,
+               VHOST_LOG_CONFIG(ERR,
                        "Failed to allocate memory for new dev.\n");
                return -1;
        }
@@ -574,7 +636,6 @@ vhost_new_device(void)
        dev->vid = i;
        dev->flags = VIRTIO_DEV_BUILTIN_VIRTIO_NET;
        dev->slave_req_fd = -1;
-       dev->vdpa_dev_id = -1;
        dev->postcopy_ufd = -1;
        rte_spinlock_init(&dev->slave_req_lock);
 
@@ -585,12 +646,10 @@ void
 vhost_destroy_device_notify(struct virtio_net *dev)
 {
        struct rte_vdpa_device *vdpa_dev;
-       int did;
 
        if (dev->flags & VIRTIO_DEV_RUNNING) {
-               did = dev->vdpa_dev_id;
-               vdpa_dev = rte_vdpa_get_device(did);
-               if (vdpa_dev && vdpa_dev->ops->dev_close)
+               vdpa_dev = dev->vdpa_dev;
+               if (vdpa_dev)
                        vdpa_dev->ops->dev_close(dev->vid);
                dev->flags &= ~VIRTIO_DEV_RUNNING;
                dev->notify_ops->destroy_device(dev->vid);
@@ -618,17 +677,14 @@ vhost_destroy_device(int vid)
 }
 
 void
-vhost_attach_vdpa_device(int vid, int did)
+vhost_attach_vdpa_device(int vid, struct rte_vdpa_device *vdpa_dev)
 {
        struct virtio_net *dev = get_device(vid);
 
        if (dev == NULL)
                return;
 
-       if (rte_vdpa_get_device(did) == NULL)
-               return;
-
-       dev->vdpa_dev_id = did;
+       dev->vdpa_dev = vdpa_dev;
 }
 
 void
@@ -657,6 +713,7 @@ vhost_enable_dequeue_zero_copy(int vid)
                return;
 
        dev->dequeue_zero_copy = 1;
+       VHOST_LOG_CONFIG(INFO, "dequeue zero copy is enabled\n");
 }
 
 void
@@ -728,7 +785,7 @@ rte_vhost_get_numa_node(int vid)
        ret = get_mempolicy(&numa_node, NULL, 0, dev,
                            MPOL_F_NODE | MPOL_F_ADDR);
        if (ret < 0) {
-               RTE_LOG(ERR, VHOST_CONFIG,
+               VHOST_LOG_CONFIG(ERR,
                        "(%d) failed to query numa node: %s\n",
                        vid, rte_strerror(errno));
                return -1;
@@ -1255,6 +1312,23 @@ vhost_enable_notify_packed(struct virtio_net *dev,
        return 0;
 }
 
+int
+vhost_enable_guest_notification(struct virtio_net *dev,
+               struct vhost_virtqueue *vq, int enable)
+{
+       /*
+        * If the virtqueue is not ready yet, it will be applied
+        * when it will become ready.
+        */
+       if (!vq->ready)
+               return 0;
+
+       if (vq_is_packed(dev))
+               return vhost_enable_notify_packed(dev, vq, enable);
+       else
+               return vhost_enable_notify_split(dev, vq, enable);
+}
+
 int
 rte_vhost_enable_guest_notification(int vid, uint16_t queue_id, int enable)
 {
@@ -1269,10 +1343,8 @@ rte_vhost_enable_guest_notification(int vid, uint16_t queue_id, int enable)
 
        rte_spinlock_lock(&vq->access_lock);
 
-       if (vq_is_packed(dev))
-               ret = vhost_enable_notify_packed(dev, vq, enable);
-       else
-               ret = vhost_enable_notify_split(dev, vq, enable);
+       vq->notif_enable = enable;
+       ret = vhost_enable_guest_notification(dev, vq, enable);
 
        rte_spinlock_unlock(&vq->access_lock);
 
@@ -1322,7 +1394,7 @@ rte_vhost_rx_queue_count(int vid, uint16_t qid)
                return 0;
 
        if (unlikely(qid >= dev->nr_vring || (qid & 1) == 0)) {
-               RTE_LOG(ERR, VHOST_DATA, "(%d) %s: invalid virtqueue idx %d.\n",
+               VHOST_LOG_DATA(ERR, "(%d) %s: invalid virtqueue idx %d.\n",
                        dev->vid, __func__, qid);
                return 0;
        }
@@ -1343,14 +1415,15 @@ out:
        return ret;
 }
 
-int rte_vhost_get_vdpa_device_id(int vid)
+struct rte_vdpa_device *
+rte_vhost_get_vdpa_device(int vid)
 {
        struct virtio_net *dev = get_device(vid);
 
        if (dev == NULL)
-               return -1;
+               return NULL;
 
-       return dev->vdpa_dev_id;
+       return dev->vdpa_dev;
 }
 
 int rte_vhost_get_log_base(int vid, uint64_t *log_base,
@@ -1457,3 +1530,125 @@ int rte_vhost_extern_callback_register(int vid,
        dev->extern_data = ctx;
        return 0;
 }
+
+int rte_vhost_async_channel_register(int vid, uint16_t queue_id,
+                                       uint32_t features,
+                                       struct rte_vhost_async_channel_ops *ops)
+{
+       struct vhost_virtqueue *vq;
+       struct virtio_net *dev = get_device(vid);
+       struct rte_vhost_async_features f;
+
+       if (dev == NULL || ops == NULL)
+               return -1;
+
+       f.intval = features;
+
+       vq = dev->virtqueue[queue_id];
+
+       if (unlikely(vq == NULL || !dev->async_copy))
+               return -1;
+
+       /* packed queue is not supported */
+       if (unlikely(vq_is_packed(dev) || !f.async_inorder)) {
+               VHOST_LOG_CONFIG(ERR,
+                       "async copy is not supported on packed queue or non-inorder mode "
+                       "(vid %d, qid: %d)\n", vid, queue_id);
+               return -1;
+       }
+
+       if (unlikely(ops->check_completed_copies == NULL ||
+               ops->transfer_data == NULL))
+               return -1;
+
+       rte_spinlock_lock(&vq->access_lock);
+
+       if (unlikely(vq->async_registered)) {
+               VHOST_LOG_CONFIG(ERR,
+                       "async register failed: channel already registered "
+                       "(vid %d, qid: %d)\n", vid, queue_id);
+               goto reg_out;
+       }
+
+       vq->async_pkts_pending = rte_malloc(NULL,
+                       vq->size * sizeof(uintptr_t),
+                       RTE_CACHE_LINE_SIZE);
+       vq->async_pending_info = rte_malloc(NULL,
+                       vq->size * sizeof(uint64_t),
+                       RTE_CACHE_LINE_SIZE);
+       if (!vq->async_pkts_pending || !vq->async_pending_info) {
+               if (vq->async_pkts_pending)
+                       rte_free(vq->async_pkts_pending);
+
+               if (vq->async_pending_info)
+                       rte_free(vq->async_pending_info);
+
+               VHOST_LOG_CONFIG(ERR,
+                               "async register failed: cannot allocate memory for vq data "
+                               "(vid %d, qid: %d)\n", vid, queue_id);
+               goto reg_out;
+       }
+
+       vq->async_ops.check_completed_copies = ops->check_completed_copies;
+       vq->async_ops.transfer_data = ops->transfer_data;
+
+       vq->async_inorder = f.async_inorder;
+       vq->async_threshold = f.async_threshold;
+
+       vq->async_registered = true;
+
+reg_out:
+       rte_spinlock_unlock(&vq->access_lock);
+
+       return 0;
+}
+
+int rte_vhost_async_channel_unregister(int vid, uint16_t queue_id)
+{
+       struct vhost_virtqueue *vq;
+       struct virtio_net *dev = get_device(vid);
+       int ret = -1;
+
+       if (dev == NULL)
+               return ret;
+
+       vq = dev->virtqueue[queue_id];
+
+       if (vq == NULL)
+               return ret;
+
+       ret = 0;
+       rte_spinlock_lock(&vq->access_lock);
+
+       if (!vq->async_registered)
+               goto out;
+
+       if (vq->async_pkts_inflight_n) {
+               VHOST_LOG_CONFIG(ERR, "Failed to unregister async channel. "
+                       "async inflight packets must be completed before unregistration.\n");
+               ret = -1;
+               goto out;
+       }
+
+       if (vq->async_pkts_pending) {
+               rte_free(vq->async_pkts_pending);
+               vq->async_pkts_pending = NULL;
+       }
+
+       if (vq->async_pending_info) {
+               rte_free(vq->async_pending_info);
+               vq->async_pending_info = NULL;
+       }
+
+       vq->async_ops.transfer_data = NULL;
+       vq->async_ops.check_completed_copies = NULL;
+       vq->async_registered = false;
+
+out:
+       rte_spinlock_unlock(&vq->access_lock);
+
+       return ret;
+}
+
+RTE_LOG_REGISTER(vhost_config_log_level, lib.vhost.config, INFO);
+RTE_LOG_REGISTER(vhost_data_log_level, lib.vhost.data, WARNING);