block/iscsi: handle failure on malloc of the allocationmap

Signed-off-by: Peter Lieven <pl@kamp.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

block/iscsi.c

@@ -317,6 +317,13 @@ static bool is_request_lun_aligned(int64_t sector_num, int nb_sectors,
     return 1;
 }
 
+static unsigned long *iscsi_allocationmap_init(IscsiLun *iscsilun)
+{
+    return bitmap_try_new(DIV_ROUND_UP(sector_lun2qemu(iscsilun->num_blocks,
+                                                       iscsilun),
+                                       iscsilun->cluster_sectors));
+}
+
 static void iscsi_allocationmap_set(IscsiLun *iscsilun, int64_t sector_num,
                                     int nb_sectors)
 {
@@ -1403,9 +1410,10 @@ static int iscsi_open(BlockDriverState *bs, QDict *options, int flags,
         iscsilun->cluster_sectors = (iscsilun->bl.opt_unmap_gran *
                                      iscsilun->block_size) >> BDRV_SECTOR_BITS;
         if (iscsilun->lbprz && !(bs->open_flags & BDRV_O_NOCACHE)) {
-            iscsilun->allocationmap =
-                bitmap_new(DIV_ROUND_UP(bs->total_sectors,
-                                        iscsilun->cluster_sectors));
+            iscsilun->allocationmap = iscsi_allocationmap_init(iscsilun);
+            if (iscsilun->allocationmap == NULL) {
+                ret = -ENOMEM;
+            }
         }
     }
 
@@ -1498,10 +1506,7 @@ static int iscsi_truncate(BlockDriverState *bs, int64_t offset)
 
     if (iscsilun->allocationmap != NULL) {
         g_free(iscsilun->allocationmap);
-        iscsilun->allocationmap =
-            bitmap_new(DIV_ROUND_UP(sector_lun2qemu(iscsilun->num_blocks,
-                                                    iscsilun),
-                                    iscsilun->cluster_sectors));
+        iscsilun->allocationmap = iscsi_allocationmap_init(iscsilun);
     }
 
     return 0;

hw/s390x/s390-virtio-bus.c

@@ -159,8 +159,9 @@ static int s390_virtio_net_init(VirtIOS390Device *s390_dev)
 static void s390_virtio_net_instance_init(Object *obj)
 {
     VirtIONetS390 *dev = VIRTIO_NET_S390(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_NET);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_NET);
 }
 
 static int s390_virtio_blk_init(VirtIOS390Device *s390_dev)
@@ -177,10 +178,9 @@ static int s390_virtio_blk_init(VirtIOS390Device *s390_dev)
 static void s390_virtio_blk_instance_init(Object *obj)
 {
     VirtIOBlkS390 *dev = VIRTIO_BLK_S390(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BLK);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-    object_unref(OBJECT(&dev->vdev));
-    qdev_alias_all_properties(DEVICE(&dev->vdev), obj);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_BLK);
     object_property_add_alias(obj, "iothread", OBJECT(&dev->vdev),"iothread",
                               &error_abort);
 }
@@ -222,8 +222,9 @@ static int s390_virtio_serial_init(VirtIOS390Device *s390_dev)
 static void s390_virtio_serial_instance_init(Object *obj)
 {
     VirtIOSerialS390 *dev = VIRTIO_SERIAL_S390(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_SERIAL);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_SERIAL);
 }
 
 static int s390_virtio_scsi_init(VirtIOS390Device *s390_dev)
@@ -254,8 +255,9 @@ static int s390_virtio_scsi_init(VirtIOS390Device *s390_dev)
 static void s390_virtio_scsi_instance_init(Object *obj)
 {
     VirtIOSCSIS390 *dev = VIRTIO_SCSI_S390(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_SCSI);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_SCSI);
 }
 
 #ifdef CONFIG_VHOST_SCSI
@@ -275,8 +277,9 @@ static int s390_vhost_scsi_init(VirtIOS390Device *s390_dev)
 static void s390_vhost_scsi_instance_init(Object *obj)
 {
     VHostSCSIS390 *dev = VHOST_SCSI_S390(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VHOST_SCSI);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VHOST_SCSI);
 }
 #endif
 
@@ -301,8 +304,9 @@ static int s390_virtio_rng_init(VirtIOS390Device *s390_dev)
 static void s390_virtio_rng_instance_init(Object *obj)
 {
     VirtIORNGS390 *dev = VIRTIO_RNG_S390(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_RNG);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_RNG);
     object_property_add_link(obj, "rng", TYPE_RNG_BACKEND,
                              (Object **)&dev->vdev.conf.rng,
                              qdev_prop_allow_set_link_before_realize,
@@ -493,10 +497,8 @@ static unsigned virtio_s390_get_features(DeviceState *d)
 /**************** S390 Virtio Bus Device Descriptions *******************/
 
 static Property s390_virtio_net_properties[] = {
-    DEFINE_NIC_PROPERTIES(VirtIONetS390, vdev.nic_conf),
     DEFINE_VIRTIO_COMMON_FEATURES(VirtIOS390Device, host_features),
     DEFINE_VIRTIO_NET_FEATURES(VirtIOS390Device, host_features),
-    DEFINE_VIRTIO_NET_PROPERTIES(VirtIONetS390, vdev.net_conf),
     DEFINE_PROP_END_OF_LIST(),
 };
 
@@ -533,7 +535,6 @@ static const TypeInfo s390_virtio_blk = {
 };
 
 static Property s390_virtio_serial_properties[] = {
-    DEFINE_VIRTIO_SERIAL_PROPERTIES(VirtIOSerialS390, vdev.serial),
     DEFINE_PROP_END_OF_LIST(),
 };
 
@@ -556,7 +557,6 @@ static const TypeInfo s390_virtio_serial = {
 
 static Property s390_virtio_rng_properties[] = {
     DEFINE_VIRTIO_COMMON_FEATURES(VirtIOS390Device, host_features),
-    DEFINE_VIRTIO_RNG_PROPERTIES(VirtIORNGS390, vdev.conf),
     DEFINE_PROP_END_OF_LIST(),
 };
 
@@ -614,7 +614,6 @@ static const TypeInfo virtio_s390_device_info = {
 };
 
 static Property s390_virtio_scsi_properties[] = {
-    DEFINE_VIRTIO_SCSI_PROPERTIES(VirtIOSCSIS390, vdev.parent_obj.conf),
     DEFINE_VIRTIO_COMMON_FEATURES(VirtIOS390Device, host_features),
     DEFINE_VIRTIO_SCSI_FEATURES(VirtIOS390Device, host_features),
     DEFINE_PROP_END_OF_LIST(),
@@ -640,7 +639,6 @@ static const TypeInfo s390_virtio_scsi = {
 #ifdef CONFIG_VHOST_SCSI
 static Property s390_vhost_scsi_properties[] = {
     DEFINE_VIRTIO_COMMON_FEATURES(VirtIOS390Device, host_features),
-    DEFINE_VHOST_SCSI_PROPERTIES(VHostSCSIS390, vdev.parent_obj.conf),
     DEFINE_PROP_END_OF_LIST(),
 };
 

hw/s390x/virtio-ccw.c

@@ -792,8 +792,9 @@ static int virtio_ccw_net_init(VirtioCcwDevice *ccw_dev)
 static void virtio_ccw_net_instance_init(Object *obj)
 {
     VirtIONetCcw *dev = VIRTIO_NET_CCW(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_NET);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_NET);
 }
 
 static int virtio_ccw_blk_init(VirtioCcwDevice *ccw_dev)
@@ -811,10 +812,9 @@ static int virtio_ccw_blk_init(VirtioCcwDevice *ccw_dev)
 static void virtio_ccw_blk_instance_init(Object *obj)
 {
     VirtIOBlkCcw *dev = VIRTIO_BLK_CCW(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BLK);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-    object_unref(OBJECT(&dev->vdev));
-    qdev_alias_all_properties(DEVICE(&dev->vdev), obj);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_BLK);
     object_property_add_alias(obj, "iothread", OBJECT(&dev->vdev),"iothread",
                               &error_abort);
 }
@@ -848,8 +848,9 @@ static int virtio_ccw_serial_init(VirtioCcwDevice *ccw_dev)
 static void virtio_ccw_serial_instance_init(Object *obj)
 {
     VirtioSerialCcw *dev = VIRTIO_SERIAL_CCW(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_SERIAL);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_SERIAL);
 }
 
 static int virtio_ccw_balloon_init(VirtioCcwDevice *ccw_dev)
@@ -896,7 +897,7 @@ static void virtio_ccw_balloon_instance_init(Object *obj)
     VirtIOBalloonCcw *dev = VIRTIO_BALLOON_CCW(obj);
     object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BALLOON);
     object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-
+    object_unref(OBJECT(&dev->vdev));
     object_property_add(obj, "guest-stats", "guest statistics",
                         balloon_ccw_stats_get_all, NULL, NULL, dev, NULL);
 
@@ -934,8 +935,11 @@ static int virtio_ccw_scsi_init(VirtioCcwDevice *ccw_dev)
 static void virtio_ccw_scsi_instance_init(Object *obj)
 {
     VirtIOSCSICcw *dev = VIRTIO_SCSI_CCW(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_SCSI);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_SCSI);
+    object_property_add_alias(obj, "iothread", OBJECT(&dev->vdev), "iothread",
+                              &error_abort);
 }
 
 #ifdef CONFIG_VHOST_SCSI
@@ -955,8 +959,9 @@ static int vhost_ccw_scsi_init(VirtioCcwDevice *ccw_dev)
 static void vhost_ccw_scsi_instance_init(Object *obj)
 {
     VHostSCSICcw *dev = VHOST_SCSI_CCW(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VHOST_SCSI);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VHOST_SCSI);
 }
 #endif
 
@@ -1374,8 +1379,6 @@ static int virtio_ccw_load_config(DeviceState *d, QEMUFile *f)
 static Property virtio_ccw_net_properties[] = {
     DEFINE_PROP_STRING("devno", VirtioCcwDevice, bus_id),
     DEFINE_VIRTIO_NET_FEATURES(VirtioCcwDevice, host_features[0]),
-    DEFINE_VIRTIO_NET_PROPERTIES(VirtIONetCcw, vdev.net_conf),
-    DEFINE_NIC_PROPERTIES(VirtIONetCcw, vdev.nic_conf),
     DEFINE_PROP_BIT("ioeventfd", VirtioCcwDevice, flags,
                     VIRTIO_CCW_FLAG_USE_IOEVENTFD_BIT, true),
     DEFINE_PROP_END_OF_LIST(),
@@ -1428,7 +1431,6 @@ static const TypeInfo virtio_ccw_blk = {
 
 static Property virtio_ccw_serial_properties[] = {
     DEFINE_PROP_STRING("devno", VirtioCcwDevice, bus_id),
-    DEFINE_VIRTIO_SERIAL_PROPERTIES(VirtioSerialCcw, vdev.serial),
     DEFINE_PROP_BIT("ioeventfd", VirtioCcwDevice, flags,
                     VIRTIO_CCW_FLAG_USE_IOEVENTFD_BIT, true),
     DEFINE_PROP_END_OF_LIST(),
@@ -1481,7 +1483,6 @@ static const TypeInfo virtio_ccw_balloon = {
 
 static Property virtio_ccw_scsi_properties[] = {
     DEFINE_PROP_STRING("devno", VirtioCcwDevice, bus_id),
-    DEFINE_VIRTIO_SCSI_PROPERTIES(VirtIOSCSICcw, vdev.parent_obj.conf),
     DEFINE_VIRTIO_SCSI_FEATURES(VirtioCcwDevice, host_features[0]),
     DEFINE_PROP_BIT("ioeventfd", VirtioCcwDevice, flags,
                     VIRTIO_CCW_FLAG_USE_IOEVENTFD_BIT, true),
@@ -1510,7 +1511,6 @@ static const TypeInfo virtio_ccw_scsi = {
 #ifdef CONFIG_VHOST_SCSI
 static Property vhost_ccw_scsi_properties[] = {
     DEFINE_PROP_STRING("devno", VirtioCcwDevice, bus_id),
-    DEFINE_VHOST_SCSI_PROPERTIES(VirtIOSCSICcw, vdev.parent_obj.conf),
     DEFINE_PROP_END_OF_LIST(),
 };
 
@@ -1537,8 +1537,9 @@ static const TypeInfo vhost_ccw_scsi = {
 static void virtio_ccw_rng_instance_init(Object *obj)
 {
     VirtIORNGCcw *dev = VIRTIO_RNG_CCW(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_RNG);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_RNG);
     object_property_add_link(obj, "rng", TYPE_RNG_BACKEND,
                              (Object **)&dev->vdev.conf.rng,
                              qdev_prop_allow_set_link_before_realize,
@@ -1547,7 +1548,6 @@ static void virtio_ccw_rng_instance_init(Object *obj)
 
 static Property virtio_ccw_rng_properties[] = {
     DEFINE_PROP_STRING("devno", VirtioCcwDevice, bus_id),
-    DEFINE_VIRTIO_RNG_PROPERTIES(VirtIORNGCcw, vdev.conf),
     DEFINE_PROP_BIT("ioeventfd", VirtioCcwDevice, flags,
                     VIRTIO_CCW_FLAG_USE_IOEVENTFD_BIT, true),
     DEFINE_PROP_END_OF_LIST(),

hw/scsi/Makefile.objs

@@ -8,6 +8,6 @@ common-obj-$(CONFIG_ESP_PCI) += esp-pci.o
 obj-$(CONFIG_PSERIES) += spapr_vscsi.o
 
 ifeq ($(CONFIG_VIRTIO),y)
-obj-y += virtio-scsi.o
+obj-y += virtio-scsi.o virtio-scsi-dataplane.o
 obj-$(CONFIG_VHOST_SCSI) += vhost-scsi.o
 endif

hw/scsi/scsi-bus.c

@@ -551,8 +551,11 @@ SCSIRequest *scsi_req_alloc(const SCSIReqOps *reqops, SCSIDevice *d,
     SCSIRequest *req;
     SCSIBus *bus = scsi_bus_from_device(d);
     BusState *qbus = BUS(bus);
+    const int memset_off = offsetof(SCSIRequest, sense)
+                           + sizeof(req->sense);
 
-    req = g_malloc0(reqops->size);
+    req = g_slice_alloc(reqops->size);
+    memset((uint8_t *)req + memset_off, 0, reqops->size - memset_off);
     req->refcount = 1;
     req->bus = bus;
     req->dev = d;
@@ -560,10 +563,10 @@ SCSIRequest *scsi_req_alloc(const SCSIReqOps *reqops, SCSIDevice *d,
     req->lun = lun;
     req->hba_private = hba_private;
     req->status = -1;
-    req->sense_len = 0;
     req->ops = reqops;
     object_ref(OBJECT(d));
     object_ref(OBJECT(qbus->parent));
+    notifier_list_init(&req->cancel_notifiers);
     trace_scsi_req_alloc(req->dev->id, req->lun, req->tag);
     return req;
 }
@@ -1603,7 +1606,7 @@ void scsi_req_unref(SCSIRequest *req)
         }
         object_unref(OBJECT(req->dev));
         object_unref(OBJECT(qbus->parent));
-        g_free(req);
+        g_slice_free1(req->ops->size, req);
     }
 }
 
@@ -1713,9 +1716,44 @@ void scsi_req_complete(SCSIRequest *req, int status)
     scsi_req_ref(req);
     scsi_req_dequeue(req);
     req->bus->info->complete(req, req->status, req->resid);
+
+    /* Cancelled requests might end up being completed instead of cancelled */
+    notifier_list_notify(&req->cancel_notifiers, req);
     scsi_req_unref(req);
 }
 
+/* Called by the devices when the request is canceled. */
+void scsi_req_cancel_complete(SCSIRequest *req)
+{
+    assert(req->io_canceled);
+    if (req->bus->info->cancel) {
+        req->bus->info->cancel(req);
+    }
+    notifier_list_notify(&req->cancel_notifiers, req);
+    scsi_req_unref(req);
+}
+
+/* Cancel @req asynchronously. @notifier is added to @req's cancellation
+ * notifier list, the bus will be notified the requests cancellation is
+ * completed.
+ * */
+void scsi_req_cancel_async(SCSIRequest *req, Notifier *notifier)
+{
+    trace_scsi_req_cancel(req->dev->id, req->lun, req->tag);
+    if (notifier) {
+        notifier_list_add(&req->cancel_notifiers, notifier);
+    }
+    if (req->io_canceled) {
+        return;
+    }
+    scsi_req_ref(req);
+    scsi_req_dequeue(req);
+    req->io_canceled = true;
+    if (req->aiocb) {
+        bdrv_aio_cancel_async(req->aiocb);
+    }
+}
+
 void scsi_req_cancel(SCSIRequest *req)
 {
     trace_scsi_req_cancel(req->dev->id, req->lun, req->tag);
@@ -1725,28 +1763,9 @@ void scsi_req_cancel(SCSIRequest *req)
     scsi_req_ref(req);
     scsi_req_dequeue(req);
     req->io_canceled = true;
-    if (req->ops->cancel_io) {
-        req->ops->cancel_io(req);
+    if (req->aiocb) {
+        bdrv_aio_cancel(req->aiocb);
     }
-    if (req->bus->info->cancel) {
-        req->bus->info->cancel(req);
-    }
-    scsi_req_unref(req);
-}
-
-void scsi_req_abort(SCSIRequest *req, int status)
-{
-    if (!req->enqueued) {
-        return;
-    }
-    scsi_req_ref(req);
-    scsi_req_dequeue(req);
-    req->io_canceled = true;
-    if (req->ops->cancel_io) {
-        req->ops->cancel_io(req);
-    }
-    scsi_req_complete(req, status);
-    scsi_req_unref(req);
 }
 
 static int scsi_ua_precedence(SCSISense sense)

hw/scsi/scsi-disk.c

@@ -105,23 +105,6 @@ static void scsi_check_condition(SCSIDiskReq *r, SCSISense sense)
     scsi_req_complete(&r->req, CHECK_CONDITION);
 }
 
-/* Cancel a pending data transfer.  */
-static void scsi_cancel_io(SCSIRequest *req)
-{
-    SCSIDiskReq *r = DO_UPCAST(SCSIDiskReq, req, req);
-
-    DPRINTF("Cancel tag=0x%x\n", req->tag);
-    if (r->req.aiocb) {
-        bdrv_aio_cancel(r->req.aiocb);
-
-        /* This reference was left in by scsi_*_data.  We take ownership of
-         * it the moment scsi_req_cancel is called, independent of whether
-         * bdrv_aio_cancel completes the request or not.  */
-        scsi_req_unref(&r->req);
-    }
-    r->req.aiocb = NULL;
-}
-
 static uint32_t scsi_init_iovec(SCSIDiskReq *r, size_t size)
 {
     SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev);
@@ -185,6 +168,7 @@ static void scsi_aio_complete(void *opaque, int ret)
     r->req.aiocb = NULL;
     block_acct_done(bdrv_get_stats(s->qdev.conf.bs), &r->acct);
     if (r->req.io_canceled) {
+        scsi_req_cancel_complete(&r->req);
         goto done;
     }
 
@@ -197,9 +181,7 @@ static void scsi_aio_complete(void *opaque, int ret)
     scsi_req_complete(&r->req, GOOD);
 
 done:
-    if (!r->req.io_canceled) {
-        scsi_req_unref(&r->req);
-    }
+    scsi_req_unref(&r->req);
 }
 
 static bool scsi_is_cmd_fua(SCSICommand *cmd)
@@ -233,6 +215,7 @@ static void scsi_write_do_fua(SCSIDiskReq *r)
     SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev);
 
     if (r->req.io_canceled) {
+        scsi_req_cancel_complete(&r->req);
         goto done;
     }
 
@@ -246,9 +229,7 @@ static void scsi_write_do_fua(SCSIDiskReq *r)
     scsi_req_complete(&r->req, GOOD);
 
 done:
-    if (!r->req.io_canceled) {
-        scsi_req_unref(&r->req);
-    }
+    scsi_req_unref(&r->req);
 }
 
 static void scsi_dma_complete_noio(void *opaque, int ret)
@@ -261,6 +242,7 @@ static void scsi_dma_complete_noio(void *opaque, int ret)
         block_acct_done(bdrv_get_stats(s->qdev.conf.bs), &r->acct);
     }
     if (r->req.io_canceled) {
+        scsi_req_cancel_complete(&r->req);
         goto done;
     }
 
@@ -280,9 +262,7 @@ static void scsi_dma_complete_noio(void *opaque, int ret)
     }
 
 done:
-    if (!r->req.io_canceled) {
-        scsi_req_unref(&r->req);
-    }
+    scsi_req_unref(&r->req);
 }
 
 static void scsi_dma_complete(void *opaque, int ret)
@@ -303,6 +283,7 @@ static void scsi_read_complete(void * opaque, int ret)
     r->req.aiocb = NULL;
     block_acct_done(bdrv_get_stats(s->qdev.conf.bs), &r->acct);
     if (r->req.io_canceled) {
+        scsi_req_cancel_complete(&r->req);
         goto done;
     }
 
@@ -320,9 +301,7 @@ static void scsi_read_complete(void * opaque, int ret)
     scsi_req_data(&r->req, r->qiov.size);
 
 done:
-    if (!r->req.io_canceled) {
-        scsi_req_unref(&r->req);
-    }
+    scsi_req_unref(&r->req);
 }
 
 /* Actually issue a read to the block device.  */
@@ -337,6 +316,7 @@ static void scsi_do_read(void *opaque, int ret)
         block_acct_done(bdrv_get_stats(s->qdev.conf.bs), &r->acct);
     }
     if (r->req.io_canceled) {
+        scsi_req_cancel_complete(&r->req);
         goto done;
     }
 
@@ -363,9 +343,7 @@ static void scsi_do_read(void *opaque, int ret)
     }
 
 done:
-    if (!r->req.io_canceled) {
-        scsi_req_unref(&r->req);
-    }
+    scsi_req_unref(&r->req);
 }
 
 /* Read more data from scsi device into buffer.  */
@@ -459,6 +437,7 @@ static void scsi_write_complete(void * opaque, int ret)
         block_acct_done(bdrv_get_stats(s->qdev.conf.bs), &r->acct);
     }
     if (r->req.io_canceled) {
+        scsi_req_cancel_complete(&r->req);
         goto done;
     }
 
@@ -481,9 +460,7 @@ static void scsi_write_complete(void * opaque, int ret)
     }
 
 done:
-    if (!r->req.io_canceled) {
-        scsi_req_unref(&r->req);
-    }
+    scsi_req_unref(&r->req);
 }
 
 static void scsi_write_data(SCSIRequest *req)
@@ -1553,6 +1530,7 @@ static void scsi_unmap_complete(void *opaque, int ret)
 
     r->req.aiocb = NULL;
     if (r->req.io_canceled) {
+        scsi_req_cancel_complete(&r->req);
         goto done;
     }
 
@@ -1582,9 +1560,7 @@ static void scsi_unmap_complete(void *opaque, int ret)
     scsi_req_complete(&r->req, GOOD);
 
 done:
-    if (!r->req.io_canceled) {
-        scsi_req_unref(&r->req);
-    }
+    scsi_req_unref(&r->req);
     g_free(data);
 }
 
@@ -1654,6 +1630,7 @@ static void scsi_write_same_complete(void *opaque, int ret)
     r->req.aiocb = NULL;
     block_acct_done(bdrv_get_stats(s->qdev.conf.bs), &r->acct);
     if (r->req.io_canceled) {
+        scsi_req_cancel_complete(&r->req);
         goto done;
     }
 
@@ -1678,9 +1655,7 @@ static void scsi_write_same_complete(void *opaque, int ret)
     scsi_req_complete(&r->req, GOOD);
 
 done:
-    if (!r->req.io_canceled) {
-        scsi_req_unref(&r->req);
-    }
+    scsi_req_unref(&r->req);
     qemu_vfree(data->iov.iov_base);
     g_free(data);
 }
@@ -2346,7 +2321,6 @@ static const SCSIReqOps scsi_disk_emulate_reqops = {
     .send_command = scsi_disk_emulate_command,
     .read_data    = scsi_disk_emulate_read_data,
     .write_data   = scsi_disk_emulate_write_data,
-    .cancel_io    = scsi_cancel_io,
     .get_buf      = scsi_get_buf,
 };
 
@@ -2356,7 +2330,6 @@ static const SCSIReqOps scsi_disk_dma_reqops = {
     .send_command = scsi_disk_dma_command,
     .read_data    = scsi_read_data,
     .write_data   = scsi_write_data,
-    .cancel_io    = scsi_cancel_io,
     .get_buf      = scsi_get_buf,
     .load_request = scsi_disk_load_request,
     .save_request = scsi_disk_save_request,

hw/scsi/scsi-generic.c

@@ -93,6 +93,10 @@ static void scsi_command_complete(void *opaque, int ret)
     SCSIGenericReq *r = (SCSIGenericReq *)opaque;
 
     r->req.aiocb = NULL;
+    if (r->req.io_canceled) {
+        scsi_req_cancel_complete(&r->req);
+        goto done;
+    }
     if (r->io_header.driver_status & SG_ERR_DRIVER_SENSE) {
         r->req.sense_len = r->io_header.sb_len_wr;
     }
@@ -133,26 +137,8 @@ static void scsi_command_complete(void *opaque, int ret)
             r, r->req.tag, status);
 
     scsi_req_complete(&r->req, status);
-    if (!r->req.io_canceled) {
-        scsi_req_unref(&r->req);
-    }
-}
-
-/* Cancel a pending data transfer.  */
-static void scsi_cancel_io(SCSIRequest *req)
-{
-    SCSIGenericReq *r = DO_UPCAST(SCSIGenericReq, req, req);
-
-    DPRINTF("Cancel tag=0x%x\n", req->tag);
-    if (r->req.aiocb) {
-        bdrv_aio_cancel(r->req.aiocb);
-
-        /* This reference was left in by scsi_*_data.  We take ownership of
-         * it independent of whether bdrv_aio_cancel completes the request
-         * or not.  */
-        scsi_req_unref(&r->req);
-    }
-    r->req.aiocb = NULL;
+done:
+    scsi_req_unref(&r->req);
 }
 
 static int execute_command(BlockDriverState *bdrv,
@@ -186,8 +172,7 @@ static void scsi_read_complete(void * opaque, int ret)
     int len;
 
     r->req.aiocb = NULL;
-    if (ret) {
-        DPRINTF("IO error ret %d\n", ret);
+    if (ret || r->req.io_canceled) {
         scsi_command_complete(r, ret);
         return;
     }
@@ -211,9 +196,7 @@ static void scsi_read_complete(void * opaque, int ret)
         bdrv_set_guest_block_size(s->conf.bs, s->blocksize);
 
         scsi_req_data(&r->req, len);
-        if (!r->req.io_canceled) {
-            scsi_req_unref(&r->req);
-        }
+        scsi_req_unref(&r->req);
     }
 }
 
@@ -246,8 +229,7 @@ static void scsi_write_complete(void * opaque, int ret)
 
     DPRINTF("scsi_write_complete() ret = %d\n", ret);
     r->req.aiocb = NULL;
-    if (ret) {
-        DPRINTF("IO error\n");
+    if (ret || r->req.io_canceled) {
         scsi_command_complete(r, ret);
         return;
     }
@@ -465,7 +447,6 @@ const SCSIReqOps scsi_generic_req_ops = {
     .send_command = scsi_send_command,
     .read_data    = scsi_read_data,
     .write_data   = scsi_write_data,
-    .cancel_io    = scsi_cancel_io,
     .get_buf      = scsi_get_buf,
     .load_request = scsi_generic_load_request,
     .save_request = scsi_generic_save_request,

hw/scsi/spapr_vscsi.c

@@ -77,8 +77,9 @@ typedef struct vscsi_req {
     SCSIRequest             *sreq;
     uint32_t                qtag; /* qemu tag != srp tag */
     bool                    active;
-    uint32_t                data_len;
     bool                    writing;
+    bool                    dma_error;
+    uint32_t                data_len;
     uint32_t                senselen;
     uint8_t                 sense[SCSI_SENSE_BUF_SIZE];
 
@@ -536,8 +537,8 @@ static void vscsi_transfer_data(SCSIRequest *sreq, uint32_t len)
     }
     if (rc < 0) {
         fprintf(stderr, "VSCSI: RDMA error rc=%d!\n", rc);
-        vscsi_makeup_sense(s, req, HARDWARE_ERROR, 0, 0);
-        scsi_req_abort(req->sreq, CHECK_CONDITION);
+        req->dma_error = true;
+        scsi_req_cancel(req->sreq);
         return;
     }
 
@@ -591,6 +592,12 @@ static void vscsi_request_cancelled(SCSIRequest *sreq)
 {
     vscsi_req *req = sreq->hba_private;
 
+    if (req->dma_error) {
+        VSCSIState *s = VIO_SPAPR_VSCSI_DEVICE(sreq->bus->qbus.parent);
+
+        vscsi_makeup_sense(s, req, HARDWARE_ERROR, 0, 0);
+        vscsi_send_rsp(s, req, CHECK_CONDITION, 0, 0);
+    }
     vscsi_put_req(req);
 }
 

hw/scsi/vhost-scsi.c

@@ -23,6 +23,7 @@
 #include "hw/virtio/vhost.h"
 #include "hw/virtio/virtio-scsi.h"
 #include "hw/virtio/virtio-bus.h"
+#include "hw/virtio/virtio-access.h"
 
 /* Features supported by host kernel. */
 static const int kernel_feature_bits[] = {
@@ -163,8 +164,8 @@ static void vhost_scsi_set_config(VirtIODevice *vdev,
     VirtIOSCSIConfig *scsiconf = (VirtIOSCSIConfig *)config;
     VirtIOSCSICommon *vs = VIRTIO_SCSI_COMMON(vdev);
 
-    if ((uint32_t) ldl_p(&scsiconf->sense_size) != vs->sense_size ||
-        (uint32_t) ldl_p(&scsiconf->cdb_size) != vs->cdb_size) {
+    if ((uint32_t) virtio_ldl_p(vdev, &scsiconf->sense_size) != vs->sense_size ||
+        (uint32_t) virtio_ldl_p(vdev, &scsiconf->cdb_size) != vs->cdb_size) {
         error_report("vhost-scsi does not support changing the sense data and CDB sizes");
         exit(1);
     }

hw/scsi/virtio-scsi-dataplane.c

@@ -0,0 +1,229 @@
+/*
+ * Virtio SCSI dataplane
+ *
+ * Copyright Red Hat, Inc. 2014
+ *
+ * Authors:
+ *   Fam Zheng <famz@redhat.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "hw/virtio/virtio-scsi.h"
+#include "qemu/error-report.h"
+#include <hw/scsi/scsi.h>
+#include <block/scsi.h>
+#include <hw/virtio/virtio-bus.h>
+#include "hw/virtio/virtio-access.h"
+#include "stdio.h"
+
+/* Context: QEMU global mutex held */
+void virtio_scsi_set_iothread(VirtIOSCSI *s, IOThread *iothread)
+{
+    BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(s)));
+    VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);
+    VirtIOSCSICommon *vs = VIRTIO_SCSI_COMMON(s);
+
+    assert(!s->ctx);
+    s->ctx = iothread_get_aio_context(vs->conf.iothread);
+
+    /* Don't try if transport does not support notifiers. */
+    if (!k->set_guest_notifiers || !k->set_host_notifier) {
+        fprintf(stderr, "virtio-scsi: Failed to set iothread "
+                   "(transport does not support notifiers)");
+        exit(1);
+    }
+}
+
+static VirtIOSCSIVring *virtio_scsi_vring_init(VirtIOSCSI *s,
+                                               VirtQueue *vq,
+                                               EventNotifierHandler *handler,
+                                               int n)
+{
+    BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(s)));
+    VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);
+    VirtIOSCSIVring *r = g_slice_new(VirtIOSCSIVring);
+
+    /* Set up virtqueue notify */
+    if (k->set_host_notifier(qbus->parent, n, true) != 0) {
+        fprintf(stderr, "virtio-scsi: Failed to set host notifier\n");
+        exit(1);
+    }
+    r->host_notifier = *virtio_queue_get_host_notifier(vq);
+    r->guest_notifier = *virtio_queue_get_guest_notifier(vq);
+    aio_set_event_notifier(s->ctx, &r->host_notifier, handler);
+
+    r->parent = s;
+
+    if (!vring_setup(&r->vring, VIRTIO_DEVICE(s), n)) {
+        fprintf(stderr, "virtio-scsi: VRing setup failed\n");
+        exit(1);
+    }
+    return r;
+}
+
+VirtIOSCSIReq *virtio_scsi_pop_req_vring(VirtIOSCSI *s,
+                                         VirtIOSCSIVring *vring)
+{
+    VirtIOSCSIReq *req = virtio_scsi_init_req(s, NULL);
+    int r;
+
+    req->vring = vring;
+    r = vring_pop((VirtIODevice *)s, &vring->vring, &req->elem);
+    if (r < 0) {
+        virtio_scsi_free_req(req);
+        req = NULL;
+    }
+    return req;
+}
+
+void virtio_scsi_vring_push_notify(VirtIOSCSIReq *req)
+{
+    vring_push(&req->vring->vring, &req->elem,
+               req->qsgl.size + req->resp_iov.size);
+    event_notifier_set(&req->vring->guest_notifier);
+}
+
+static void virtio_scsi_iothread_handle_ctrl(EventNotifier *notifier)
+{
+    VirtIOSCSIVring *vring = container_of(notifier,
+                                          VirtIOSCSIVring, host_notifier);
+    VirtIOSCSI *s = VIRTIO_SCSI(vring->parent);
+    VirtIOSCSIReq *req;
+
+    event_notifier_test_and_clear(notifier);
+    while ((req = virtio_scsi_pop_req_vring(s, vring))) {
+        virtio_scsi_handle_ctrl_req(s, req);
+    }
+}
+
+static void virtio_scsi_iothread_handle_event(EventNotifier *notifier)
+{
+    VirtIOSCSIVring *vring = container_of(notifier,
+                                          VirtIOSCSIVring, host_notifier);
+    VirtIOSCSI *s = vring->parent;
+    VirtIODevice *vdev = VIRTIO_DEVICE(s);
+
+    event_notifier_test_and_clear(notifier);
+
+    if (!(vdev->status & VIRTIO_CONFIG_S_DRIVER_OK)) {
+        return;
+    }
+
+    if (s->events_dropped) {
+        virtio_scsi_push_event(s, NULL, VIRTIO_SCSI_T_NO_EVENT, 0);
+    }
+}
+
+static void virtio_scsi_iothread_handle_cmd(EventNotifier *notifier)
+{
+    VirtIOSCSIVring *vring = container_of(notifier,
+                                          VirtIOSCSIVring, host_notifier);
+    VirtIOSCSI *s = (VirtIOSCSI *)vring->parent;
+    VirtIOSCSIReq *req, *next;
+    QTAILQ_HEAD(, VirtIOSCSIReq) reqs = QTAILQ_HEAD_INITIALIZER(reqs);
+
+    event_notifier_test_and_clear(notifier);
+    while ((req = virtio_scsi_pop_req_vring(s, vring))) {
+        if (virtio_scsi_handle_cmd_req_prepare(s, req)) {
+            QTAILQ_INSERT_TAIL(&reqs, req, next);
+        }
+    }
+
+    QTAILQ_FOREACH_SAFE(req, &reqs, next, next) {
+        virtio_scsi_handle_cmd_req_submit(s, req);
+    }
+}
+
+/* Context: QEMU global mutex held */
+void virtio_scsi_dataplane_start(VirtIOSCSI *s)
+{
+    int i;
+    int rc;
+    BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(s)));
+    VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);
+    VirtIOSCSICommon *vs = VIRTIO_SCSI_COMMON(s);
+
+    if (s->dataplane_started ||
+        s->dataplane_starting ||
+        s->ctx != iothread_get_aio_context(vs->conf.iothread)) {
+        return;
+    }
+
+    s->dataplane_starting = true;
+
+    /* Set up guest notifier (irq) */
+    rc = k->set_guest_notifiers(qbus->parent, vs->conf.num_queues + 2, true);
+    if (rc != 0) {
+        fprintf(stderr, "virtio-scsi: Failed to set guest notifiers, "
+                "ensure -enable-kvm is set\n");
+        exit(1);
+    }
+
+    aio_context_acquire(s->ctx);
+    s->ctrl_vring = virtio_scsi_vring_init(s, vs->ctrl_vq,
+                                           virtio_scsi_iothread_handle_ctrl,
+                                           0);
+    s->event_vring = virtio_scsi_vring_init(s, vs->event_vq,
+                                            virtio_scsi_iothread_handle_event,
+                                            1);
+    s->cmd_vrings = g_malloc0(sizeof(VirtIOSCSIVring) * vs->conf.num_queues);
+    for (i = 0; i < vs->conf.num_queues; i++) {
+        s->cmd_vrings[i] =
+            virtio_scsi_vring_init(s, vs->cmd_vqs[i],
+                                   virtio_scsi_iothread_handle_cmd,
+                                   i + 2);
+    }
+
+    aio_context_release(s->ctx);
+    s->dataplane_starting = false;
+    s->dataplane_started = true;
+}
+
+/* Context: QEMU global mutex held */
+void virtio_scsi_dataplane_stop(VirtIOSCSI *s)
+{
+    BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(s)));
+    VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);
+    VirtIODevice *vdev = VIRTIO_DEVICE(s);
+    VirtIOSCSICommon *vs = VIRTIO_SCSI_COMMON(s);
+    int i;
+
+    if (!s->dataplane_started || s->dataplane_stopping) {
+        return;
+    }
+    s->dataplane_stopping = true;
+    assert(s->ctx == iothread_get_aio_context(vs->conf.iothread));
+
+    aio_context_acquire(s->ctx);
+
+    aio_set_event_notifier(s->ctx, &s->ctrl_vring->host_notifier, NULL);
+    aio_set_event_notifier(s->ctx, &s->event_vring->host_notifier, NULL);
+    for (i = 0; i < vs->conf.num_queues; i++) {
+        aio_set_event_notifier(s->ctx, &s->cmd_vrings[i]->host_notifier, NULL);
+    }
+
+    bdrv_drain_all(); /* ensure there are no in-flight requests */
+
+    aio_context_release(s->ctx);
+
+    /* Sync vring state back to virtqueue so that non-dataplane request
+     * processing can continue when we disable the host notifier below.
+     */
+    vring_teardown(&s->ctrl_vring->vring, vdev, 0);
+    vring_teardown(&s->event_vring->vring, vdev, 1);
+    for (i = 0; i < vs->conf.num_queues; i++) {
+        vring_teardown(&s->cmd_vrings[i]->vring, vdev, 2 + i);
+    }
+
+    for (i = 0; i < vs->conf.num_queues + 2; i++) {
+        k->set_host_notifier(qbus->parent, i, false);
+    }
+
+    /* Clean up guest notifier (irq) */
+    k->set_guest_notifiers(qbus->parent, vs->conf.num_queues + 2, false);
+    s->dataplane_stopping = false;
+    s->dataplane_started = false;
+}

hw/scsi/virtio-scsi.c

@@ -20,34 +20,7 @@
 #include <block/scsi.h>
 #include <hw/virtio/virtio-bus.h>
 #include "hw/virtio/virtio-access.h"
-
-typedef struct VirtIOSCSIReq {
-    VirtIOSCSI *dev;
-    VirtQueue *vq;
-    VirtQueueElement elem;
-    QEMUSGList qsgl;
-    SCSIRequest *sreq;
-    size_t resp_size;
-    enum SCSIXferMode mode;
-    QEMUIOVector resp_iov;
-    union {
-        VirtIOSCSICmdResp     cmd;
-        VirtIOSCSICtrlTMFResp tmf;
-        VirtIOSCSICtrlANResp  an;
-        VirtIOSCSIEvent       event;
-    } resp;
-    union {
-        struct {
-            VirtIOSCSICmdReq  cmd;
-            uint8_t           cdb[];
-        } QEMU_PACKED;
-        VirtIOSCSICtrlTMFReq  tmf;
-        VirtIOSCSICtrlANReq   an;
-    } req;
-} VirtIOSCSIReq;
-
-QEMU_BUILD_BUG_ON(offsetof(VirtIOSCSIReq, req.cdb) !=
-                  offsetof(VirtIOSCSIReq, req.cmd) + sizeof(VirtIOSCSICmdReq));
+#include "migration/migration.h"
 
 static inline int virtio_scsi_get_lun(uint8_t *lun)
 {
@@ -65,26 +38,29 @@ static inline SCSIDevice *virtio_scsi_device_find(VirtIOSCSI *s, uint8_t *lun)
     return scsi_device_find(&s->bus, 0, lun[1], virtio_scsi_get_lun(lun));
 }
 
-static VirtIOSCSIReq *virtio_scsi_init_req(VirtIOSCSI *s, VirtQueue *vq)
+VirtIOSCSIReq *virtio_scsi_init_req(VirtIOSCSI *s, VirtQueue *vq)
 {
     VirtIOSCSIReq *req;
-    VirtIOSCSICommon *vs = VIRTIO_SCSI_COMMON(s);
-
-    req = g_malloc0(sizeof(*req) + vs->cdb_size);
+    VirtIOSCSICommon *vs = (VirtIOSCSICommon *)s;
+    const size_t zero_skip = offsetof(VirtIOSCSIReq, elem)
+                             + sizeof(VirtQueueElement);
 
+    req = g_slice_alloc(sizeof(*req) + vs->cdb_size);
     req->vq = vq;
     req->dev = s;
-    req->sreq = NULL;
     qemu_sglist_init(&req->qsgl, DEVICE(s), 8, &address_space_memory);
     qemu_iovec_init(&req->resp_iov, 1);
+    memset((uint8_t *)req + zero_skip, 0, sizeof(*req) - zero_skip);
     return req;
 }
 
-static void virtio_scsi_free_req(VirtIOSCSIReq *req)
+void virtio_scsi_free_req(VirtIOSCSIReq *req)
 {
+    VirtIOSCSICommon *vs = (VirtIOSCSICommon *)req->dev;
+
     qemu_iovec_destroy(&req->resp_iov);
     qemu_sglist_destroy(&req->qsgl);
-    g_free(req);
+    g_slice_free1(sizeof(*req) + vs->cdb_size, req);
 }
 
 static void virtio_scsi_complete_req(VirtIOSCSIReq *req)
@@ -94,13 +70,19 @@ static void virtio_scsi_complete_req(VirtIOSCSIReq *req)
     VirtIODevice *vdev = VIRTIO_DEVICE(s);
 
     qemu_iovec_from_buf(&req->resp_iov, 0, &req->resp, req->resp_size);
-    virtqueue_push(vq, &req->elem, req->qsgl.size + req->resp_iov.size);
+    if (req->vring) {
+        assert(req->vq == NULL);
+        virtio_scsi_vring_push_notify(req);
+    } else {
+        virtqueue_push(vq, &req->elem, req->qsgl.size + req->resp_iov.size);
+        virtio_notify(vdev, vq);
+    }
+
     if (req->sreq) {
         req->sreq->hba_private = NULL;
         scsi_req_unref(req->sreq);
     }
     virtio_scsi_free_req(req);
-    virtio_notify(vdev, vq);
 }
 
 static void virtio_scsi_bad_req(void)
@@ -226,13 +208,39 @@ static void *virtio_scsi_load_request(QEMUFile *f, SCSIRequest *sreq)
     return req;
 }
 
-static void virtio_scsi_do_tmf(VirtIOSCSI *s, VirtIOSCSIReq *req)
+typedef struct {
+    Notifier        notifier;
+    VirtIOSCSIReq  *tmf_req;
+} VirtIOSCSICancelNotifier;
+
+static void virtio_scsi_cancel_notify(Notifier *notifier, void *data)
+{
+    VirtIOSCSICancelNotifier *n = container_of(notifier,
+                                               VirtIOSCSICancelNotifier,
+                                               notifier);
+
+    if (--n->tmf_req->remaining == 0) {
+        virtio_scsi_complete_req(n->tmf_req);
+    }
+    g_slice_free(VirtIOSCSICancelNotifier, n);
+}
+
+/* Return 0 if the request is ready to be completed and return to guest;
+ * -EINPROGRESS if the request is submitted and will be completed later, in the
+ *  case of async cancellation. */
+static int virtio_scsi_do_tmf(VirtIOSCSI *s, VirtIOSCSIReq *req)
 {
     SCSIDevice *d = virtio_scsi_device_find(s, req->req.tmf.lun);
     SCSIRequest *r, *next;
     BusChild *kid;
     int target;
+    int ret = 0;
 
+    if (s->dataplane_started && bdrv_get_aio_context(d->conf.bs) != s->ctx) {
+        aio_context_acquire(s->ctx);
+        bdrv_set_aio_context(d->conf.bs, s->ctx);
+        aio_context_release(s->ctx);
+    }
     /* Here VIRTIO_SCSI_S_OK means "FUNCTION COMPLETE".  */
     req->resp.tmf.response = VIRTIO_SCSI_S_OK;
 
@@ -264,7 +272,14 @@ static void virtio_scsi_do_tmf(VirtIOSCSI *s, VirtIOSCSIReq *req)
                  */
                 req->resp.tmf.response = VIRTIO_SCSI_S_FUNCTION_SUCCEEDED;
             } else {
-                scsi_req_cancel(r);
+                VirtIOSCSICancelNotifier *notifier;
+
+                req->remaining = 1;
+                notifier = g_slice_new(VirtIOSCSICancelNotifier);
+                notifier->tmf_req = req;
+                notifier->notifier.notify = virtio_scsi_cancel_notify;
+                scsi_req_cancel_async(r, &notifier->notifier);
+                ret = -EINPROGRESS;
             }
         }
         break;
@@ -290,6 +305,13 @@ static void virtio_scsi_do_tmf(VirtIOSCSI *s, VirtIOSCSIReq *req)
         if (d->lun != virtio_scsi_get_lun(req->req.tmf.lun)) {
             goto incorrect_lun;
         }
+
+        /* Add 1 to "remaining" until virtio_scsi_do_tmf returns.
+         * This way, if the bus starts calling back to the notifiers
+         * even before we finish the loop, virtio_scsi_cancel_notify
+         * will not complete the TMF too early.
+         */
+        req->remaining = 1;
         QTAILQ_FOREACH_SAFE(r, &d->requests, next, next) {
             if (r->hba_private) {
                 if (req->req.tmf.subtype == VIRTIO_SCSI_T_TMF_QUERY_TASK_SET) {
@@ -299,10 +321,19 @@ static void virtio_scsi_do_tmf(VirtIOSCSI *s, VirtIOSCSIReq *req)
                     req->resp.tmf.response = VIRTIO_SCSI_S_FUNCTION_SUCCEEDED;
                     break;
                 } else {
-                    scsi_req_cancel(r);
+                    VirtIOSCSICancelNotifier *notifier;
+
+                    req->remaining++;
+                    notifier = g_slice_new(VirtIOSCSICancelNotifier);
+                    notifier->notifier.notify = virtio_scsi_cancel_notify;
+                    notifier->tmf_req = req;
+                    scsi_req_cancel_async(r, &notifier->notifier);
                 }
             }
         }
+        if (--req->remaining > 0) {
+            ret = -EINPROGRESS;
+        }
         break;
 
     case VIRTIO_SCSI_T_TMF_I_T_NEXUS_RESET:
@@ -323,14 +354,53 @@ static void virtio_scsi_do_tmf(VirtIOSCSI *s, VirtIOSCSIReq *req)
         break;
     }
 
-    return;
+    return ret;
 
 incorrect_lun:
     req->resp.tmf.response = VIRTIO_SCSI_S_INCORRECT_LUN;
-    return;
+    return ret;
 
 fail:
     req->resp.tmf.response = VIRTIO_SCSI_S_BAD_TARGET;
+    return ret;
+}
+
+void virtio_scsi_handle_ctrl_req(VirtIOSCSI *s, VirtIOSCSIReq *req)
+{
+    VirtIODevice *vdev = (VirtIODevice *)s;
+    int type;
+    int r = 0;
+
+    if (iov_to_buf(req->elem.out_sg, req->elem.out_num, 0,
+                &type, sizeof(type)) < sizeof(type)) {
+        virtio_scsi_bad_req();
+        return;
+    }
+
+    virtio_tswap32s(vdev, &req->req.tmf.type);
+    if (req->req.tmf.type == VIRTIO_SCSI_T_TMF) {
+        if (virtio_scsi_parse_req(req, sizeof(VirtIOSCSICtrlTMFReq),
+                    sizeof(VirtIOSCSICtrlTMFResp)) < 0) {
+            virtio_scsi_bad_req();
+        } else {
+            r = virtio_scsi_do_tmf(s, req);
+        }
+
+    } else if (req->req.tmf.type == VIRTIO_SCSI_T_AN_QUERY ||
+            req->req.tmf.type == VIRTIO_SCSI_T_AN_SUBSCRIBE) {
+        if (virtio_scsi_parse_req(req, sizeof(VirtIOSCSICtrlANReq),
+                    sizeof(VirtIOSCSICtrlANResp)) < 0) {
+            virtio_scsi_bad_req();
+        } else {
+            req->resp.an.event_actual = 0;
+            req->resp.an.response = VIRTIO_SCSI_S_OK;
+        }
+    }
+    if (r == 0) {
+        virtio_scsi_complete_req(req);
+    } else {
+        assert(r == -EINPROGRESS);
+    }
 }
 
 static void virtio_scsi_handle_ctrl(VirtIODevice *vdev, VirtQueue *vq)
@@ -338,35 +408,12 @@ static void virtio_scsi_handle_ctrl(VirtIODevice *vdev, VirtQueue *vq)
     VirtIOSCSI *s = (VirtIOSCSI *)vdev;
     VirtIOSCSIReq *req;
 
+    if (s->ctx && !s->dataplane_disabled) {
+        virtio_scsi_dataplane_start(s);
+        return;
+    }
     while ((req = virtio_scsi_pop_req(s, vq))) {
-        int type;
-
-        if (iov_to_buf(req->elem.out_sg, req->elem.out_num, 0,
-                       &type, sizeof(type)) < sizeof(type)) {
-            virtio_scsi_bad_req();
-            continue;
-        }
-
-        virtio_tswap32s(vdev, &req->req.tmf.type);
-        if (req->req.tmf.type == VIRTIO_SCSI_T_TMF) {
-            if (virtio_scsi_parse_req(req, sizeof(VirtIOSCSICtrlTMFReq),
-                                      sizeof(VirtIOSCSICtrlTMFResp)) < 0) {
-                virtio_scsi_bad_req();
-            } else {
-                virtio_scsi_do_tmf(s, req);
-            }
-
-        } else if (req->req.tmf.type == VIRTIO_SCSI_T_AN_QUERY ||
-                   req->req.tmf.type == VIRTIO_SCSI_T_AN_SUBSCRIBE) {
-            if (virtio_scsi_parse_req(req, sizeof(VirtIOSCSICtrlANReq),
-                                      sizeof(VirtIOSCSICtrlANResp)) < 0) {
-                virtio_scsi_bad_req();
-            } else {
-                req->resp.an.event_actual = 0;
-                req->resp.an.response = VIRTIO_SCSI_S_OK;
-            }
-        }
-        virtio_scsi_complete_req(req);
+        virtio_scsi_handle_ctrl_req(s, req);
     }
 }
 
@@ -420,13 +467,7 @@ static int virtio_scsi_parse_cdb(SCSIDevice *dev, SCSICommand *cmd,
      * host device passthrough.
      */
     cmd->xfer = req->qsgl.size;
-    if (cmd->xfer == 0) {
-        cmd->mode = SCSI_XFER_NONE;
-    } else if (iov_size(req->elem.in_sg, req->elem.in_num) > req->resp_size) {
-        cmd->mode = SCSI_XFER_FROM_DEV;
-    } else {
-        cmd->mode = SCSI_XFER_TO_DEV;
-    }
+    cmd->mode = req->mode;
     return 0;
 }
 
@@ -458,52 +499,78 @@ static void virtio_scsi_fail_cmd_req(VirtIOSCSIReq *req)
     virtio_scsi_complete_cmd_req(req);
 }
 
+bool virtio_scsi_handle_cmd_req_prepare(VirtIOSCSI *s, VirtIOSCSIReq *req)
+{
+    VirtIOSCSICommon *vs = &s->parent_obj;
+    SCSIDevice *d;
+    int rc;
+
+    rc = virtio_scsi_parse_req(req, sizeof(VirtIOSCSICmdReq) + vs->cdb_size,
+                               sizeof(VirtIOSCSICmdResp) + vs->sense_size);
+    if (rc < 0) {
+        if (rc == -ENOTSUP) {
+            virtio_scsi_fail_cmd_req(req);
+        } else {
+            virtio_scsi_bad_req();
+        }
+        return false;
+    }
+
+    d = virtio_scsi_device_find(s, req->req.cmd.lun);
+    if (!d) {
+        req->resp.cmd.response = VIRTIO_SCSI_S_BAD_TARGET;
+        virtio_scsi_complete_cmd_req(req);
+        return false;
+    }
+    if (s->dataplane_started && bdrv_get_aio_context(d->conf.bs) != s->ctx) {
+        aio_context_acquire(s->ctx);
+        bdrv_set_aio_context(d->conf.bs, s->ctx);
+        aio_context_release(s->ctx);
+    }
+    req->sreq = scsi_req_new(d, req->req.cmd.tag,
+                             virtio_scsi_get_lun(req->req.cmd.lun),
+                             req->req.cdb, req);
+
+    if (req->sreq->cmd.mode != SCSI_XFER_NONE
+        && (req->sreq->cmd.mode != req->mode ||
+            req->sreq->cmd.xfer > req->qsgl.size)) {
+        req->resp.cmd.response = VIRTIO_SCSI_S_OVERRUN;
+        virtio_scsi_complete_cmd_req(req);
+        return false;
+    }
+    scsi_req_ref(req->sreq);
+    bdrv_io_plug(d->conf.bs);
+    return true;
+}
+
+void virtio_scsi_handle_cmd_req_submit(VirtIOSCSI *s, VirtIOSCSIReq *req)
+{
+    if (scsi_req_enqueue(req->sreq)) {
+        scsi_req_continue(req->sreq);
+    }
+    bdrv_io_unplug(req->sreq->dev->conf.bs);
+    scsi_req_unref(req->sreq);
+}
+
 static void virtio_scsi_handle_cmd(VirtIODevice *vdev, VirtQueue *vq)
 {
     /* use non-QOM casts in the data path */
     VirtIOSCSI *s = (VirtIOSCSI *)vdev;
-    VirtIOSCSICommon *vs = &s->parent_obj;
-
-    VirtIOSCSIReq *req;
-    int n;
+    VirtIOSCSIReq *req, *next;
+    QTAILQ_HEAD(, VirtIOSCSIReq) reqs = QTAILQ_HEAD_INITIALIZER(reqs);
 
+    if (s->ctx && !s->dataplane_disabled) {
+        virtio_scsi_dataplane_start(s);
+        return;
+    }
     while ((req = virtio_scsi_pop_req(s, vq))) {
-        SCSIDevice *d;
-        int rc;
-
-        rc = virtio_scsi_parse_req(req, sizeof(VirtIOSCSICmdReq) + vs->cdb_size,
-                                   sizeof(VirtIOSCSICmdResp) + vs->sense_size);
-        if (rc < 0) {
-            if (rc == -ENOTSUP) {
-                virtio_scsi_fail_cmd_req(req);
-            } else {
-                virtio_scsi_bad_req();
-            }
-            continue;
+        if (virtio_scsi_handle_cmd_req_prepare(s, req)) {
+            QTAILQ_INSERT_TAIL(&reqs, req, next);
         }
+    }
 
-        d = virtio_scsi_device_find(s, req->req.cmd.lun);
-        if (!d) {
-            req->resp.cmd.response = VIRTIO_SCSI_S_BAD_TARGET;
-            virtio_scsi_complete_cmd_req(req);
-            continue;
-        }
-        req->sreq = scsi_req_new(d, req->req.cmd.tag,
-                                 virtio_scsi_get_lun(req->req.cmd.lun),
-                                 req->req.cdb, req);
-
-        if (req->sreq->cmd.mode != SCSI_XFER_NONE
-            && (req->sreq->cmd.mode != req->mode ||
-                req->sreq->cmd.xfer > req->qsgl.size)) {
-            req->resp.cmd.response = VIRTIO_SCSI_S_OVERRUN;
-            virtio_scsi_complete_cmd_req(req);
-            continue;
-        }
-
-        n = scsi_req_enqueue(req->sreq);
-        if (n) {
-            scsi_req_continue(req->sreq);
-        }
+    QTAILQ_FOREACH_SAFE(req, &reqs, next, next) {
+        virtio_scsi_handle_cmd_req_submit(s, req);
     }
 }
 
@@ -552,6 +619,9 @@ static void virtio_scsi_reset(VirtIODevice *vdev)
     VirtIOSCSI *s = VIRTIO_SCSI(vdev);
     VirtIOSCSICommon *vs = VIRTIO_SCSI_COMMON(vdev);
 
+    if (s->ctx) {
+        virtio_scsi_dataplane_stop(s);
+    }
     s->resetting++;
     qbus_reset_all(&s->bus.qbus);
     s->resetting--;
@@ -582,8 +652,8 @@ static int virtio_scsi_load(QEMUFile *f, void *opaque, int version_id)
     return 0;
 }
 
-static void virtio_scsi_push_event(VirtIOSCSI *s, SCSIDevice *dev,
-                                   uint32_t event, uint32_t reason)
+void virtio_scsi_push_event(VirtIOSCSI *s, SCSIDevice *dev,
+                            uint32_t event, uint32_t reason)
 {
     VirtIOSCSICommon *vs = VIRTIO_SCSI_COMMON(s);
     VirtIOSCSIReq *req;
@@ -594,10 +664,19 @@ static void virtio_scsi_push_event(VirtIOSCSI *s, SCSIDevice *dev,
         return;
     }
 
-    req = virtio_scsi_pop_req(s, vs->event_vq);
+    if (s->dataplane_started) {
+        assert(s->ctx);
+        aio_context_acquire(s->ctx);
+    }
+
+    if (s->dataplane_started) {
+        req = virtio_scsi_pop_req_vring(s, s->event_vring);
+    } else {
+        req = virtio_scsi_pop_req(s, vs->event_vq);
+    }
     if (!req) {
         s->events_dropped = true;
-        return;
+        goto out;
     }
 
     if (s->events_dropped) {
@@ -626,12 +705,20 @@ static void virtio_scsi_push_event(VirtIOSCSI *s, SCSIDevice *dev,
         evt->lun[3] = dev->lun & 0xFF;
     }
     virtio_scsi_complete_req(req);
+out:
+    if (s->dataplane_started) {
+        aio_context_release(s->ctx);
+    }
 }
 
 static void virtio_scsi_handle_event(VirtIODevice *vdev, VirtQueue *vq)
 {
     VirtIOSCSI *s = VIRTIO_SCSI(vdev);
 
+    if (s->ctx && !s->dataplane_disabled) {
+        virtio_scsi_dataplane_start(s);
+        return;
+    }
     if (s->events_dropped) {
         virtio_scsi_push_event(s, NULL, VIRTIO_SCSI_T_NO_EVENT, 0);
     }
@@ -717,6 +804,35 @@ void virtio_scsi_common_realize(DeviceState *dev, Error **errp,
         s->cmd_vqs[i] = virtio_add_queue(vdev, VIRTIO_SCSI_VQ_SIZE,
                                          cmd);
     }
+
+    if (s->conf.iothread) {
+        virtio_scsi_set_iothread(VIRTIO_SCSI(s), s->conf.iothread);
+    }
+}
+
+/* Disable dataplane thread during live migration since it does not
+ * update the dirty memory bitmap yet.
+ */
+static void virtio_scsi_migration_state_changed(Notifier *notifier, void *data)
+{
+    VirtIOSCSI *s = container_of(notifier, VirtIOSCSI,
+                                 migration_state_notifier);
+    MigrationState *mig = data;
+
+    if (migration_in_setup(mig)) {
+        if (!s->dataplane_started) {
+            return;
+        }
+        virtio_scsi_dataplane_stop(s);
+        s->dataplane_disabled = true;
+    } else if (migration_has_finished(mig) ||
+               migration_has_failed(mig)) {
+        if (s->dataplane_started) {
+            return;
+        }
+        bdrv_drain_all(); /* complete in-flight non-dataplane requests */
+        s->dataplane_disabled = false;
+    }
 }
 
 static void virtio_scsi_device_realize(DeviceState *dev, Error **errp)
@@ -747,6 +863,18 @@ static void virtio_scsi_device_realize(DeviceState *dev, Error **errp)
 
     register_savevm(dev, "virtio-scsi", virtio_scsi_id++, 1,
                     virtio_scsi_save, virtio_scsi_load, s);
+    s->migration_state_notifier.notify = virtio_scsi_migration_state_changed;
+    add_migration_state_change_notifier(&s->migration_state_notifier);
+}
+
+static void virtio_scsi_instance_init(Object *obj)
+{
+    VirtIOSCSICommon *vs = VIRTIO_SCSI_COMMON(obj);
+
+    object_property_add_link(obj, "iothread", TYPE_IOTHREAD,
+                             (Object **)&vs->conf.iothread,
+                             qdev_prop_allow_set_link_before_realize,
+                             OBJ_PROP_LINK_UNREF_ON_RELEASE, &error_abort);
 }
 
 void virtio_scsi_common_unrealize(DeviceState *dev, Error **errp)
@@ -763,6 +891,7 @@ static void virtio_scsi_device_unrealize(DeviceState *dev, Error **errp)
     VirtIOSCSI *s = VIRTIO_SCSI(dev);
 
     unregister_savevm(dev, "virtio-scsi", s);
+    remove_migration_state_change_notifier(&s->migration_state_notifier);
 
     virtio_scsi_common_unrealize(dev, errp);
 }
@@ -807,6 +936,7 @@ static const TypeInfo virtio_scsi_info = {
     .name = TYPE_VIRTIO_SCSI,
     .parent = TYPE_VIRTIO_SCSI_COMMON,
     .instance_size = sizeof(VirtIOSCSI),
+    .instance_init = virtio_scsi_instance_init,
     .class_init = virtio_scsi_class_init,
 };
 

hw/virtio/virtio-pci.c

@@ -926,7 +926,6 @@ static Property virtio_9p_pci_properties[] = {
     DEFINE_PROP_BIT("ioeventfd", VirtIOPCIProxy, flags,
                     VIRTIO_PCI_FLAG_USE_IOEVENTFD_BIT, true),
     DEFINE_PROP_UINT32("vectors", VirtIOPCIProxy, nvectors, 2),
-    DEFINE_VIRTIO_9P_PROPERTIES(V9fsPCIState, vdev.fsconf),
     DEFINE_PROP_END_OF_LIST(),
 };
 
@@ -948,8 +947,9 @@ static void virtio_9p_pci_class_init(ObjectClass *klass, void *data)
 static void virtio_9p_pci_instance_init(Object *obj)
 {
     V9fsPCIState *dev = VIRTIO_9P_PCI(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_9P);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_9P);
 }
 
 static const TypeInfo virtio_9p_pci_info = {
@@ -1111,10 +1111,9 @@ static void virtio_blk_pci_class_init(ObjectClass *klass, void *data)
 static void virtio_blk_pci_instance_init(Object *obj)
 {
     VirtIOBlkPCI *dev = VIRTIO_BLK_PCI(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BLK);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-    object_unref(OBJECT(&dev->vdev));
-    qdev_alias_all_properties(DEVICE(&dev->vdev), obj);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_BLK);
     object_property_add_alias(obj, "iothread", OBJECT(&dev->vdev),"iothread",
                               &error_abort);
 }
@@ -1135,7 +1134,6 @@ static Property virtio_scsi_pci_properties[] = {
     DEFINE_PROP_UINT32("vectors", VirtIOPCIProxy, nvectors,
                        DEV_NVECTORS_UNSPECIFIED),
     DEFINE_VIRTIO_SCSI_FEATURES(VirtIOPCIProxy, host_features),
-    DEFINE_VIRTIO_SCSI_PROPERTIES(VirtIOSCSIPCI, vdev.parent_obj.conf),
     DEFINE_PROP_END_OF_LIST(),
 };
 
@@ -1185,8 +1183,11 @@ static void virtio_scsi_pci_class_init(ObjectClass *klass, void *data)
 static void virtio_scsi_pci_instance_init(Object *obj)
 {
     VirtIOSCSIPCI *dev = VIRTIO_SCSI_PCI(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_SCSI);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_SCSI);
+    object_property_add_alias(obj, "iothread", OBJECT(&dev->vdev), "iothread",
+                              &error_abort);
 }
 
 static const TypeInfo virtio_scsi_pci_info = {
@@ -1203,7 +1204,6 @@ static const TypeInfo virtio_scsi_pci_info = {
 static Property vhost_scsi_pci_properties[] = {
     DEFINE_PROP_UINT32("vectors", VirtIOPCIProxy, nvectors,
                        DEV_NVECTORS_UNSPECIFIED),
-    DEFINE_VHOST_SCSI_PROPERTIES(VHostSCSIPCI, vdev.parent_obj.conf),
     DEFINE_PROP_END_OF_LIST(),
 };
 
@@ -1241,8 +1241,9 @@ static void vhost_scsi_pci_class_init(ObjectClass *klass, void *data)
 static void vhost_scsi_pci_instance_init(Object *obj)
 {
     VHostSCSIPCI *dev = VHOST_SCSI_PCI(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VHOST_SCSI);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VHOST_SCSI);
 }
 
 static const TypeInfo vhost_scsi_pci_info = {
@@ -1323,7 +1324,7 @@ static void virtio_balloon_pci_instance_init(Object *obj)
     VirtIOBalloonPCI *dev = VIRTIO_BALLOON_PCI(obj);
     object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BALLOON);
     object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-
+    object_unref(OBJECT(&dev->vdev));
     object_property_add(obj, "guest-stats", "guest statistics",
                         balloon_pci_stats_get_all, NULL, NULL, dev,
                         NULL);
@@ -1385,7 +1386,6 @@ static Property virtio_serial_pci_properties[] = {
                     VIRTIO_PCI_FLAG_USE_IOEVENTFD_BIT, true),
     DEFINE_PROP_UINT32("vectors", VirtIOPCIProxy, nvectors, 2),
     DEFINE_PROP_UINT32("class", VirtIOPCIProxy, class_code, 0),
-    DEFINE_VIRTIO_SERIAL_PROPERTIES(VirtIOSerialPCI, vdev.serial),
     DEFINE_PROP_END_OF_LIST(),
 };
 
@@ -1406,8 +1406,9 @@ static void virtio_serial_pci_class_init(ObjectClass *klass, void *data)
 static void virtio_serial_pci_instance_init(Object *obj)
 {
     VirtIOSerialPCI *dev = VIRTIO_SERIAL_PCI(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_SERIAL);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_SERIAL);
 }
 
 static const TypeInfo virtio_serial_pci_info = {
@@ -1425,8 +1426,6 @@ static Property virtio_net_properties[] = {
                     VIRTIO_PCI_FLAG_USE_IOEVENTFD_BIT, false),
     DEFINE_PROP_UINT32("vectors", VirtIOPCIProxy, nvectors, 3),
     DEFINE_VIRTIO_NET_FEATURES(VirtIOPCIProxy, host_features),
-    DEFINE_NIC_PROPERTIES(VirtIONetPCI, vdev.nic_conf),
-    DEFINE_VIRTIO_NET_PROPERTIES(VirtIONetPCI, vdev.net_conf),
     DEFINE_PROP_END_OF_LIST(),
 };
 
@@ -1465,8 +1464,9 @@ static void virtio_net_pci_class_init(ObjectClass *klass, void *data)
 static void virtio_net_pci_instance_init(Object *obj)
 {
     VirtIONetPCI *dev = VIRTIO_NET_PCI(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_NET);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_NET);
 }
 
 static const TypeInfo virtio_net_pci_info = {
@@ -1480,7 +1480,6 @@ static const TypeInfo virtio_net_pci_info = {
 /* virtio-rng-pci */
 
 static Property virtio_rng_pci_properties[] = {
-    DEFINE_VIRTIO_RNG_PROPERTIES(VirtIORngPCI, vdev.conf),
     DEFINE_PROP_END_OF_LIST(),
 };
 
@@ -1520,8 +1519,9 @@ static void virtio_rng_pci_class_init(ObjectClass *klass, void *data)
 static void virtio_rng_initfn(Object *obj)
 {
     VirtIORngPCI *dev = VIRTIO_RNG_PCI(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_RNG);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
+
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_RNG);
     object_property_add_link(obj, "rng", TYPE_RNG_BACKEND,
                              (Object **)&dev->vdev.conf.rng,
                              qdev_prop_allow_set_link_before_realize,

hw/virtio/virtio.c

@@ -1123,6 +1123,17 @@ static void virtio_vmstate_change(void *opaque, int running, RunState state)
     }
 }
 
+void virtio_instance_init_common(Object *proxy_obj, void *data,
+                                 size_t vdev_size, const char *vdev_name)
+{
+    DeviceState *vdev = data;
+
+    object_initialize(vdev, vdev_size, vdev_name);
+    object_property_add_child(proxy_obj, "virtio-backend", OBJECT(vdev), NULL);
+    object_unref(OBJECT(vdev));
+    qdev_alias_all_properties(vdev, proxy_obj);
+}
+
 void virtio_init(VirtIODevice *vdev, const char *name,
                  uint16_t device_id, size_t config_size)
 {

include/hw/scsi/scsi.h

@@ -5,6 +5,7 @@
 #include "block/block.h"
 #include "hw/block/block.h"
 #include "sysemu/sysemu.h"
+#include "qemu/notify.h"
 
 #define MAX_SCSI_DEVS	255
 
@@ -50,17 +51,25 @@ struct SCSIRequest {
     uint32_t          tag;
     uint32_t          lun;
     uint32_t          status;
+    void              *hba_private;
     size_t            resid;
     SCSICommand       cmd;
+    NotifierList      cancel_notifiers;
+
+    /* Note:
+     * - fields before sense are initialized by scsi_req_alloc;
+     * - sense[] is uninitialized;
+     * - fields after sense are memset to 0 by scsi_req_alloc.
+     * */
+
+    uint8_t           sense[SCSI_SENSE_BUF_SIZE];
+    uint32_t          sense_len;
+    bool              enqueued;
+    bool              io_canceled;
+    bool              retry;
+    bool              dma_started;
     BlockDriverAIOCB  *aiocb;
     QEMUSGList        *sg;
-    bool              dma_started;
-    uint8_t sense[SCSI_SENSE_BUF_SIZE];
-    uint32_t sense_len;
-    bool enqueued;
-    bool io_canceled;
-    bool retry;
-    void *hba_private;
     QTAILQ_ENTRY(SCSIRequest) next;
 };
 
@@ -123,7 +132,6 @@ struct SCSIReqOps {
     int32_t (*send_command)(SCSIRequest *req, uint8_t *buf);
     void (*read_data)(SCSIRequest *req);
     void (*write_data)(SCSIRequest *req);
-    void (*cancel_io)(SCSIRequest *req);
     uint8_t *(*get_buf)(SCSIRequest *req);
 
     void (*save_request)(QEMUFile *f, SCSIRequest *req);
@@ -258,8 +266,9 @@ void scsi_req_data(SCSIRequest *req, int len);
 void scsi_req_complete(SCSIRequest *req, int status);
 uint8_t *scsi_req_get_buf(SCSIRequest *req);
 int scsi_req_get_sense(SCSIRequest *req, uint8_t *buf, int len);
-void scsi_req_abort(SCSIRequest *req, int status);
+void scsi_req_cancel_complete(SCSIRequest *req);
 void scsi_req_cancel(SCSIRequest *req);
+void scsi_req_cancel_async(SCSIRequest *req, Notifier *notifier);
 void scsi_req_retry(SCSIRequest *req);
 void scsi_device_purge_requests(SCSIDevice *sdev, SCSISense sense);
 void scsi_device_set_ua(SCSIDevice *sdev, SCSISense sense);

include/hw/virtio/virtio-scsi.h

@@ -17,6 +17,8 @@
 #include "hw/virtio/virtio.h"
 #include "hw/pci/pci.h"
 #include "hw/scsi/scsi.h"
+#include "sysemu/iothread.h"
+#include "hw/virtio/dataplane/vring.h"
 
 #define TYPE_VIRTIO_SCSI_COMMON "virtio-scsi-common"
 #define VIRTIO_SCSI_COMMON(obj) \
@@ -151,8 +153,18 @@ struct VirtIOSCSIConf {
     uint32_t cmd_per_lun;
     char *vhostfd;
     char *wwpn;
+    IOThread *iothread;
 };
 
+struct VirtIOSCSI;
+
+typedef struct {
+    struct VirtIOSCSI *parent;
+    Vring vring;
+    EventNotifier host_notifier;
+    EventNotifier guest_notifier;
+} VirtIOSCSIVring;
+
 typedef struct VirtIOSCSICommon {
     VirtIODevice parent_obj;
     VirtIOSCSIConf conf;
@@ -164,14 +176,74 @@ typedef struct VirtIOSCSICommon {
     VirtQueue **cmd_vqs;
 } VirtIOSCSICommon;
 
-typedef struct {
+typedef struct VirtIOSCSI {
     VirtIOSCSICommon parent_obj;
 
     SCSIBus bus;
     int resetting;
     bool events_dropped;
+
+    /* Fields for dataplane below */
+    AioContext *ctx; /* one iothread per virtio-scsi-pci for now */
+
+    /* Vring is used instead of vq in dataplane code, because of the underlying
+     * memory layer thread safety */
+    VirtIOSCSIVring *ctrl_vring;
+    VirtIOSCSIVring *event_vring;
+    VirtIOSCSIVring **cmd_vrings;
+    bool dataplane_started;
+    bool dataplane_starting;
+    bool dataplane_stopping;
+    bool dataplane_disabled;
+    Notifier migration_state_notifier;
 } VirtIOSCSI;
 
+typedef struct VirtIOSCSIReq {
+    VirtIOSCSI *dev;
+    VirtQueue *vq;
+    QEMUSGList qsgl;
+    QEMUIOVector resp_iov;
+
+    /* Note:
+     * - fields before elem are initialized by virtio_scsi_init_req;
+     * - elem is uninitialized at the time of allocation.
+     * - fields after elem are zeroed by virtio_scsi_init_req.
+     * */
+
+    VirtQueueElement elem;
+    /* Set by dataplane code. */
+    VirtIOSCSIVring *vring;
+
+    union {
+        /* Used for two-stage request submission */
+        QTAILQ_ENTRY(VirtIOSCSIReq) next;
+
+        /* Used for cancellation of request during TMFs */
+        int remaining;
+    };
+
+    SCSIRequest *sreq;
+    size_t resp_size;
+    enum SCSIXferMode mode;
+    union {
+        VirtIOSCSICmdResp     cmd;
+        VirtIOSCSICtrlTMFResp tmf;
+        VirtIOSCSICtrlANResp  an;
+        VirtIOSCSIEvent       event;
+    } resp;
+    union {
+        struct {
+            VirtIOSCSICmdReq  cmd;
+            uint8_t           cdb[];
+        } QEMU_PACKED;
+        VirtIOSCSICtrlTMFReq  tmf;
+        VirtIOSCSICtrlANReq   an;
+    } req;
+} VirtIOSCSIReq;
+
+QEMU_BUILD_BUG_ON(offsetof(VirtIOSCSIReq, req.cdb) !=
+                  offsetof(VirtIOSCSIReq, req.cmd) + sizeof(VirtIOSCSICmdReq));
+
 #define DEFINE_VIRTIO_SCSI_PROPERTIES(_state, _conf_field)                     \
     DEFINE_PROP_UINT32("num_queues", _state, _conf_field.num_queues, 1),       \
     DEFINE_PROP_UINT32("max_sectors", _state, _conf_field.max_sectors, 0xFFFF),\
@@ -192,5 +264,19 @@ void virtio_scsi_common_realize(DeviceState *dev, Error **errp,
                                 HandleOutput cmd);
 
 void virtio_scsi_common_unrealize(DeviceState *dev, Error **errp);
+void virtio_scsi_handle_ctrl_req(VirtIOSCSI *s, VirtIOSCSIReq *req);
+bool virtio_scsi_handle_cmd_req_prepare(VirtIOSCSI *s, VirtIOSCSIReq *req);
+void virtio_scsi_handle_cmd_req_submit(VirtIOSCSI *s, VirtIOSCSIReq *req);
+VirtIOSCSIReq *virtio_scsi_init_req(VirtIOSCSI *s, VirtQueue *vq);
+void virtio_scsi_free_req(VirtIOSCSIReq *req);
+void virtio_scsi_push_event(VirtIOSCSI *s, SCSIDevice *dev,
+                            uint32_t event, uint32_t reason);
+
+void virtio_scsi_set_iothread(VirtIOSCSI *s, IOThread *iothread);
+void virtio_scsi_dataplane_start(VirtIOSCSI *s);
+void virtio_scsi_dataplane_stop(VirtIOSCSI *s);
+void virtio_scsi_vring_push_notify(VirtIOSCSIReq *req);
+VirtIOSCSIReq *virtio_scsi_pop_req_vring(VirtIOSCSI *s,
+                                         VirtIOSCSIVring *vring);
 
 #endif /* _QEMU_VIRTIO_SCSI_H */

include/hw/virtio/virtio.h

@@ -161,6 +161,9 @@ typedef struct VirtioDeviceClass {
     int (*load)(VirtIODevice *vdev, QEMUFile *f, int version_id);
 } VirtioDeviceClass;
 
+void virtio_instance_init_common(Object *proxy_obj, void *data,
+                                 size_t vdev_size, const char *vdev_name);
+
 void virtio_init(VirtIODevice *vdev, const char *name,
                          uint16_t device_id, size_t config_size);
 void virtio_cleanup(VirtIODevice *vdev);

include/qemu/bitmap.h

@@ -88,10 +88,19 @@ int slow_bitmap_andnot(unsigned long *dst, const unsigned long *bitmap1,
 int slow_bitmap_intersects(const unsigned long *bitmap1,
                            const unsigned long *bitmap2, long bits);
 
-static inline unsigned long *bitmap_new(long nbits)
+static inline unsigned long *bitmap_try_new(long nbits)
 {
     long len = BITS_TO_LONGS(nbits) * sizeof(unsigned long);
-    return g_malloc0(len);
+    return g_try_malloc0(len);
+}
+
+static inline unsigned long *bitmap_new(long nbits)
+{
+    unsigned long *ptr = bitmap_try_new(nbits);
+    if (ptr == NULL) {
+        abort();
+    }
+    return ptr;
 }
 
 static inline void bitmap_zero(unsigned long *dst, long nbits)