Merge tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost

Pull virtio updates from Michael Tsirkin:
"Several new features here:

- virtio-net is finally supported in vduse

- virtio (balloon and mem) interaction with suspend is improved

- vhost-scsi now handles signals better/faster

And fixes, cleanups all over the place"

* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost: (48 commits)
virtio-pci: Check if is_avq is NULL
virtio: delete vq in vp_find_vqs_msix() when request_irq() fails
MAINTAINERS: add Eugenio Pérez as reviewer
vhost-vdpa: Remove usage of the deprecated ida_simple_xx() API
vp_vdpa: don't allocate unused msix vectors
sound: virtio: drop owner assignment
fuse: virtio: drop owner assignment
scsi: virtio: drop owner assignment
rpmsg: virtio: drop owner assignment
nvdimm: virtio_pmem: drop owner assignment
wifi: mac80211_hwsim: drop owner assignment
vsock/virtio: drop owner assignment
net: 9p: virtio: drop owner assignment
net: virtio: drop owner assignment
net: caif: virtio: drop owner assignment
misc: nsm: drop owner assignment
iommu: virtio: drop owner assignment
drm/virtio: drop owner assignment
gpio: virtio: drop owner assignment
firmware: arm_scmi: virtio: drop owner assignment
...

MAINTAINERS

···
 F:	net/ieee802154/
 F:	net/mac802154/
 
-IFCVF VIRTIO DATA PATH ACCELERATOR
-R:	Zhu Lingshan <[e-mail]>
+Intel VIRTIO DATA PATH ACCELERATOR
+M:	Zhu Lingshan <[e-mail]>
+L:	[e-mail]
+S:	Supported
 F:	drivers/vdpa/ifcvf/
 
 IFE PROTOCOL
···
 M:	Jason Wang <[e-mail]>
 R:	Paolo Bonzini <[e-mail]>
 R:	Stefan Hajnoczi <[e-mail]>
+R:	Eugenio Pérez <[e-mail]>
 L:	[e-mail]
 S:	Maintained
 F:	drivers/block/virtio_blk.c
···
 M:	"Michael S. Tsirkin" <[e-mail]>
 M:	Jason Wang <[e-mail]>
 R:	Xuan Zhuo <[e-mail]>
+R:	Eugenio Pérez <[e-mail]>
 L:	[e-mail]
 S:	Maintained
 F:	Documentation/ABI/testing/sysfs-bus-vdpa
···
 M:	Vivek Goyal <[e-mail]>
 M:	Stefan Hajnoczi <[e-mail]>
 M:	Miklos Szeredi <[e-mail]>
+R:	Eugenio Pérez <[e-mail]>
 L:	[e-mail]
 L:	[e-mail]
 S:	Supported
···
 VIRTIO HOST (VHOST)
 M:	"Michael S. Tsirkin" <[e-mail]>
 M:	Jason Wang <[e-mail]>
+R:	Eugenio Pérez <[e-mail]>
 L:	[e-mail]
 L:	[e-mail]
 L:	[e-mail]

arch/um/drivers/virt-pci.c

···
 
 static struct virtio_driver um_pci_virtio_driver = {
 	.driver.name = "virtio-pci",
-	.driver.owner = THIS_MODULE,
 	.id_table = id_table,
 	.probe = um_pci_virtio_probe,
 	.remove = um_pci_virtio_remove,

drivers/block/virtio_blk.c

···
 	.feature_table_legacy		= features_legacy,
 	.feature_table_size_legacy	= ARRAY_SIZE(features_legacy),
 	.driver.name			= KBUILD_MODNAME,
-	.driver.owner			= THIS_MODULE,
 	.id_table			= id_table,
 	.probe				= virtblk_probe,
 	.remove				= virtblk_remove,

drivers/bluetooth/virtio_bt.c

···
 
 static struct virtio_driver virtbt_driver = {
 	.driver.name         = KBUILD_MODNAME,
-	.driver.owner        = THIS_MODULE,
 	.feature_table       = virtbt_features,
 	.feature_table_size  = ARRAY_SIZE(virtbt_features),
 	.id_table            = virtbt_table,

drivers/char/hw_random/virtio-rng.c

···
 
 static struct virtio_driver virtio_rng_driver = {
 	.driver.name =	KBUILD_MODNAME,
-	.driver.owner =	THIS_MODULE,
 	.id_table =	id_table,
 	.probe =	virtrng_probe,
 	.remove =	virtrng_remove,

drivers/char/virtio_console.c

···
 	.feature_table = features,
 	.feature_table_size = ARRAY_SIZE(features),
 	.driver.name =	KBUILD_MODNAME,
-	.driver.owner =	THIS_MODULE,
 	.id_table =	id_table,
 	.probe =	virtcons_probe,
 	.remove =	virtcons_remove,
···
 	.feature_table = rproc_serial_features,
 	.feature_table_size = ARRAY_SIZE(rproc_serial_features),
 	.driver.name =	"virtio_rproc_serial",
-	.driver.owner =	THIS_MODULE,
 	.id_table =	rproc_serial_id_table,
 	.probe =	virtcons_probe,
 	.remove =	virtcons_remove,

drivers/crypto/virtio/virtio_crypto_core.c

···
 
 static struct virtio_driver virtio_crypto_driver = {
 	.driver.name         = KBUILD_MODNAME,
-	.driver.owner        = THIS_MODULE,
 	.feature_table       = features,
 	.feature_table_size  = ARRAY_SIZE(features),
 	.id_table            = id_table,

drivers/firmware/arm_scmi/virtio.c

···
 
 static struct virtio_driver virtio_scmi_driver = {
 	.driver.name = "scmi-virtio",
-	.driver.owner = THIS_MODULE,
 	.feature_table = features,
 	.feature_table_size = ARRAY_SIZE(features),
 	.id_table = id_table,

drivers/gpio/gpio-virtio.c

···
 	.remove			= virtio_gpio_remove,
 	.driver			= {
 		.name		= KBUILD_MODNAME,
-		.owner		= THIS_MODULE,
 	},
 };
 module_virtio_driver(virtio_gpio_driver);

drivers/gpu/drm/virtio/virtgpu_drv.c

···
 	.feature_table = features,
 	.feature_table_size = ARRAY_SIZE(features),
 	.driver.name = KBUILD_MODNAME,
-	.driver.owner = THIS_MODULE,
 	.id_table = id_table,
 	.probe = virtio_gpu_probe,
 	.remove = virtio_gpu_remove,

drivers/iommu/virtio-iommu.c

···
 
 static struct virtio_driver virtio_iommu_drv = {
 	.driver.name		= KBUILD_MODNAME,
-	.driver.owner		= THIS_MODULE,
 	.id_table		= id_table,
 	.feature_table		= features,
 	.feature_table_size	= ARRAY_SIZE(features),

drivers/misc/nsm.c

···
 	.feature_table_legacy      = 0,
 	.feature_table_size_legacy = 0,
 	.driver.name               = KBUILD_MODNAME,
-	.driver.owner              = THIS_MODULE,
 	.id_table                  = id_table,
 	.probe                     = nsm_device_probe,
 	.remove                    = nsm_device_remove,

drivers/net/caif/caif_virtio.c

···
 	.feature_table		= features,
 	.feature_table_size	= ARRAY_SIZE(features),
 	.driver.name		= KBUILD_MODNAME,
-	.driver.owner		= THIS_MODULE,
 	.id_table		= id_table,
 	.probe			= cfv_probe,
 	.remove			= cfv_remove,

drivers/net/virtio_net.c

···
 	.feature_table_legacy = features_legacy,
 	.feature_table_size_legacy = ARRAY_SIZE(features_legacy),
 	.driver.name =	KBUILD_MODNAME,
-	.driver.owner =	THIS_MODULE,
 	.id_table =	id_table,
 	.validate =	virtnet_validate,
 	.probe =	virtnet_probe,

drivers/net/wireless/virtual/mac80211_hwsim.c

···
 
 static struct virtio_driver virtio_hwsim = {
 	.driver.name = KBUILD_MODNAME,
-	.driver.owner = THIS_MODULE,
 	.id_table = id_table,
 	.probe = hwsim_virtio_probe,
 	.remove = hwsim_virtio_remove,

drivers/nvdimm/virtio_pmem.c

···
 	.feature_table		= features,
 	.feature_table_size	= ARRAY_SIZE(features),
 	.driver.name		= KBUILD_MODNAME,
-	.driver.owner		= THIS_MODULE,
 	.id_table		= id_table,
 	.validate		= virtio_pmem_validate,
 	.probe			= virtio_pmem_probe,

drivers/rpmsg/virtio_rpmsg_bus.c

···
 	.feature_table	= features,
 	.feature_table_size = ARRAY_SIZE(features),
 	.driver.name	= KBUILD_MODNAME,
-	.driver.owner	= THIS_MODULE,
 	.id_table	= id_table,
 	.probe		= rpmsg_probe,
 	.remove		= rpmsg_remove,

drivers/scsi/virtio_scsi.c

···
 	.feature_table = features,
 	.feature_table_size = ARRAY_SIZE(features),
 	.driver.name = KBUILD_MODNAME,
-	.driver.owner = THIS_MODULE,
 	.id_table = id_table,
 	.probe = virtscsi_probe,
 #ifdef CONFIG_PM_SLEEP

drivers/vdpa/vdpa.c

···
 	ssize_t len;
 
 	device_lock(dev);
-	len = snprintf(buf, PAGE_SIZE, "%s\n", vdev->driver_override);
+	len = sysfs_emit(buf, "%s\n", vdev->driver_override);
 	device_unlock(dev);
 
 	return len;

drivers/vdpa/vdpa_user/vduse_dev.c

···
  *
  */
 
+#include "linux/virtio_net.h"
 #include <linux/init.h>
 #include <linux/module.h>
 #include <linux/cdev.h>
···
 #include <uapi/linux/virtio_config.h>
 #include <uapi/linux/virtio_ids.h>
 #include <uapi/linux/virtio_blk.h>
+#include <uapi/linux/virtio_ring.h>
 #include <linux/mod_devicetable.h>
 
 #include "iova_domain.h"
···
 
 static u32 allowed_device_id[] = {
 	VIRTIO_ID_BLOCK,
+	VIRTIO_ID_NET,
 };
 
 static inline struct vduse_dev *vdpa_to_vduse(struct vdpa_device *vdpa)
···
 	return false;
 }
 
-static bool features_is_valid(u64 features)
+static bool features_is_valid(struct vduse_dev_config *config)
 {
-	if (!(features & (1ULL << VIRTIO_F_ACCESS_PLATFORM)))
+	if (!(config->features & BIT_ULL(VIRTIO_F_ACCESS_PLATFORM)))
 		return false;
 
 	/* Now we only support read-only configuration space */
-	if (features & (1ULL << VIRTIO_BLK_F_CONFIG_WCE))
+	if ((config->device_id == VIRTIO_ID_BLOCK) &&
+			(config->features & BIT_ULL(VIRTIO_BLK_F_CONFIG_WCE)))
+		return false;
+	else if ((config->device_id == VIRTIO_ID_NET) &&
+			(config->features & BIT_ULL(VIRTIO_NET_F_CTRL_VQ)))
+		return false;
+
+	if ((config->device_id == VIRTIO_ID_NET) &&
+			!(config->features & BIT_ULL(VIRTIO_F_VERSION_1)))
 		return false;
 
 	return true;
···
 	if (!device_is_allowed(config->device_id))
 		return false;
 
-	if (!features_is_valid(config->features))
+	if (!features_is_valid(config))
 		return false;
 
 	return true;
···
 {
 	int ret;
 	struct vduse_dev *dev;
+
+	ret = -EPERM;
+	if ((config->device_id == VIRTIO_ID_NET) && !capable(CAP_NET_ADMIN))
+		goto err;
 
 	ret = -EEXIST;
 	if (vduse_find_dev(config->name))
···
 
 static struct virtio_device_id id_table[] = {
 	{ VIRTIO_ID_BLOCK, VIRTIO_DEV_ANY_ID },
+	{ VIRTIO_ID_NET, VIRTIO_DEV_ANY_ID },
 	{ 0 },
 };
 

drivers/vdpa/virtio_pci/vp_vdpa.c

···
 	struct pci_dev *pdev = mdev->pci_dev;
 	int i, ret, irq;
 	int queues = vp_vdpa->queues;
-	int vectors = queues + 1;
+	int vectors = 1;
+	int msix_vec = 0;
+
+	for (i = 0; i < queues; i++) {
+		if (vp_vdpa->vring[i].cb.callback)
+			vectors++;
+	}
 
 	ret = pci_alloc_irq_vectors(pdev, vectors, vectors, PCI_IRQ_MSIX);
 	if (ret != vectors) {
···
 	vp_vdpa->vectors = vectors;
 
 	for (i = 0; i < queues; i++) {
+		if (!vp_vdpa->vring[i].cb.callback)
+			continue;
+
 		snprintf(vp_vdpa->vring[i].msix_name, VP_VDPA_NAME_SIZE,
 			"vp-vdpa[%s]-%d\n", pci_name(pdev), i);
-		irq = pci_irq_vector(pdev, i);
+		irq = pci_irq_vector(pdev, msix_vec);
 		ret = devm_request_irq(&pdev->dev, irq,
 				       vp_vdpa_vq_handler,
 				       0, vp_vdpa->vring[i].msix_name,
···
 				"vp_vdpa: fail to request irq for vq %d\n", i);
 			goto err;
 		}
-		vp_modern_queue_vector(mdev, i, i);
+		vp_modern_queue_vector(mdev, i, msix_vec);
 		vp_vdpa->vring[i].irq = irq;
+		msix_vec++;
 	}
 
 	snprintf(vp_vdpa->msix_name, VP_VDPA_NAME_SIZE, "vp-vdpa[%s]-config\n",
 		 pci_name(pdev));
-	irq = pci_irq_vector(pdev, queues);
+	irq = pci_irq_vector(pdev, msix_vec);
 	ret = devm_request_irq(&pdev->dev, irq,	vp_vdpa_config_handler, 0,
 			       vp_vdpa->msix_name, vp_vdpa);
 	if (ret) {
 		dev_err(&pdev->dev,
-			"vp_vdpa: fail to request irq for vq %d\n", i);
+			"vp_vdpa: fail to request irq for config: %d\n", ret);
 			goto err;
 	}
-	vp_modern_config_vector(mdev, queues);
+	vp_modern_config_vector(mdev, msix_vec);
 	vp_vdpa->config_irq = irq;
 
 	return 0;
···
 
 	if (status & VIRTIO_CONFIG_S_DRIVER_OK &&
 	    !(s & VIRTIO_CONFIG_S_DRIVER_OK)) {
-		vp_vdpa_request_irq(vp_vdpa);
+		if (vp_vdpa_request_irq(vp_vdpa)) {
+			WARN_ON(1);
+			return;
+		}
 	}
 
 	vp_modern_set_status(mdev, status);

drivers/vhost/scsi.c

···
 
 struct vhost_scsi_tmf {
 	struct vhost_work vwork;
+	struct work_struct flush_work;
 	struct vhost_scsi *vhost;
 	struct vhost_scsi_virtqueue *svq;
 
···
 	vhost_scsi_put_inflight(inflight);
 }
 
+static void vhost_scsi_drop_cmds(struct vhost_scsi_virtqueue *svq)
+{
+	struct vhost_scsi_cmd *cmd, *t;
+	struct llist_node *llnode;
+
+	llnode = llist_del_all(&svq->completion_list);
+	llist_for_each_entry_safe(cmd, t, llnode, tvc_completion_list)
+		vhost_scsi_release_cmd_res(&cmd->tvc_se_cmd);
+}
+
 static void vhost_scsi_release_cmd(struct se_cmd *se_cmd)
 {
 	if (se_cmd->se_cmd_flags & SCF_SCSI_TMR_CDB) {
 		struct vhost_scsi_tmf *tmf = container_of(se_cmd,
 					struct vhost_scsi_tmf, se_cmd);
-		struct vhost_virtqueue *vq = &tmf->svq->vq;
 
-		vhost_vq_work_queue(vq, &tmf->vwork);
+		schedule_work(&tmf->flush_work);
 	} else {
 		struct vhost_scsi_cmd *cmd = container_of(se_cmd,
 					struct vhost_scsi_cmd, tvc_se_cmd);
···
 					struct vhost_scsi_virtqueue, vq);
 
 		llist_add(&cmd->tvc_completion_list, &svq->completion_list);
-		vhost_vq_work_queue(&svq->vq, &svq->completion_work);
+		if (!vhost_vq_work_queue(&svq->vq, &svq->completion_work))
+			vhost_scsi_drop_cmds(svq);
 	}
 }
 
···
 		vq_err(vq, "Faulted on vhost_scsi_send_event\n");
 }
 
-static void vhost_scsi_evt_work(struct vhost_work *work)
+static void vhost_scsi_complete_events(struct vhost_scsi *vs, bool drop)
 {
-	struct vhost_scsi *vs = container_of(work, struct vhost_scsi,
-					vs_event_work);
 	struct vhost_virtqueue *vq = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
 	struct vhost_scsi_evt *evt, *t;
 	struct llist_node *llnode;
···
 	mutex_lock(&vq->mutex);
 	llnode = llist_del_all(&vs->vs_event_list);
 	llist_for_each_entry_safe(evt, t, llnode, list) {
-		vhost_scsi_do_evt_work(vs, evt);
+		if (!drop)
+			vhost_scsi_do_evt_work(vs, evt);
 		vhost_scsi_free_evt(vs, evt);
 	}
 	mutex_unlock(&vq->mutex);
+}
+
+static void vhost_scsi_evt_work(struct vhost_work *work)
+{
+	struct vhost_scsi *vs = container_of(work, struct vhost_scsi,
+					     vs_event_work);
+	vhost_scsi_complete_events(vs, false);
 }
 
 static int vhost_scsi_copy_sgl_to_iov(struct vhost_scsi_cmd *cmd)
···
 {
 	struct vhost_scsi_tmf *tmf = container_of(work, struct vhost_scsi_tmf,
 						  vwork);
-	struct vhost_virtqueue *ctl_vq, *vq;
-	int resp_code, i;
+	int resp_code;
 
-	if (tmf->scsi_resp == TMR_FUNCTION_COMPLETE) {
-		/*
-		 * Flush IO vqs that don't share a worker with the ctl to make
-		 * sure they have sent their responses before us.
-		 */
-		ctl_vq = &tmf->vhost->vqs[VHOST_SCSI_VQ_CTL].vq;
-		for (i = VHOST_SCSI_VQ_IO; i < tmf->vhost->dev.nvqs; i++) {
-			vq = &tmf->vhost->vqs[i].vq;
-
-			if (vhost_vq_is_setup(vq) &&
-			    vq->worker != ctl_vq->worker)
-				vhost_vq_flush(vq);
-		}
-
+	if (tmf->scsi_resp == TMR_FUNCTION_COMPLETE)
 		resp_code = VIRTIO_SCSI_S_FUNCTION_SUCCEEDED;
-	} else {
+	else
 		resp_code = VIRTIO_SCSI_S_FUNCTION_REJECTED;
-	}
 
 	vhost_scsi_send_tmf_resp(tmf->vhost, &tmf->svq->vq, tmf->in_iovs,
 				 tmf->vq_desc, &tmf->resp_iov, resp_code);
 	vhost_scsi_release_tmf_res(tmf);
+}
+
+static void vhost_scsi_tmf_flush_work(struct work_struct *work)
+{
+	struct vhost_scsi_tmf *tmf = container_of(work, struct vhost_scsi_tmf,
+						 flush_work);
+	struct vhost_virtqueue *vq = &tmf->svq->vq;
+	/*
+	 * Make sure we have sent responses for other commands before we
+	 * send our response.
+	 */
+	vhost_dev_flush(vq->dev);
+	if (!vhost_vq_work_queue(vq, &tmf->vwork))
+		vhost_scsi_release_tmf_res(tmf);
 }
 
 static void
···
 	if (!tmf)
 		goto send_reject;
 
+	INIT_WORK(&tmf->flush_work, vhost_scsi_tmf_flush_work);
 	vhost_work_init(&tmf->vwork, vhost_scsi_tmf_resp_work);
 	tmf->vhost = vs;
 	tmf->svq = svq;
···
 	}
 
 	llist_add(&evt->list, &vs->vs_event_list);
-	vhost_vq_work_queue(vq, &vs->vs_event_work);
+	if (!vhost_vq_work_queue(vq, &vs->vs_event_work))
+		vhost_scsi_complete_events(vs, true);
 }
 
 static void vhost_scsi_evt_handle_kick(struct vhost_work *work)

drivers/vhost/vdpa.c

···
 	struct vhost_vdpa *v =
 	       container_of(device, struct vhost_vdpa, dev);
 
-	ida_simple_remove(&vhost_vdpa_ida, v->minor);
+	ida_free(&vhost_vdpa_ida, v->minor);
 	kfree(v->vqs);
 	kfree(v);
 }
···
 	if (!v)
 		return -ENOMEM;
 
-	minor = ida_simple_get(&vhost_vdpa_ida, 0,
-			       VHOST_VDPA_DEV_MAX, GFP_KERNEL);
+	minor = ida_alloc_max(&vhost_vdpa_ida, VHOST_VDPA_DEV_MAX - 1,
+			      GFP_KERNEL);
 	if (minor < 0) {
 		kfree(v);
 		return minor;

drivers/vhost/vhost.c

···
 }
 EXPORT_SYMBOL_GPL(vhost_vq_work_queue);
 
-void vhost_vq_flush(struct vhost_virtqueue *vq)
-{
-	struct vhost_flush_struct flush;
-
-	init_completion(&flush.wait_event);
-	vhost_work_init(&flush.work, vhost_flush_work);
-
-	if (vhost_vq_work_queue(vq, &flush.work))
-		wait_for_completion(&flush.wait_event);
-}
-EXPORT_SYMBOL_GPL(vhost_vq_flush);
-
 /**
- * vhost_worker_flush - flush a worker
+ * __vhost_worker_flush - flush a worker
  * @worker: worker to flush
  *
- * This does not use RCU to protect the worker, so the device or worker
- * mutex must be held.
+ * The worker's flush_mutex must be held.
  */
-static void vhost_worker_flush(struct vhost_worker *worker)
+static void __vhost_worker_flush(struct vhost_worker *worker)
 {
 	struct vhost_flush_struct flush;
+
+	if (!worker->attachment_cnt || worker->killed)
+		return;
 
 	init_completion(&flush.wait_event);
 	vhost_work_init(&flush.work, vhost_flush_work);
 
 	vhost_worker_queue(worker, &flush.work);
+	/*
+	 * Drop mutex in case our worker is killed and it needs to take the
+	 * mutex to force cleanup.
+	 */
+	mutex_unlock(&worker->mutex);
 	wait_for_completion(&flush.wait_event);
+	mutex_lock(&worker->mutex);
+}
+
+static void vhost_worker_flush(struct vhost_worker *worker)
+{
+	mutex_lock(&worker->mutex);
+	__vhost_worker_flush(worker);
+	mutex_unlock(&worker->mutex);
 }
 
 void vhost_dev_flush(struct vhost_dev *dev)
···
 	struct vhost_worker *worker;
 	unsigned long i;
 
-	xa_for_each(&dev->worker_xa, i, worker) {
-		mutex_lock(&worker->mutex);
-		if (!worker->attachment_cnt) {
-			mutex_unlock(&worker->mutex);
-			continue;
-		}
+	xa_for_each(&dev->worker_xa, i, worker)
 		vhost_worker_flush(worker);
-		mutex_unlock(&worker->mutex);
-	}
 }
 EXPORT_SYMBOL_GPL(vhost_dev_flush);
 
···
 	__vhost_vq_meta_reset(vq);
 }
 
-static bool vhost_worker(void *data)
+static bool vhost_run_work_list(void *data)
 {
 	struct vhost_worker *worker = data;
 	struct vhost_work *work, *work_next;
···
 	}
 
 	return !!node;
+}
+
+static void vhost_worker_killed(void *data)
+{
+	struct vhost_worker *worker = data;
+	struct vhost_dev *dev = worker->dev;
+	struct vhost_virtqueue *vq;
+	int i, attach_cnt = 0;
+
+	mutex_lock(&worker->mutex);
+	worker->killed = true;
+
+	for (i = 0; i < dev->nvqs; i++) {
+		vq = dev->vqs[i];
+
+		mutex_lock(&vq->mutex);
+		if (worker ==
+		    rcu_dereference_check(vq->worker,
+					  lockdep_is_held(&vq->mutex))) {
+			rcu_assign_pointer(vq->worker, NULL);
+			attach_cnt++;
+		}
+		mutex_unlock(&vq->mutex);
+	}
+
+	worker->attachment_cnt -= attach_cnt;
+	if (attach_cnt)
+		synchronize_rcu();
+	/*
+	 * Finish vhost_worker_flush calls and any other works that snuck in
+	 * before the synchronize_rcu.
+	 */
+	vhost_run_work_list(worker);
+	mutex_unlock(&worker->mutex);
 }
 
 static void vhost_vq_free_iovecs(struct vhost_virtqueue *vq)
···
 	if (!worker)
 		return NULL;
 
+	worker->dev = dev;
 	snprintf(name, sizeof(name), "vhost-%d", current->pid);
 
-	vtsk = vhost_task_create(vhost_worker, worker, name);
+	vtsk = vhost_task_create(vhost_run_work_list, vhost_worker_killed,
+				 worker, name);
 	if (!vtsk)
 		goto free_worker;
 
···
 {
 	struct vhost_worker *old_worker;
 
-	old_worker = rcu_dereference_check(vq->worker,
-					   lockdep_is_held(&vq->dev->mutex));
-
 	mutex_lock(&worker->mutex);
-	worker->attachment_cnt++;
-	mutex_unlock(&worker->mutex);
-	rcu_assign_pointer(vq->worker, worker);
-
-	if (!old_worker)
+	if (worker->killed) {
+		mutex_unlock(&worker->mutex);
 		return;
+	}
+
+	mutex_lock(&vq->mutex);
+
+	old_worker = rcu_dereference_check(vq->worker,
+					   lockdep_is_held(&vq->mutex));
+	rcu_assign_pointer(vq->worker, worker);
+	worker->attachment_cnt++;
+
+	if (!old_worker) {
+		mutex_unlock(&vq->mutex);
+		mutex_unlock(&worker->mutex);
+		return;
+	}
+	mutex_unlock(&vq->mutex);
+	mutex_unlock(&worker->mutex);
+
 	/*
 	 * Take the worker mutex to make sure we see the work queued from
 	 * device wide flushes which doesn't use RCU for execution.
 	 */
 	mutex_lock(&old_worker->mutex);
-	old_worker->attachment_cnt--;
+	if (old_worker->killed) {
+		mutex_unlock(&old_worker->mutex);
+		return;
+	}
+
 	/*
 	 * We don't want to call synchronize_rcu for every vq during setup
 	 * because it will slow down VM startup. If we haven't done
···
 	mutex_lock(&vq->mutex);
 	if (!vhost_vq_get_backend(vq) && !vq->kick) {
 		mutex_unlock(&vq->mutex);
+
+		old_worker->attachment_cnt--;
 		mutex_unlock(&old_worker->mutex);
 		/*
 		 * vsock can queue anytime after VHOST_VSOCK_SET_GUEST_CID.
···
 	/* Make sure new vq queue/flush/poll calls see the new worker */
 	synchronize_rcu();
 	/* Make sure whatever was queued gets run */
-	vhost_worker_flush(old_worker);
+	__vhost_worker_flush(old_worker);
+	old_worker->attachment_cnt--;
 	mutex_unlock(&old_worker->mutex);
 }
 
···
 		return -ENODEV;
 
 	mutex_lock(&worker->mutex);
-	if (worker->attachment_cnt) {
+	if (worker->attachment_cnt || worker->killed) {
 		mutex_unlock(&worker->mutex);
 		return -EBUSY;
 	}
+	/*
+	 * A flush might have raced and snuck in before attachment_cnt was set
+	 * to zero. Make sure flushes are flushed from the queue before
+	 * freeing.
+	 */
+	__vhost_worker_flush(worker);
 	mutex_unlock(&worker->mutex);
 
 	vhost_worker_destroy(dev, worker);

drivers/vhost/vhost.h

···
 
 struct vhost_worker {
 	struct vhost_task	*vtsk;
+	struct vhost_dev	*dev;
 	/* Used to serialize device wide flushing with worker swapping. */
 	struct mutex		mutex;
 	struct llist_head	work_list;
 	u64			kcov_handle;
 	u32			id;
 	int			attachment_cnt;
+	bool			killed;
 };
 
 /* Poll a file (eventfd or socket) */
···
 		      struct vhost_log *log, unsigned int *log_num);
 void vhost_discard_vq_desc(struct vhost_virtqueue *, int n);
 
-void vhost_vq_flush(struct vhost_virtqueue *vq);
 bool vhost_vq_work_queue(struct vhost_virtqueue *vq, struct vhost_work *work);
 bool vhost_vq_has_work(struct vhost_virtqueue *vq);
 bool vhost_vq_is_setup(struct vhost_virtqueue *vq);

drivers/virtio/virtio_balloon.c

···
 	struct page_reporting_dev_info pr_dev_info;
 
 	/* State for keeping the wakeup_source active while adjusting the balloon */
-	spinlock_t adjustment_lock;
-	bool adjustment_signal_pending;
-	bool adjustment_in_progress;
+	spinlock_t wakeup_lock;
+	bool processing_wakeup_event;
+	u32 wakeup_signal_mask;
 };
+
+#define VIRTIO_BALLOON_WAKEUP_SIGNAL_ADJUST (1 << 0)
+#define VIRTIO_BALLOON_WAKEUP_SIGNAL_STATS (1 << 1)
 
 static const struct virtio_device_id id_table[] = {
 	{ VIRTIO_ID_BALLOON, VIRTIO_DEV_ANY_ID },
···
 	BUILD_BUG_ON(PAGE_SHIFT < VIRTIO_BALLOON_PFN_SHIFT);
 	/* Convert pfn from Linux page size to balloon page size. */
 	return pfn * VIRTIO_BALLOON_PAGES_PER_PAGE;
+}
+
+static void start_wakeup_event(struct virtio_balloon *vb, u32 mask)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&vb->wakeup_lock, flags);
+	vb->wakeup_signal_mask |= mask;
+	if (!vb->processing_wakeup_event) {
+		vb->processing_wakeup_event = true;
+		pm_stay_awake(&vb->vdev->dev);
+	}
+	spin_unlock_irqrestore(&vb->wakeup_lock, flags);
+}
+
+static void process_wakeup_event(struct virtio_balloon *vb, u32 mask)
+{
+	spin_lock_irq(&vb->wakeup_lock);
+	vb->wakeup_signal_mask &= ~mask;
+	spin_unlock_irq(&vb->wakeup_lock);
+}
+
+static void finish_wakeup_event(struct virtio_balloon *vb)
+{
+	spin_lock_irq(&vb->wakeup_lock);
+	if (!vb->wakeup_signal_mask && vb->processing_wakeup_event) {
+		vb->processing_wakeup_event = false;
+		pm_relax(&vb->vdev->dev);
+	}
+	spin_unlock_irq(&vb->wakeup_lock);
 }
 
 static void balloon_ack(struct virtqueue *vq)
···
 	struct virtio_balloon *vb = vq->vdev->priv;
 
 	spin_lock(&vb->stop_update_lock);
-	if (!vb->stop_update)
+	if (!vb->stop_update) {
+		start_wakeup_event(vb, VIRTIO_BALLOON_WAKEUP_SIGNAL_STATS);
 		queue_work(system_freezable_wq, &vb->update_balloon_stats_work);
+	}
 	spin_unlock(&vb->stop_update_lock);
 }
 
···
 
 static void start_update_balloon_size(struct virtio_balloon *vb)
 {
-	unsigned long flags;
-
-	spin_lock_irqsave(&vb->adjustment_lock, flags);
-	vb->adjustment_signal_pending = true;
-	if (!vb->adjustment_in_progress) {
-		vb->adjustment_in_progress = true;
-		pm_stay_awake(vb->vdev->dev.parent);
-	}
-	spin_unlock_irqrestore(&vb->adjustment_lock, flags);
-
+	start_wakeup_event(vb, VIRTIO_BALLOON_WAKEUP_SIGNAL_ADJUST);
 	queue_work(system_freezable_wq, &vb->update_balloon_size_work);
-}
-
-static void end_update_balloon_size(struct virtio_balloon *vb)
-{
-	spin_lock_irq(&vb->adjustment_lock);
-	if (!vb->adjustment_signal_pending && vb->adjustment_in_progress) {
-		vb->adjustment_in_progress = false;
-		pm_relax(vb->vdev->dev.parent);
-	}
-	spin_unlock_irq(&vb->adjustment_lock);
 }
 
 static void virtballoon_changed(struct virtio_device *vdev)
···
 
 	vb = container_of(work, struct virtio_balloon,
 			  update_balloon_stats_work);
+
+	process_wakeup_event(vb, VIRTIO_BALLOON_WAKEUP_SIGNAL_STATS);
 	stats_handle_request(vb);
+	finish_wakeup_event(vb);
 }
 
 static void update_balloon_size_func(struct work_struct *work)
···
 	vb = container_of(work, struct virtio_balloon,
 			  update_balloon_size_work);
 
-	spin_lock_irq(&vb->adjustment_lock);
-	vb->adjustment_signal_pending = false;
-	spin_unlock_irq(&vb->adjustment_lock);
+	process_wakeup_event(vb, VIRTIO_BALLOON_WAKEUP_SIGNAL_ADJUST);
 
 	diff = towards_target(vb);
 
···
 	if (diff)
 		queue_work(system_freezable_wq, work);
 	else
-		end_update_balloon_size(vb);
+		finish_wakeup_event(vb);
 }
 
 static int init_vqs(struct virtio_balloon *vb)
···
 			goto out_unregister_oom;
 	}
 
-	spin_lock_init(&vb->adjustment_lock);
+	spin_lock_init(&vb->wakeup_lock);
+
+	/*
+	 * The virtio balloon itself can't wake up the device, but it is
+	 * responsible for processing wakeup events passed up from the transport
+	 * layer. Wakeup sources don't support nesting/chaining calls, so we use
+	 * our own wakeup source to ensure wakeup events are properly handled
+	 * without trampling on the transport layer's wakeup source.
+	 */
+	device_set_wakeup_capable(&vb->vdev->dev, true);
 
 	virtio_device_ready(vdev);
 
···
 	.feature_table = features,
 	.feature_table_size = ARRAY_SIZE(features),
 	.driver.name =	KBUILD_MODNAME,
-	.driver.owner =	THIS_MODULE,
 	.id_table =	id_table,
 	.validate =	virtballoon_validate,
 	.probe =	virtballoon_probe,

drivers/virtio/virtio_input.c

···
 
 static struct virtio_driver virtio_input_driver = {
 	.driver.name         = KBUILD_MODNAME,
-	.driver.owner        = THIS_MODULE,
 	.feature_table       = features,
 	.feature_table_size  = ARRAY_SIZE(features),
 	.id_table            = id_table,

drivers/virtio/virtio_mem.c

···
 #include <linux/lockdep.h>
 #include <linux/log2.h>
 #include <linux/vmalloc.h>
+#include <linux/suspend.h>
 
 #include <acpi/acpi_numa.h>
 
···
 
 	/* Memory notifier (online/offline events). */
 	struct notifier_block memory_notifier;
+
+	/* Notifier to block hibernation image storing/reloading. */
+	struct notifier_block pm_notifier;
 
 #ifdef CONFIG_PROC_VMCORE
 	/* vmcore callback for /proc/vmcore handling in kdump mode */
···
 	lockdep_on();
 
 	return rc;
+}
+
+static int virtio_mem_pm_notifier_cb(struct notifier_block *nb,
+				     unsigned long action, void *arg)
+{
+	struct virtio_mem *vm = container_of(nb, struct virtio_mem,
+					     pm_notifier);
+	switch (action) {
+	case PM_HIBERNATION_PREPARE:
+	case PM_RESTORE_PREPARE:
+		/*
+		 * When restarting the VM, all memory is unplugged. Don't
+		 * allow to hibernate and restore from an image.
+		 */
+		dev_err(&vm->vdev->dev, "hibernation is not supported.\n");
+		return NOTIFY_BAD;
+	default:
+		return NOTIFY_OK;
+	}
 }
 
 /*
···
 	rc = register_memory_notifier(&vm->memory_notifier);
 	if (rc)
 		goto out_unreg_group;
-	rc = register_virtio_mem_device(vm);
+	/* Block hibernation as early as possible. */
+	vm->pm_notifier.priority = INT_MAX;
+	vm->pm_notifier.notifier_call = virtio_mem_pm_notifier_cb;
+	rc = register_pm_notifier(&vm->pm_notifier);
 	if (rc)
 		goto out_unreg_mem;
+	rc = register_virtio_mem_device(vm);
+	if (rc)
+		goto out_unreg_pm;
 
 	return 0;
+out_unreg_pm:
+	unregister_pm_notifier(&vm->pm_notifier);
 out_unreg_mem:
 	unregister_memory_notifier(&vm->memory_notifier);
 out_unreg_group:
···
 
 	/* unregister callbacks */
 	unregister_virtio_mem_device(vm);
+	unregister_pm_notifier(&vm->pm_notifier);
 	unregister_memory_notifier(&vm->memory_notifier);
 
 	/*
···
 #ifdef CONFIG_PM_SLEEP
 static int virtio_mem_freeze(struct virtio_device *vdev)
 {
+	struct virtio_mem *vm = vdev->priv;
+
 	/*
-	 * When restarting the VM, all memory is usually unplugged. Don't
-	 * allow to suspend/hibernate.
+	 * We block hibernation using the PM notifier completely. The workqueue
+	 * is already frozen by the PM core at this point, so we simply
+	 * reset the device and cleanup the queues.
 	 */
-	dev_err(&vdev->dev, "save/restore not supported.\n");
-	return -EPERM;
+	if (pm_suspend_target_state != PM_SUSPEND_TO_IDLE &&
+	    vm->plugged_size &&
+	    !virtio_has_feature(vm->vdev, VIRTIO_MEM_F_PERSISTENT_SUSPEND)) {
+		dev_err(&vm->vdev->dev,
+			"suspending with plugged memory is not supported\n");
+		return -EPERM;
+	}
+
+	virtio_reset_device(vdev);
+	vdev->config->del_vqs(vdev);
+	vm->vq = NULL;
+	return 0;
 }
 
 static int virtio_mem_restore(struct virtio_device *vdev)
 {
-	return -EPERM;
+	struct virtio_mem *vm = vdev->priv;
+	int ret;
+
+	ret = virtio_mem_init_vq(vm);
+	if (ret)
+		return ret;
+	virtio_device_ready(vdev);
+
+	/* Let's check if anything changed. */
+	virtio_mem_config_changed(vdev);
+	return 0;
 }
 #endif
 
···
 	VIRTIO_MEM_F_ACPI_PXM,
 #endif
 	VIRTIO_MEM_F_UNPLUGGED_INACCESSIBLE,
+	VIRTIO_MEM_F_PERSISTENT_SUSPEND,
 };
 
 static const struct virtio_device_id virtio_mem_id_table[] = {
···
 	.feature_table = virtio_mem_features,
 	.feature_table_size = ARRAY_SIZE(virtio_mem_features),
 	.driver.name = KBUILD_MODNAME,
-	.driver.owner = THIS_MODULE,
 	.id_table = virtio_mem_id_table,
 	.probe = virtio_mem_probe,
 	.remove = virtio_mem_remove,

drivers/virtio/virtio_mmio.c

···
 	return rc;
 }
 
-static int virtio_mmio_remove(struct platform_device *pdev)
+static void virtio_mmio_remove(struct platform_device *pdev)
 {
 	struct virtio_mmio_device *vm_dev = platform_get_drvdata(pdev);
 	unregister_virtio_device(&vm_dev->vdev);
-
-	return 0;
 }
 
 
···
 
 static struct platform_driver virtio_mmio_driver = {
 	.probe		= virtio_mmio_probe,
-	.remove		= virtio_mmio_remove,
+	.remove_new	= virtio_mmio_remove,
 	.driver		= {
 		.name	= "virtio-mmio",
 		.of_match_table	= virtio_mmio_match,

drivers/virtio/virtio_pci_common.c

···
 	int i;
 
 	list_for_each_entry_safe(vq, n, &vdev->vqs, list) {
-		if (vp_dev->is_avq(vdev, vq->index))
+		if (vp_dev->is_avq && vp_dev->is_avq(vdev, vq->index))
 			continue;
 
 		if (vp_dev->per_vq_vectors) {
···
 				  vring_interrupt, 0,
 				  vp_dev->msix_names[msix_vec],
 				  vqs[i]);
-		if (err)
+		if (err) {
+			vp_del_vq(vqs[i]);
 			goto error_find;
+		}
 	}
 	return 0;
 

fs/coredump.c

···
 		if (t != current && !(t->flags & PF_POSTCOREDUMP)) {
 			sigaddset(&t->pending.signal, SIGKILL);
 			signal_wake_up(t, 1);
-			/* The vhost_worker does not particpate in coredumps */
-			if ((t->flags & (PF_USER_WORKER | PF_IO_WORKER)) != PF_USER_WORKER)
-				nr++;
+			nr++;
 		}
 	}
 

fs/fuse/virtio_fs.c

···
 
 static struct virtio_driver virtio_fs_driver = {
 	.driver.name		= KBUILD_MODNAME,
-	.driver.owner		= THIS_MODULE,
 	.id_table		= id_table,
 	.feature_table		= feature_table,
 	.feature_table_size	= ARRAY_SIZE(feature_table),

include/linux/sched/vhost_task.h

···
 
 struct vhost_task;
 
-struct vhost_task *vhost_task_create(bool (*fn)(void *), void *arg,
+struct vhost_task *vhost_task_create(bool (*fn)(void *),
+				     void (*handle_kill)(void *), void *arg,
 				     const char *name);
 void vhost_task_start(struct vhost_task *vtsk);
 void vhost_task_stop(struct vhost_task *vtsk);

include/uapi/linux/virtio_mem.h

···
 #define VIRTIO_MEM_F_ACPI_PXM		0
 /* unplugged memory must not be accessed */
 #define VIRTIO_MEM_F_UNPLUGGED_INACCESSIBLE	1
+/* plugged memory will remain plugged when suspending+resuming */
+#define VIRTIO_MEM_F_PERSISTENT_SUSPEND		2
 
 
 /* --- virtio-mem: guest -> host requests --- */

kernel/exit.c

···
 	tsk->flags |= PF_POSTCOREDUMP;
 	core_state = tsk->signal->core_state;
 	spin_unlock_irq(&tsk->sighand->siglock);
-
-	/* The vhost_worker does not particpate in coredumps */
-	if (core_state &&
-	    ((tsk->flags & (PF_IO_WORKER | PF_USER_WORKER)) != PF_USER_WORKER)) {
+	if (core_state) {
 		struct core_thread self;
 
 		self.task = current;

kernel/signal.c

···
 
 	for_other_threads(p, t) {
 		task_clear_jobctl_pending(t, JOBCTL_PENDING_MASK);
-		/* Don't require de_thread to wait for the vhost_worker */
-		if ((t->flags & (PF_IO_WORKER | PF_USER_WORKER)) != PF_USER_WORKER)
-			count++;
+		count++;
 
 		/* Don't bother with already dead threads */
 		if (t->exit_state)

kernel/vhost_task.c

···
 
 enum vhost_task_flags {
 	VHOST_TASK_FLAGS_STOP,
+	VHOST_TASK_FLAGS_KILLED,
 };
 
 struct vhost_task {
 	bool (*fn)(void *data);
+	void (*handle_sigkill)(void *data);
 	void *data;
 	struct completion exited;
 	unsigned long flags;
 	struct task_struct *task;
+	/* serialize SIGKILL and vhost_task_stop calls */
+	struct mutex exit_mutex;
 };
 
 static int vhost_task_fn(void *data)
 {
 	struct vhost_task *vtsk = data;
-	bool dead = false;
 
 	for (;;) {
 		bool did_work;
 
-		if (!dead && signal_pending(current)) {
+		if (signal_pending(current)) {
 			struct ksignal ksig;
-			/*
-			 * Calling get_signal will block in SIGSTOP,
-			 * or clear fatal_signal_pending, but remember
-			 * what was set.
-			 *
-			 * This thread won't actually exit until all
-			 * of the file descriptors are closed, and
-			 * the release function is called.
-			 */
-			dead = get_signal(&ksig);
-			if (dead)
-				clear_thread_flag(TIF_SIGPENDING);
+
+			if (get_signal(&ksig))
+				break;
 		}
 
 		/* mb paired w/ vhost_task_stop */
···
 			schedule();
 	}
 
+	mutex_lock(&vtsk->exit_mutex);
+	/*
+	 * If a vhost_task_stop and SIGKILL race, we can ignore the SIGKILL.
+	 * When the vhost layer has called vhost_task_stop it's already stopped
+	 * new work and flushed.
+	 */
+	if (!test_bit(VHOST_TASK_FLAGS_STOP, &vtsk->flags)) {
+		set_bit(VHOST_TASK_FLAGS_KILLED, &vtsk->flags);
+		vtsk->handle_sigkill(vtsk->data);
+	}
+	mutex_unlock(&vtsk->exit_mutex);
 	complete(&vtsk->exited);
+
 	do_exit(0);
 }
 
···
  * @vtsk: vhost_task to stop
  *
  * vhost_task_fn ensures the worker thread exits after
- * VHOST_TASK_FLAGS_SOP becomes true.
+ * VHOST_TASK_FLAGS_STOP becomes true.
  */
 void vhost_task_stop(struct vhost_task *vtsk)
 {
-	set_bit(VHOST_TASK_FLAGS_STOP, &vtsk->flags);
-	vhost_task_wake(vtsk);
+	mutex_lock(&vtsk->exit_mutex);
+	if (!test_bit(VHOST_TASK_FLAGS_KILLED, &vtsk->flags)) {
+		set_bit(VHOST_TASK_FLAGS_STOP, &vtsk->flags);
+		vhost_task_wake(vtsk);
+	}
+	mutex_unlock(&vtsk->exit_mutex);
+
 	/*
 	 * Make sure vhost_task_fn is no longer accessing the vhost_task before
 	 * freeing it below.
···
 /**
  * vhost_task_create - create a copy of a task to be used by the kernel
  * @fn: vhost worker function
- * @arg: data to be passed to fn
+ * @handle_sigkill: vhost function to handle when we are killed
+ * @arg: data to be passed to fn and handled_kill
  * @name: the thread's name
  *
  * This returns a specialized task for use by the vhost layer or NULL on
  * failure. The returned task is inactive, and the caller must fire it up
  * through vhost_task_start().
  */
-struct vhost_task *vhost_task_create(bool (*fn)(void *), void *arg,
+struct vhost_task *vhost_task_create(bool (*fn)(void *),
+				     void (*handle_sigkill)(void *), void *arg,
 				     const char *name)
 {
 	struct kernel_clone_args args = {
···
 	if (!vtsk)
 		return NULL;
 	init_completion(&vtsk->exited);
+	mutex_init(&vtsk->exit_mutex);
 	vtsk->data = arg;
 	vtsk->fn = fn;
+	vtsk->handle_sigkill = handle_sigkill;
 
 	args.fn_arg = vtsk;
 

net/9p/trans_virtio.c

···
 	.feature_table  = features,
 	.feature_table_size = ARRAY_SIZE(features),
 	.driver.name    = KBUILD_MODNAME,
-	.driver.owner	= THIS_MODULE,
 	.id_table	= id_table,
 	.probe		= p9_virtio_probe,
 	.remove		= p9_virtio_remove,

net/vmw_vsock/virtio_transport.c

···
 	.feature_table = features,
 	.feature_table_size = ARRAY_SIZE(features),
 	.driver.name = KBUILD_MODNAME,
-	.driver.owner = THIS_MODULE,
 	.id_table = id_table,
 	.probe = virtio_vsock_probe,
 	.remove = virtio_vsock_remove,

sound/virtio/virtio_card.c

···
 
 static struct virtio_driver virtsnd_driver = {
 	.driver.name = KBUILD_MODNAME,
-	.driver.owner = THIS_MODULE,
 	.id_table = id_table,
 	.feature_table = features,
 	.feature_table_size = ARRAY_SIZE(features),