Merge tag 'vfio-v6.2-rc1' of https://github.com/awilliam/linux-vfio

Pull VFIO updates from Alex Williamson:

- Replace deprecated git://github.com link in MAINTAINERS (Palmer
Dabbelt)

- Simplify vfio/mlx5 with module_pci_driver() helper (Shang XiaoJing)

- Drop unnecessary buffer from ACPI call (Rafael Mendonca)

- Correct latent missing include issue in iova-bitmap and fix support
for unaligned bitmaps. Follow-up with better fix through refactor
(Joao Martins)

- Rework ccw mdev driver to split private data from parent structure,
better aligning with the mdev lifecycle and allowing us to remove a
temporary workaround (Eric Farman)

- Add an interface to get an estimated migration data size for a
device, allowing userspace to make informed decisions, ex. more
accurately predicting VM downtime (Yishai Hadas)

- Fix minor typo in vfio/mlx5 array declaration (Yishai Hadas)

- Simplify module and Kconfig through consolidating SPAPR/EEH code and
config options and folding virqfd module into main vfio module (Jason
Gunthorpe)

- Fix error path from device_register() across all vfio mdev and sample
drivers (Alex Williamson)

- Define migration pre-copy interface and implement for vfio/mlx5
devices, allowing portions of the device state to be saved while the
device continues operation, towards reducing the stop-copy state size
(Jason Gunthorpe, Yishai Hadas, Shay Drory)

- Implement pre-copy for hisi_acc devices (Shameer Kolothum)

- Fixes to mdpy mdev driver remove path and error path on probe (Shang
XiaoJing)

- vfio/mlx5 fixes for incorrect return after copy_to_user() fault and
incorrect buffer freeing (Dan Carpenter)

* tag 'vfio-v6.2-rc1' of https://github.com/awilliam/linux-vfio: (42 commits)
vfio/mlx5: error pointer dereference in error handling
vfio/mlx5: fix error code in mlx5vf_precopy_ioctl()
samples: vfio-mdev: Fix missing pci_disable_device() in mdpy_fb_probe()
hisi_acc_vfio_pci: Enable PRE_COPY flag
hisi_acc_vfio_pci: Move the dev compatibility tests for early check
hisi_acc_vfio_pci: Introduce support for PRE_COPY state transitions
hisi_acc_vfio_pci: Add support for precopy IOCTL
vfio/mlx5: Enable MIGRATION_PRE_COPY flag
vfio/mlx5: Fallback to STOP_COPY upon specific PRE_COPY error
vfio/mlx5: Introduce multiple loads
vfio/mlx5: Consider temporary end of stream as part of PRE_COPY
vfio/mlx5: Introduce vfio precopy ioctl implementation
vfio/mlx5: Introduce SW headers for migration states
vfio/mlx5: Introduce device transitions of PRE_COPY
vfio/mlx5: Refactor to use queue based data chunks
vfio/mlx5: Refactor migration file state
vfio/mlx5: Refactor MKEY usage
vfio/mlx5: Refactor PD usage
vfio/mlx5: Enforce a single SAVE command at a time
vfio: Extend the device migration protocol with PRE_COPY
...

MAINTAINERS

···
 R:	Cornelia Huck <[e-mail]>
 L:	[e-mail]
 S:	Maintained
-T:	git git://github.com/awilliam/linux-vfio.git
+T:	git https://github.com/awilliam/linux-vfio.git
 F:	Documentation/ABI/testing/sysfs-devices-vfio-dev
 F:	Documentation/driver-api/vfio.rst
 F:	drivers/vfio/

drivers/gpu/drm/i915/gvt/kvmgt.c

···
 	struct intel_vgpu *vgpu = vfio_dev_to_vgpu(vfio_dev);
 
 	intel_gvt_destroy_vgpu(vgpu);
-	vfio_free_device(vfio_dev);
 }
 
 static const struct vfio_device_ops intel_vgpu_dev_ops = {

drivers/s390/cio/vfio_ccw_chp.c

···
 					  char __user *buf, size_t count,
 					  loff_t *ppos)
 {
+	struct subchannel *sch = to_subchannel(private->vdev.dev->parent);
 	unsigned int i = VFIO_CCW_OFFSET_TO_INDEX(*ppos) - VFIO_CCW_NUM_REGIONS;
 	loff_t pos = *ppos & VFIO_CCW_OFFSET_MASK;
 	struct ccw_schib_region *region;
···
 	mutex_lock(&private->io_mutex);
 	region = private->region[i].data;
 
-	if (cio_update_schib(private->sch)) {
+	if (cio_update_schib(sch)) {
 		ret = -ENODEV;
 		goto out;
 	}
 
-	memcpy(region, &private->sch->schib, sizeof(*region));
+	memcpy(region, &sch->schib, sizeof(*region));
 
 	if (copy_to_user(buf, (void *)region + pos, count)) {
 		ret = -EFAULT;

drivers/s390/cio/vfio_ccw_drv.c

···
 #include "vfio_ccw_private.h"
 
 struct workqueue_struct *vfio_ccw_work_q;
-static struct kmem_cache *vfio_ccw_io_region;
-static struct kmem_cache *vfio_ccw_cmd_region;
-static struct kmem_cache *vfio_ccw_schib_region;
-static struct kmem_cache *vfio_ccw_crw_region;
+struct kmem_cache *vfio_ccw_io_region;
+struct kmem_cache *vfio_ccw_cmd_region;
+struct kmem_cache *vfio_ccw_schib_region;
+struct kmem_cache *vfio_ccw_crw_region;
 
 debug_info_t *vfio_ccw_debug_msg_id;
 debug_info_t *vfio_ccw_debug_trace_id;
···
  */
 int vfio_ccw_sch_quiesce(struct subchannel *sch)
 {
-	struct vfio_ccw_private *private = dev_get_drvdata(&sch->dev);
+	struct vfio_ccw_parent *parent = dev_get_drvdata(&sch->dev);
+	struct vfio_ccw_private *private = dev_get_drvdata(&parent->dev);
 	DECLARE_COMPLETION_ONSTACK(completion);
 	int iretry, ret = 0;
+
+	/*
+	 * Probably an impossible situation, after being called through
+	 * FSM callbacks. But in the event it did, register a warning
+	 * and return as if things were fine.
+	 */
+	if (WARN_ON(!private))
+		return 0;
 
 	iretry = 255;
 	do {
···
 	return ret;
 }
 
-static void vfio_ccw_sch_io_todo(struct work_struct *work)
+void vfio_ccw_sch_io_todo(struct work_struct *work)
 {
 	struct vfio_ccw_private *private;
 	struct irb *irb;
···
 		eventfd_signal(private->io_trigger, 1);
 }
 
-static void vfio_ccw_crw_todo(struct work_struct *work)
+void vfio_ccw_crw_todo(struct work_struct *work)
 {
 	struct vfio_ccw_private *private;
 
···
  */
 static void vfio_ccw_sch_irq(struct subchannel *sch)
 {
-	struct vfio_ccw_private *private = dev_get_drvdata(&sch->dev);
+	struct vfio_ccw_parent *parent = dev_get_drvdata(&sch->dev);
+	struct vfio_ccw_private *private = dev_get_drvdata(&parent->dev);
+
+	/*
+	 * The subchannel should still be disabled at this point,
+	 * so an interrupt would be quite surprising. As with an
+	 * interrupt while the FSM is closed, let's attempt to
+	 * disable the subchannel again.
+	 */
+	if (!private) {
+		VFIO_CCW_MSG_EVENT(2, "sch %x.%x.%04x: unexpected interrupt\n",
+				   sch->schid.cssid, sch->schid.ssid,
+				   sch->schid.sch_no);
+
+		cio_disable_subchannel(sch);
+		return;
+	}
 
 	inc_irq_stat(IRQIO_CIO);
 	vfio_ccw_fsm_event(private, VFIO_CCW_EVENT_INTERRUPT);
 }
 
-static struct vfio_ccw_private *vfio_ccw_alloc_private(struct subchannel *sch)
+static void vfio_ccw_free_parent(struct device *dev)
 {
-	struct vfio_ccw_private *private;
+	struct vfio_ccw_parent *parent = container_of(dev, struct vfio_ccw_parent, dev);
 
-	private = kzalloc(sizeof(*private), GFP_KERNEL);
-	if (!private)
-		return ERR_PTR(-ENOMEM);
-
-	private->sch = sch;
-	mutex_init(&private->io_mutex);
-	private->state = VFIO_CCW_STATE_STANDBY;
-	INIT_LIST_HEAD(&private->crw);
-	INIT_WORK(&private->io_work, vfio_ccw_sch_io_todo);
-	INIT_WORK(&private->crw_work, vfio_ccw_crw_todo);
-
-	private->cp.guest_cp = kcalloc(CCWCHAIN_LEN_MAX, sizeof(struct ccw1),
-				       GFP_KERNEL);
-	if (!private->cp.guest_cp)
-		goto out_free_private;
-
-	private->io_region = kmem_cache_zalloc(vfio_ccw_io_region,
-					       GFP_KERNEL | GFP_DMA);
-	if (!private->io_region)
-		goto out_free_cp;
-
-	private->cmd_region = kmem_cache_zalloc(vfio_ccw_cmd_region,
-						GFP_KERNEL | GFP_DMA);
-	if (!private->cmd_region)
-		goto out_free_io;
-
-	private->schib_region = kmem_cache_zalloc(vfio_ccw_schib_region,
-						  GFP_KERNEL | GFP_DMA);
-
-	if (!private->schib_region)
-		goto out_free_cmd;
-
-	private->crw_region = kmem_cache_zalloc(vfio_ccw_crw_region,
-						GFP_KERNEL | GFP_DMA);
-
-	if (!private->crw_region)
-		goto out_free_schib;
-	return private;
-
-out_free_schib:
-	kmem_cache_free(vfio_ccw_schib_region, private->schib_region);
-out_free_cmd:
-	kmem_cache_free(vfio_ccw_cmd_region, private->cmd_region);
-out_free_io:
-	kmem_cache_free(vfio_ccw_io_region, private->io_region);
-out_free_cp:
-	kfree(private->cp.guest_cp);
-out_free_private:
-	mutex_destroy(&private->io_mutex);
-	kfree(private);
-	return ERR_PTR(-ENOMEM);
+	kfree(parent);
 }
 
-static void vfio_ccw_free_private(struct vfio_ccw_private *private)
-{
-	struct vfio_ccw_crw *crw, *temp;
-
-	list_for_each_entry_safe(crw, temp, &private->crw, next) {
-		list_del(&crw->next);
-		kfree(crw);
-	}
-
-	kmem_cache_free(vfio_ccw_crw_region, private->crw_region);
-	kmem_cache_free(vfio_ccw_schib_region, private->schib_region);
-	kmem_cache_free(vfio_ccw_cmd_region, private->cmd_region);
-	kmem_cache_free(vfio_ccw_io_region, private->io_region);
-	kfree(private->cp.guest_cp);
-	mutex_destroy(&private->io_mutex);
-	kfree(private);
-}
 static int vfio_ccw_sch_probe(struct subchannel *sch)
 {
 	struct pmcw *pmcw = &sch->schib.pmcw;
-	struct vfio_ccw_private *private;
+	struct vfio_ccw_parent *parent;
 	int ret = -ENOMEM;
 
 	if (pmcw->qf) {
···
 		return -ENODEV;
 	}
 
-	private = vfio_ccw_alloc_private(sch);
-	if (IS_ERR(private))
-		return PTR_ERR(private);
+	parent = kzalloc(sizeof(*parent), GFP_KERNEL);
+	if (!parent)
+		return -ENOMEM;
 
-	dev_set_drvdata(&sch->dev, private);
-
-	private->mdev_type.sysfs_name = "io";
-	private->mdev_type.pretty_name = "I/O subchannel (Non-QDIO)";
-	private->mdev_types[0] = &private->mdev_type;
-	ret = mdev_register_parent(&private->parent, &sch->dev,
-				   &vfio_ccw_mdev_driver,
-				   private->mdev_types, 1);
+	dev_set_name(&parent->dev, "parent");
+	parent->dev.parent = &sch->dev;
+	parent->dev.release = &vfio_ccw_free_parent;
+	ret = device_register(&parent->dev);
 	if (ret)
 		goto out_free;
+
+	dev_set_drvdata(&sch->dev, parent);
+
+	parent->mdev_type.sysfs_name = "io";
+	parent->mdev_type.pretty_name = "I/O subchannel (Non-QDIO)";
+	parent->mdev_types[0] = &parent->mdev_type;
+	ret = mdev_register_parent(&parent->parent, &sch->dev,
+				   &vfio_ccw_mdev_driver,
+				   parent->mdev_types, 1);
+	if (ret)
+		goto out_unreg;
 
 	VFIO_CCW_MSG_EVENT(4, "bound to subchannel %x.%x.%04x\n",
 			   sch->schid.cssid, sch->schid.ssid,
 			   sch->schid.sch_no);
 	return 0;
 
+out_unreg:
+	device_del(&parent->dev);
 out_free:
+	put_device(&parent->dev);
 	dev_set_drvdata(&sch->dev, NULL);
-	vfio_ccw_free_private(private);
 	return ret;
 }
 
 static void vfio_ccw_sch_remove(struct subchannel *sch)
 {
-	struct vfio_ccw_private *private = dev_get_drvdata(&sch->dev);
+	struct vfio_ccw_parent *parent = dev_get_drvdata(&sch->dev);
 
-	mdev_unregister_parent(&private->parent);
+	mdev_unregister_parent(&parent->parent);
 
+	device_unregister(&parent->dev);
 	dev_set_drvdata(&sch->dev, NULL);
-
-	vfio_ccw_free_private(private);
 
 	VFIO_CCW_MSG_EVENT(4, "unbound from subchannel %x.%x.%04x\n",
 			   sch->schid.cssid, sch->schid.ssid,
···
 
 static void vfio_ccw_sch_shutdown(struct subchannel *sch)
 {
-	struct vfio_ccw_private *private = dev_get_drvdata(&sch->dev);
+	struct vfio_ccw_parent *parent = dev_get_drvdata(&sch->dev);
+	struct vfio_ccw_private *private = dev_get_drvdata(&parent->dev);
+
+	if (WARN_ON(!private))
+		return;
 
 	vfio_ccw_fsm_event(private, VFIO_CCW_EVENT_CLOSE);
 	vfio_ccw_fsm_event(private, VFIO_CCW_EVENT_NOT_OPER);
···
  */
 static int vfio_ccw_sch_event(struct subchannel *sch, int process)
 {
-	struct vfio_ccw_private *private = dev_get_drvdata(&sch->dev);
+	struct vfio_ccw_parent *parent = dev_get_drvdata(&sch->dev);
+	struct vfio_ccw_private *private = dev_get_drvdata(&parent->dev);
 	unsigned long flags;
 	int rc = -EAGAIN;
 
···
 
 	rc = 0;
 
-	if (cio_update_schib(sch))
-		vfio_ccw_fsm_event(private, VFIO_CCW_EVENT_NOT_OPER);
+	if (cio_update_schib(sch)) {
+		if (private)
+			vfio_ccw_fsm_event(private, VFIO_CCW_EVENT_NOT_OPER);
+	}
 
 out_unlock:
 	spin_unlock_irqrestore(sch->lock, flags);
···
 static int vfio_ccw_chp_event(struct subchannel *sch,
 			      struct chp_link *link, int event)
 {
-	struct vfio_ccw_private *private = dev_get_drvdata(&sch->dev);
+	struct vfio_ccw_parent *parent = dev_get_drvdata(&sch->dev);
+	struct vfio_ccw_private *private = dev_get_drvdata(&parent->dev);
 	int mask = chp_ssd_get_mask(&sch->ssd_info, link);
 	int retry = 255;
 
 	if (!private || !mask)
 		return 0;
 
-	trace_vfio_ccw_chp_event(private->sch->schid, mask, event);
+	trace_vfio_ccw_chp_event(sch->schid, mask, event);
 	VFIO_CCW_MSG_EVENT(2, "sch %x.%x.%04x: mask=0x%x event=%d\n",
 			   sch->schid.cssid,
 			   sch->schid.ssid, sch->schid.sch_no,

drivers/s390/cio/vfio_ccw_fsm.c

···
 
 static int fsm_io_helper(struct vfio_ccw_private *private)
 {
-	struct subchannel *sch;
+	struct subchannel *sch = to_subchannel(private->vdev.dev->parent);
 	union orb *orb;
 	int ccode;
 	__u8 lpm;
 	unsigned long flags;
 	int ret;
-
-	sch = private->sch;
 
 	spin_lock_irqsave(sch->lock, flags);
 
···
 
 static int fsm_do_halt(struct vfio_ccw_private *private)
 {
-	struct subchannel *sch;
+	struct subchannel *sch = to_subchannel(private->vdev.dev->parent);
 	unsigned long flags;
 	int ccode;
 	int ret;
-
-	sch = private->sch;
 
 	spin_lock_irqsave(sch->lock, flags);
 
···
 
 static int fsm_do_clear(struct vfio_ccw_private *private)
 {
-	struct subchannel *sch;
+	struct subchannel *sch = to_subchannel(private->vdev.dev->parent);
 	unsigned long flags;
 	int ccode;
 	int ret;
-
-	sch = private->sch;
 
 	spin_lock_irqsave(sch->lock, flags);
 
···
 static void fsm_notoper(struct vfio_ccw_private *private,
 			enum vfio_ccw_event event)
 {
-	struct subchannel *sch = private->sch;
+	struct subchannel *sch = to_subchannel(private->vdev.dev->parent);
 
 	VFIO_CCW_MSG_EVENT(2, "sch %x.%x.%04x: notoper event %x state %x\n",
 			   sch->schid.cssid,
···
 static void fsm_disabled_irq(struct vfio_ccw_private *private,
 			     enum vfio_ccw_event event)
 {
-	struct subchannel *sch = private->sch;
+	struct subchannel *sch = to_subchannel(private->vdev.dev->parent);
 
 	/*
 	 * An interrupt in a disabled state means a previous disable was not
···
 }
 inline struct subchannel_id get_schid(struct vfio_ccw_private *p)
 {
-	return p->sch->schid;
+	struct subchannel *sch = to_subchannel(p->vdev.dev->parent);
+
+	return sch->schid;
 }
 
 /*
···
 static void fsm_irq(struct vfio_ccw_private *private,
 		    enum vfio_ccw_event event)
 {
+	struct subchannel *sch = to_subchannel(private->vdev.dev->parent);
 	struct irb *irb = this_cpu_ptr(&cio_irb);
 
 	VFIO_CCW_TRACE_EVENT(6, "IRQ");
-	VFIO_CCW_TRACE_EVENT(6, dev_name(&private->sch->dev));
+	VFIO_CCW_TRACE_EVENT(6, dev_name(&sch->dev));
 
 	memcpy(&private->irb, irb, sizeof(*irb));
 
···
 static void fsm_open(struct vfio_ccw_private *private,
 		     enum vfio_ccw_event event)
 {
-	struct subchannel *sch = private->sch;
+	struct subchannel *sch = to_subchannel(private->vdev.dev->parent);
 	int ret;
 
 	spin_lock_irq(sch->lock);
···
 static void fsm_close(struct vfio_ccw_private *private,
 		      enum vfio_ccw_event event)
 {
-	struct subchannel *sch = private->sch;
+	struct subchannel *sch = to_subchannel(private->vdev.dev->parent);
 	int ret;
 
 	spin_lock_irq(sch->lock);

drivers/s390/cio/vfio_ccw_ops.c

···
 	struct vfio_ccw_private *private =
 		container_of(vdev, struct vfio_ccw_private, vdev);
 
-	init_completion(&private->release_comp);
+	mutex_init(&private->io_mutex);
+	private->state = VFIO_CCW_STATE_STANDBY;
+	INIT_LIST_HEAD(&private->crw);
+	INIT_WORK(&private->io_work, vfio_ccw_sch_io_todo);
+	INIT_WORK(&private->crw_work, vfio_ccw_crw_todo);
+
+	private->cp.guest_cp = kcalloc(CCWCHAIN_LEN_MAX, sizeof(struct ccw1),
+				       GFP_KERNEL);
+	if (!private->cp.guest_cp)
+		goto out_free_private;
+
+	private->io_region = kmem_cache_zalloc(vfio_ccw_io_region,
+					       GFP_KERNEL | GFP_DMA);
+	if (!private->io_region)
+		goto out_free_cp;
+
+	private->cmd_region = kmem_cache_zalloc(vfio_ccw_cmd_region,
+						GFP_KERNEL | GFP_DMA);
+	if (!private->cmd_region)
+		goto out_free_io;
+
+	private->schib_region = kmem_cache_zalloc(vfio_ccw_schib_region,
+						  GFP_KERNEL | GFP_DMA);
+	if (!private->schib_region)
+		goto out_free_cmd;
+
+	private->crw_region = kmem_cache_zalloc(vfio_ccw_crw_region,
+						GFP_KERNEL | GFP_DMA);
+	if (!private->crw_region)
+		goto out_free_schib;
+
 	return 0;
+
+out_free_schib:
+	kmem_cache_free(vfio_ccw_schib_region, private->schib_region);
+out_free_cmd:
+	kmem_cache_free(vfio_ccw_cmd_region, private->cmd_region);
+out_free_io:
+	kmem_cache_free(vfio_ccw_io_region, private->io_region);
+out_free_cp:
+	kfree(private->cp.guest_cp);
+out_free_private:
+	mutex_destroy(&private->io_mutex);
+	return -ENOMEM;
 }
 
 static int vfio_ccw_mdev_probe(struct mdev_device *mdev)
 {
-	struct vfio_ccw_private *private = dev_get_drvdata(mdev->dev.parent);
+	struct subchannel *sch = to_subchannel(mdev->dev.parent);
+	struct vfio_ccw_parent *parent = dev_get_drvdata(&sch->dev);
+	struct vfio_ccw_private *private;
 	int ret;
 
-	if (private->state == VFIO_CCW_STATE_NOT_OPER)
-		return -ENODEV;
+	private = vfio_alloc_device(vfio_ccw_private, vdev, &mdev->dev,
+				    &vfio_ccw_dev_ops);
+	if (IS_ERR(private))
+		return PTR_ERR(private);
 
-	ret = vfio_init_device(&private->vdev, &mdev->dev, &vfio_ccw_dev_ops);
-	if (ret)
-		return ret;
+	dev_set_drvdata(&parent->dev, private);
 
 	VFIO_CCW_MSG_EVENT(2, "sch %x.%x.%04x: create\n",
-			   private->sch->schid.cssid,
-			   private->sch->schid.ssid,
-			   private->sch->schid.sch_no);
+			   sch->schid.cssid,
+			   sch->schid.ssid,
+			   sch->schid.sch_no);
 
 	ret = vfio_register_emulated_iommu_dev(&private->vdev);
 	if (ret)
···
 	return 0;
 
 err_put_vdev:
+	dev_set_drvdata(&parent->dev, NULL);
 	vfio_put_device(&private->vdev);
 	return ret;
 }
···
 {
 	struct vfio_ccw_private *private =
 		container_of(vdev, struct vfio_ccw_private, vdev);
+	struct vfio_ccw_crw *crw, *temp;
 
-	/*
-	 * We cannot free vfio_ccw_private here because it includes
-	 * parent info which must be free'ed by css driver.
-	 *
-	 * Use a workaround by memset'ing the core device part and
-	 * then notifying the remove path that all active references
-	 * to this device have been released.
-	 */
-	memset(vdev, 0, sizeof(*vdev));
-	complete(&private->release_comp);
+	list_for_each_entry_safe(crw, temp, &private->crw, next) {
+		list_del(&crw->next);
+		kfree(crw);
+	}
+
+	kmem_cache_free(vfio_ccw_crw_region, private->crw_region);
+	kmem_cache_free(vfio_ccw_schib_region, private->schib_region);
+	kmem_cache_free(vfio_ccw_cmd_region, private->cmd_region);
+	kmem_cache_free(vfio_ccw_io_region, private->io_region);
+	kfree(private->cp.guest_cp);
+	mutex_destroy(&private->io_mutex);
 }
 
 static void vfio_ccw_mdev_remove(struct mdev_device *mdev)
 {
-	struct vfio_ccw_private *private = dev_get_drvdata(mdev->dev.parent);
+	struct subchannel *sch = to_subchannel(mdev->dev.parent);
+	struct vfio_ccw_parent *parent = dev_get_drvdata(&sch->dev);
+	struct vfio_ccw_private *private = dev_get_drvdata(&parent->dev);
 
 	VFIO_CCW_MSG_EVENT(2, "sch %x.%x.%04x: remove\n",
-			   private->sch->schid.cssid,
-			   private->sch->schid.ssid,
-			   private->sch->schid.sch_no);
+			   sch->schid.cssid,
+			   sch->schid.ssid,
+			   sch->schid.sch_no);
 
 	vfio_unregister_group_dev(&private->vdev);
 
+	dev_set_drvdata(&parent->dev, NULL);
 	vfio_put_device(&private->vdev);
-	/*
-	 * Wait for all active references on mdev are released so it
-	 * is safe to defer kfree() to a later point.
-	 *
-	 * TODO: the clean fix is to split parent/mdev info from ccw
-	 * private structure so each can be managed in its own life
-	 * cycle.
-	 */
-	wait_for_completion(&private->release_comp);
 }
 
 static int vfio_ccw_mdev_open_device(struct vfio_device *vdev)

drivers/s390/cio/vfio_ccw_private.h

···
 };
 
 /**
+ * struct vfio_ccw_parent
+ *
+ * @dev: embedded device struct
+ * @parent: parent data structures for mdevs created
+ * @mdev_type(s): identifying information for mdevs created
+ */
+struct vfio_ccw_parent {
+	struct device		dev;
+
+	struct mdev_parent	parent;
+	struct mdev_type	mdev_type;
+	struct mdev_type	*mdev_types[1];
+};
+
+/**
  * struct vfio_ccw_private
  * @vdev: Embedded VFIO device
- * @sch: pointer to the subchannel
  * @state: internal state of the device
  * @completion: synchronization helper of the I/O completion
  * @io_region: MMIO region to input/output I/O arguments/results
···
  * @req_trigger: eventfd ctx for signaling userspace to return device
  * @io_work: work for deferral process of I/O handling
  * @crw_work: work for deferral process of CRW handling
- * @release_comp: synchronization helper for vfio device release
- * @parent: parent data structures for mdevs created
  */
 struct vfio_ccw_private {
 	struct vfio_device vdev;
-	struct subchannel	*sch;
 	int			state;
 	struct completion	*completion;
 	struct ccw_io_region	*io_region;
···
 	struct eventfd_ctx	*req_trigger;
 	struct work_struct	io_work;
 	struct work_struct	crw_work;
-
-	struct completion	release_comp;
-
-	struct mdev_parent	parent;
-	struct mdev_type	mdev_type;
-	struct mdev_type	*mdev_types[1];
 } __aligned(8);
 
 int vfio_ccw_sch_quiesce(struct subchannel *sch);
+void vfio_ccw_sch_io_todo(struct work_struct *work);
+void vfio_ccw_crw_todo(struct work_struct *work);
 
 extern struct mdev_driver vfio_ccw_mdev_driver;
 
···
 static inline void vfio_ccw_fsm_event(struct vfio_ccw_private *private,
 				      enum vfio_ccw_event event)
 {
-	trace_vfio_ccw_fsm_event(private->sch->schid, private->state, event);
+	struct subchannel *sch = to_subchannel(private->vdev.dev->parent);
+
+	if (sch)
+		trace_vfio_ccw_fsm_event(sch->schid, private->state, event);
 	vfio_ccw_jumptable[private->state][event](private, event);
 }
 
 extern struct workqueue_struct *vfio_ccw_work_q;
-
+extern struct kmem_cache *vfio_ccw_io_region;
+extern struct kmem_cache *vfio_ccw_cmd_region;
+extern struct kmem_cache *vfio_ccw_schib_region;
+extern struct kmem_cache *vfio_ccw_crw_region;
 
 /* s390 debug feature, similar to base cio */
 extern debug_info_t *vfio_ccw_debug_msg_id;

drivers/s390/crypto/vfio_ap_drv.c

···
 	return 0;
 
 matrix_drv_err:
-	device_unregister(&matrix_dev->device);
+	device_del(&matrix_dev->device);
 matrix_reg_err:
 	put_device(&matrix_dev->device);
 matrix_alloc_err:

drivers/s390/crypto/vfio_ap_ops.c

···
 	}
 }
 
-static void vfio_ap_mdev_release_dev(struct vfio_device *vdev)
-{
-	vfio_free_device(vdev);
-}
-
 static void vfio_ap_mdev_remove(struct mdev_device *mdev)
 {
 	struct ap_matrix_mdev *matrix_mdev = dev_get_drvdata(&mdev->dev);
···
 
 static const struct vfio_device_ops vfio_ap_matrix_dev_ops = {
 	.init = vfio_ap_mdev_init_dev,
-	.release = vfio_ap_mdev_release_dev,
 	.open_device = vfio_ap_mdev_open_device,
 	.close_device = vfio_ap_mdev_close_device,
 	.ioctl = vfio_ap_mdev_ioctl,

drivers/vfio/Kconfig

···
 	  If you don't know what to do here, say N.
 endif
 
-config VFIO_SPAPR_EEH
-	tristate
-	depends on EEH && VFIO_IOMMU_SPAPR_TCE
-	default VFIO
-
 config VFIO_VIRQFD
-	tristate
+	bool
 	select EVENTFD
 	default n
 

drivers/vfio/Makefile

···
 # SPDX-License-Identifier: GPL-2.0
-vfio_virqfd-y := virqfd.o
-
 obj-$(CONFIG_VFIO) += vfio.o
 
 vfio-y += vfio_main.o \
···
 	  iova_bitmap.o
 vfio-$(CONFIG_IOMMUFD) += iommufd.o
 vfio-$(CONFIG_VFIO_CONTAINER) += container.o
+vfio-$(CONFIG_VFIO_VIRQFD) += virqfd.o
 
-obj-$(CONFIG_VFIO_VIRQFD) += vfio_virqfd.o
 obj-$(CONFIG_VFIO_IOMMU_TYPE1) += vfio_iommu_type1.o
 obj-$(CONFIG_VFIO_IOMMU_SPAPR_TCE) += vfio_iommu_spapr_tce.o
-obj-$(CONFIG_VFIO_SPAPR_EEH) += vfio_spapr_eeh.o
 obj-$(CONFIG_VFIO_PCI) += pci/
 obj-$(CONFIG_VFIO_PLATFORM) += platform/
 obj-$(CONFIG_VFIO_MDEV) += mdev/

drivers/vfio/fsl-mc/vfio_fsl_mc.c

···
 
 	vfio_fsl_uninit_device(vdev);
 	mutex_destroy(&vdev->igate);
-	vfio_free_device(core_vdev);
 }
 
 static int vfio_fsl_mc_remove(struct fsl_mc_device *mc_dev)

drivers/vfio/iova_bitmap.c

···
  */
 #include <linux/iova_bitmap.h>
 #include <linux/mm.h>
+#include <linux/slab.h>
 #include <linux/highmem.h>
 
 #define BITS_PER_PAGE (PAGE_SIZE * BITS_PER_BYTE)
···
  */
 static unsigned long iova_bitmap_mapped_remaining(struct iova_bitmap *bitmap)
 {
-	unsigned long remaining;
+	unsigned long remaining, bytes;
+
+	bytes = (bitmap->mapped.npages << PAGE_SHIFT) - bitmap->mapped.pgoff;
 
 	remaining = bitmap->mapped_total_index - bitmap->mapped_base_index;
 	remaining = min_t(unsigned long, remaining,
-	      (bitmap->mapped.npages << PAGE_SHIFT) / sizeof(*bitmap->bitmap));
+			  bytes / sizeof(*bitmap->bitmap));
 
 	return remaining;
 }
···
  * Set the bits corresponding to the range [iova .. iova+length-1] in
  * the user bitmap.
  *
- * Return: The number of bits set.
  */
 void iova_bitmap_set(struct iova_bitmap *bitmap,
 		     unsigned long iova, size_t length)
 {
 	struct iova_bitmap_map *mapped = &bitmap->mapped;
-	unsigned long offset = (iova - mapped->iova) >> mapped->pgshift;
-	unsigned long nbits = max_t(unsigned long, 1, length >> mapped->pgshift);
-	unsigned long page_idx = offset / BITS_PER_PAGE;
-	unsigned long page_offset = mapped->pgoff;
-	void *kaddr;
-
-	offset = offset % BITS_PER_PAGE;
+	unsigned long cur_bit = ((iova - mapped->iova) >>
+			mapped->pgshift) + mapped->pgoff * BITS_PER_BYTE;
+	unsigned long last_bit = (((iova + length - 1) - mapped->iova) >>
+			mapped->pgshift) + mapped->pgoff * BITS_PER_BYTE;
 
 	do {
-		unsigned long size = min(BITS_PER_PAGE - offset, nbits);
+		unsigned int page_idx = cur_bit / BITS_PER_PAGE;
+		unsigned int offset = cur_bit % BITS_PER_PAGE;
+		unsigned int nbits = min(BITS_PER_PAGE - offset,
+					 last_bit - cur_bit + 1);
+		void *kaddr;
 
 		kaddr = kmap_local_page(mapped->pages[page_idx]);
-		bitmap_set(kaddr + page_offset, offset, size);
+		bitmap_set(kaddr, offset, nbits);
 		kunmap_local(kaddr);
-		page_offset = offset = 0;
-		nbits -= size;
-		page_idx++;
-	} while (nbits > 0);
+		cur_bit += nbits;
+	} while (cur_bit <= last_bit);
 }
 EXPORT_SYMBOL_GPL(iova_bitmap_set);

drivers/vfio/pci/hisilicon/hisi_acc_vfio_pci.c

···
 	u32 que_iso_state;
 	int ret;
 
-	if (migf->total_length < QM_MATCH_SIZE)
-		return -EINVAL;
+	if (migf->total_length < QM_MATCH_SIZE || hisi_acc_vdev->match_done)
+		return 0;
 
 	if (vf_data->acc_magic != ACC_DEV_MAGIC) {
 		dev_err(dev, "failed to match ACC_DEV_MAGIC\n");
···
 	}
 
 	hisi_acc_vdev->vf_qm_state = vf_data->vf_qm_state;
+	hisi_acc_vdev->match_done = true;
 	return 0;
 }
 
···
 	struct hisi_qm *vf_qm = &hisi_acc_vdev->vf_qm;
 	struct device *dev = &vf_qm->pdev->dev;
 	int ret;
-
-	ret = vf_qm_get_match_data(hisi_acc_vdev, vf_data);
-	if (ret)
-		return ret;
 
 	if (unlikely(qm_wait_dev_not_ready(vf_qm))) {
 		/* Update state and return with match data */
···
 	struct hisi_acc_vf_migration_file *migf = hisi_acc_vdev->resuming_migf;
 	int ret;
 
-	/* Check dev compatibility */
-	ret = vf_qm_check_match(hisi_acc_vdev, migf);
-	if (ret) {
-		dev_err(dev, "failed to match the VF!\n");
-		return ret;
-	}
 	/* Recover data to VF */
 	ret = vf_qm_load_data(hisi_acc_vdev, migf);
 	if (ret) {
···
 	*pos += len;
 	done = len;
 	migf->total_length += len;
+
+	ret = vf_qm_check_match(migf->hisi_acc_vdev, migf);
+	if (ret)
+		done = -EFAULT;
 out_unlock:
 	mutex_unlock(&migf->lock);
 	return done;
···
 
 	stream_open(migf->filp->f_inode, migf->filp);
 	mutex_init(&migf->lock);
+	migf->hisi_acc_vdev = hisi_acc_vdev;
 	return migf;
+}
+
+static long hisi_acc_vf_precopy_ioctl(struct file *filp,
+				      unsigned int cmd, unsigned long arg)
+{
+	struct hisi_acc_vf_migration_file *migf = filp->private_data;
+	struct hisi_acc_vf_core_device *hisi_acc_vdev = migf->hisi_acc_vdev;
+	loff_t *pos = &filp->f_pos;
+	struct vfio_precopy_info info;
+	unsigned long minsz;
+	int ret;
+
+	if (cmd != VFIO_MIG_GET_PRECOPY_INFO)
+		return -ENOTTY;
+
+	minsz = offsetofend(struct vfio_precopy_info, dirty_bytes);
+
+	if (copy_from_user(&info, (void __user *)arg, minsz))
+		return -EFAULT;
+	if (info.argsz < minsz)
+		return -EINVAL;
+
+	mutex_lock(&hisi_acc_vdev->state_mutex);
+	if (hisi_acc_vdev->mig_state != VFIO_DEVICE_STATE_PRE_COPY) {
+		mutex_unlock(&hisi_acc_vdev->state_mutex);
+		return -EINVAL;
+	}
+
+	mutex_lock(&migf->lock);
+
+	if (migf->disabled) {
+		ret = -ENODEV;
+		goto out;
+	}
+
+	if (*pos > migf->total_length) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	info.dirty_bytes = 0;
+	info.initial_bytes = migf->total_length - *pos;
+
+	ret = copy_to_user((void __user *)arg, &info, minsz) ? -EFAULT : 0;
+out:
+	mutex_unlock(&migf->lock);
+	mutex_unlock(&hisi_acc_vdev->state_mutex);
+	return ret;
 }
 
 static ssize_t hisi_acc_vf_save_read(struct file *filp, char __user *buf, size_t len,
···
 static const struct file_operations hisi_acc_vf_save_fops = {
 	.owner = THIS_MODULE,
 	.read = hisi_acc_vf_save_read,
+	.unlocked_ioctl = hisi_acc_vf_precopy_ioctl,
+	.compat_ioctl = compat_ptr_ioctl,
 	.release = hisi_acc_vf_release_file,
 	.llseek = no_llseek,
 };
 
 static struct hisi_acc_vf_migration_file *
-hisi_acc_vf_stop_copy(struct hisi_acc_vf_core_device *hisi_acc_vdev)
+hisi_acc_open_saving_migf(struct hisi_acc_vf_core_device *hisi_acc_vdev)
 {
 	struct hisi_acc_vf_migration_file *migf;
 	int ret;
···
 
 	stream_open(migf->filp->f_inode, migf->filp);
 	mutex_init(&migf->lock);
+	migf->hisi_acc_vdev = hisi_acc_vdev;
 
-	ret = vf_qm_state_save(hisi_acc_vdev, migf);
+	ret = vf_qm_get_match_data(hisi_acc_vdev, &migf->vf_data);
 	if (ret) {
 		fput(migf->filp);
 		return ERR_PTR(ret);
 	}
 
 	return migf;
+}
+
+static struct hisi_acc_vf_migration_file *
+hisi_acc_vf_pre_copy(struct hisi_acc_vf_core_device *hisi_acc_vdev)
+{
+	struct hisi_acc_vf_migration_file *migf;
+
+	migf = hisi_acc_open_saving_migf(hisi_acc_vdev);
+	if (IS_ERR(migf))
+		return migf;
+
+	migf->total_length = QM_MATCH_SIZE;
+	return migf;
+}
+
+static struct hisi_acc_vf_migration_file *
+hisi_acc_vf_stop_copy(struct hisi_acc_vf_core_device *hisi_acc_vdev, bool open)
+{
+	int ret;
+	struct hisi_acc_vf_migration_file *migf = NULL;
+
+	if (open) {
+		/*
+		 * Userspace didn't use PRECOPY support. Hence saving_migf
+		 * is not opened yet.
+		 */
+		migf = hisi_acc_open_saving_migf(hisi_acc_vdev);
+		if (IS_ERR(migf))
+			return migf;
+	} else {
+		migf = hisi_acc_vdev->saving_migf;
+	}
+
+	ret = vf_qm_state_save(hisi_acc_vdev, migf);
+	if (ret)
+		return ERR_PTR(ret);
+
+	return open ? migf : NULL;
 }
 
 static int hisi_acc_vf_stop_device(struct hisi_acc_vf_core_device *hisi_acc_vdev)
···
 	u32 cur = hisi_acc_vdev->mig_state;
 	int ret;
 
+	if (cur == VFIO_DEVICE_STATE_RUNNING && new == VFIO_DEVICE_STATE_PRE_COPY) {
+		struct hisi_acc_vf_migration_file *migf;
+
+		migf = hisi_acc_vf_pre_copy(hisi_acc_vdev);
+		if (IS_ERR(migf))
+			return ERR_CAST(migf);
+		get_file(migf->filp);
+		hisi_acc_vdev->saving_migf = migf;
+		return migf->filp;
+	}
+
+	if (cur == VFIO_DEVICE_STATE_PRE_COPY && new == VFIO_DEVICE_STATE_STOP_COPY) {
+		struct hisi_acc_vf_migration_file *migf;
+
+		ret = hisi_acc_vf_stop_device(hisi_acc_vdev);
+		if (ret)
+			return ERR_PTR(ret);
+
+		migf = hisi_acc_vf_stop_copy(hisi_acc_vdev, false);
+		if (IS_ERR(migf))
+			return ERR_CAST(migf);
+
+		return NULL;
+	}
+
 	if (cur == VFIO_DEVICE_STATE_RUNNING && new == VFIO_DEVICE_STATE_STOP) {
 		ret = hisi_acc_vf_stop_device(hisi_acc_vdev);
 		if (ret)
···
 	if (cur == VFIO_DEVICE_STATE_STOP && new == VFIO_DEVICE_STATE_STOP_COPY) {
 		struct hisi_acc_vf_migration_file *migf;
 
-		migf = hisi_acc_vf_stop_copy(hisi_acc_vdev);
+		migf = hisi_acc_vf_stop_copy(hisi_acc_vdev, true);
 		if (IS_ERR(migf))
 			return ERR_CAST(migf);
 		get_file(migf->filp);
···
 		ret = hisi_acc_vf_load_state(hisi_acc_vdev);
 		if (ret)
 			return ERR_PTR(ret);
+		hisi_acc_vf_disable_fds(hisi_acc_vdev);
+		return NULL;
+	}
+
+	if (cur == VFIO_DEVICE_STATE_PRE_COPY && new == VFIO_DEVICE_STATE_RUNNING) {
 		hisi_acc_vf_disable_fds(hisi_acc_vdev);
 		return NULL;
 	}
···
 	}
 	hisi_acc_vf_state_mutex_unlock(hisi_acc_vdev);
 	return res;
+}
+
+static int
+hisi_acc_vfio_pci_get_data_size(struct vfio_device *vdev,
+				unsigned long *stop_copy_length)
+{
+	*stop_copy_length = sizeof(struct acc_vf_data);
+	return 0;
 }
 
 static int
···
 static const struct vfio_migration_ops hisi_acc_vfio_pci_migrn_state_ops = {
 	.migration_set_state = hisi_acc_vfio_pci_set_device_state,
 	.migration_get_state = hisi_acc_vfio_pci_get_device_state,
+	.migration_get_data_size = hisi_acc_vfio_pci_get_data_size,
 };
 
 static int hisi_acc_vfio_pci_migrn_init_dev(struct vfio_device *core_vdev)
···
 	hisi_acc_vdev->vf_dev = pdev;
 	mutex_init(&hisi_acc_vdev->state_mutex);
 
-	core_vdev->migration_flags = VFIO_MIGRATION_STOP_COPY;
+	core_vdev->migration_flags = VFIO_MIGRATION_STOP_COPY | VFIO_MIGRATION_PRE_COPY;
 	core_vdev->mig_ops = &hisi_acc_vfio_pci_migrn_state_ops;
 
 	return vfio_pci_core_init_dev(core_vdev);

drivers/vfio/pci/hisilicon/hisi_acc_vfio_pci.h

···
 	struct mutex lock;
 	bool disabled;
 
+	struct hisi_acc_vf_core_device *hisi_acc_vdev;
 	struct acc_vf_data vf_data;
 	size_t total_length;
 };
 
 struct hisi_acc_vf_core_device {
 	struct vfio_pci_core_device core_device;
+	u8 match_done:1;
 	u8 deferred_reset:1;
 	/* For migration state */
 	struct mutex state_mutex;

drivers/vfio/pci/mlx5/cmd.c

···
 
 int mlx5vf_cmd_suspend_vhca(struct mlx5vf_pci_core_device *mvdev, u16 op_mod)
 {
+	struct mlx5_vf_migration_file *migf = mvdev->saving_migf;
 	u32 out[MLX5_ST_SZ_DW(suspend_vhca_out)] = {};
 	u32 in[MLX5_ST_SZ_DW(suspend_vhca_in)] = {};
+	int err;
 
 	lockdep_assert_held(&mvdev->state_mutex);
 	if (mvdev->mdev_detach)
 		return -ENOTCONN;
 
+	/*
+	 * In case PRE_COPY is used, saving_migf is exposed while the device is
+	 * running. Make sure to run only once there is no active save command.
+	 * Running both in parallel, might end-up with a failure in the save
+	 * command once it will try to turn on 'tracking' on a suspended device.
+	 */
+	if (migf) {
+		err = wait_for_completion_interruptible(&migf->save_comp);
+		if (err)
+			return err;
+	}
+
 	MLX5_SET(suspend_vhca_in, in, opcode, MLX5_CMD_OP_SUSPEND_VHCA);
 	MLX5_SET(suspend_vhca_in, in, vhca_id, mvdev->vhca_id);
 	MLX5_SET(suspend_vhca_in, in, op_mod, op_mod);
 
-	return mlx5_cmd_exec_inout(mvdev->mdev, suspend_vhca, in, out);
+	err = mlx5_cmd_exec_inout(mvdev->mdev, suspend_vhca, in, out);
+	if (migf)
+		complete(&migf->save_comp);
+
+	return err;
 }
 
 int mlx5vf_cmd_resume_vhca(struct mlx5vf_pci_core_device *mvdev, u16 op_mod)
···
 }
 
 int mlx5vf_cmd_query_vhca_migration_state(struct mlx5vf_pci_core_device *mvdev,
-					  size_t *state_size)
+					  size_t *state_size, u8 query_flags)
 {
 	u32 out[MLX5_ST_SZ_DW(query_vhca_migration_state_out)] = {};
 	u32 in[MLX5_ST_SZ_DW(query_vhca_migration_state_in)] = {};
+	bool inc = query_flags & MLX5VF_QUERY_INC;
 	int ret;
 
 	lockdep_assert_held(&mvdev->state_mutex);
 	if (mvdev->mdev_detach)
 		return -ENOTCONN;
 
+	/*
+	 * In case PRE_COPY is used, saving_migf is exposed while device is
+	 * running. Make sure to run only once there is no active save command.
+	 * Running both in parallel, might end-up with a failure in the
+	 * incremental query command on un-tracked vhca.
+	 */
+	if (inc) {
+		ret = wait_for_completion_interruptible(&mvdev->saving_migf->save_comp);
+		if (ret)
+			return ret;
+		if (mvdev->saving_migf->state ==
+		    MLX5_MIGF_STATE_PRE_COPY_ERROR) {
+			/*
+			 * In case we had a PRE_COPY error, only query full
+			 * image for final image
+			 */
+			if (!(query_flags & MLX5VF_QUERY_FINAL)) {
+				*state_size = 0;
+				complete(&mvdev->saving_migf->save_comp);
+				return 0;
+			}
+			query_flags &= ~MLX5VF_QUERY_INC;
+		}
+	}
+
 	MLX5_SET(query_vhca_migration_state_in, in, opcode,
 		 MLX5_CMD_OP_QUERY_VHCA_MIGRATION_STATE);
 	MLX5_SET(query_vhca_migration_state_in, in, vhca_id, mvdev->vhca_id);
 	MLX5_SET(query_vhca_migration_state_in, in, op_mod, 0);
+	MLX5_SET(query_vhca_migration_state_in, in, incremental,
+		 query_flags & MLX5VF_QUERY_INC);
 
 	ret = mlx5_cmd_exec_inout(mvdev->mdev, query_vhca_migration_state, in,
 				  out);
+	if (inc)
+		complete(&mvdev->saving_migf->save_comp);
+
 	if (ret)
 		return ret;
 
···
 	if (MLX5_CAP_GEN(mvdev->mdev, adv_virtualization))
 		mvdev->core_device.vdev.log_ops = log_ops;
 
+	if (MLX5_CAP_GEN_2(mvdev->mdev, migration_multi_load) &&
+	    MLX5_CAP_GEN_2(mvdev->mdev, migration_tracking_state))
+		mvdev->core_device.vdev.migration_flags |=
+			VFIO_MIGRATION_PRE_COPY;
+
 end:
 	mlx5_vf_put_core_dev(mvdev->mdev);
 }
···
 }
 
 static int _create_mkey(struct mlx5_core_dev *mdev, u32 pdn,
-			struct mlx5_vf_migration_file *migf,
+			struct mlx5_vhca_data_buffer *buf,
 			struct mlx5_vhca_recv_buf *recv_buf,
 			u32 *mkey)
 {
-	size_t npages = migf ? DIV_ROUND_UP(migf->total_length, PAGE_SIZE) :
+	size_t npages = buf ? DIV_ROUND_UP(buf->allocated_length, PAGE_SIZE) :
 				recv_buf->npages;
 	int err = 0, inlen;
 	__be64 *mtt;
···
 		 DIV_ROUND_UP(npages, 2));
 	mtt = (__be64 *)MLX5_ADDR_OF(create_mkey_in, in, klm_pas_mtt);
 
-	if (migf) {
+	if (buf) {
 		struct sg_dma_page_iter dma_iter;
 
-		for_each_sgtable_dma_page(&migf->table.sgt, &dma_iter, 0)
+		for_each_sgtable_dma_page(&buf->table.sgt, &dma_iter, 0)
 			*mtt++ = cpu_to_be64(sg_page_iter_dma_address(&dma_iter));
 	} else {
 		int i;
···
 	MLX5_SET(mkc, mkc, qpn, 0xffffff);
 	MLX5_SET(mkc, mkc, log_page_size, PAGE_SHIFT);
 	MLX5_SET(mkc, mkc, translations_octword_size, DIV_ROUND_UP(npages, 2));
-	MLX5_SET64(mkc, mkc, len,
-		   migf ? migf->total_length : (npages * PAGE_SIZE));
+	MLX5_SET64(mkc, mkc, len, npages * PAGE_SIZE);
 	err = mlx5_core_create_mkey(mdev, mkey, in, inlen);
 	kvfree(in);
 	return err;
+}
+
+static int mlx5vf_dma_data_buffer(struct mlx5_vhca_data_buffer *buf)
+{
+	struct mlx5vf_pci_core_device *mvdev = buf->migf->mvdev;
+	struct mlx5_core_dev *mdev = mvdev->mdev;
+	int ret;
+
+	lockdep_assert_held(&mvdev->state_mutex);
+	if (mvdev->mdev_detach)
+		return -ENOTCONN;
+
+	if (buf->dmaed || !buf->allocated_length)
+		return -EINVAL;
+
+	ret = dma_map_sgtable(mdev->device, &buf->table.sgt, buf->dma_dir, 0);
+	if (ret)
+		return ret;
+
+	ret = _create_mkey(mdev, buf->migf->pdn, buf, NULL, &buf->mkey);
+	if (ret)
+		goto err;
+
+	buf->dmaed = true;
+
+	return 0;
+err:
+	dma_unmap_sgtable(mdev->device, &buf->table.sgt, buf->dma_dir, 0);
+	return ret;
+}
+
+void mlx5vf_free_data_buffer(struct mlx5_vhca_data_buffer *buf)
+{
+	struct mlx5_vf_migration_file *migf = buf->migf;
+	struct sg_page_iter sg_iter;
+
+	lockdep_assert_held(&migf->mvdev->state_mutex);
+	WARN_ON(migf->mvdev->mdev_detach);
+
+	if (buf->dmaed) {
+		mlx5_core_destroy_mkey(migf->mvdev->mdev, buf->mkey);
+		dma_unmap_sgtable(migf->mvdev->mdev->device, &buf->table.sgt,
+				  buf->dma_dir, 0);
+	}
+
+	/* Undo alloc_pages_bulk_array() */
+	for_each_sgtable_page(&buf->table.sgt, &sg_iter, 0)
+		__free_page(sg_page_iter_page(&sg_iter));
+	sg_free_append_table(&buf->table);
+	kfree(buf);
+}
+
+struct mlx5_vhca_data_buffer *
+mlx5vf_alloc_data_buffer(struct mlx5_vf_migration_file *migf,
+			 size_t length,
+			 enum dma_data_direction dma_dir)
+{
+	struct mlx5_vhca_data_buffer *buf;
+	int ret;
+
+	buf = kzalloc(sizeof(*buf), GFP_KERNEL);
+	if (!buf)
+		return ERR_PTR(-ENOMEM);
+
+	buf->dma_dir = dma_dir;
+	buf->migf = migf;
+	if (length) {
+		ret = mlx5vf_add_migration_pages(buf,
+				DIV_ROUND_UP_ULL(length, PAGE_SIZE));
+		if (ret)
+			goto end;
+
+		if (dma_dir != DMA_NONE) {
+			ret = mlx5vf_dma_data_buffer(buf);
+			if (ret)
+				goto end;
+		}
+	}
+
+	return buf;
+end:
+	mlx5vf_free_data_buffer(buf);
+	return ERR_PTR(ret);
+}
+
+void mlx5vf_put_data_buffer(struct mlx5_vhca_data_buffer *buf)
+{
+	spin_lock_irq(&buf->migf->list_lock);
+	list_add_tail(&buf->buf_elm, &buf->migf->avail_list);
+	spin_unlock_irq(&buf->migf->list_lock);
+}
+
+struct mlx5_vhca_data_buffer *
+mlx5vf_get_data_buffer(struct mlx5_vf_migration_file *migf,
+		       size_t length, enum dma_data_direction dma_dir)
+{
+	struct mlx5_vhca_data_buffer *buf, *temp_buf;
+	struct list_head free_list;
+
+	lockdep_assert_held(&migf->mvdev->state_mutex);
+	if (migf->mvdev->mdev_detach)
+		return ERR_PTR(-ENOTCONN);
+
+	INIT_LIST_HEAD(&free_list);
+
+	spin_lock_irq(&migf->list_lock);
+	list_for_each_entry_safe(buf, temp_buf, &migf->avail_list, buf_elm) {
+		if (buf->dma_dir == dma_dir) {
+			list_del_init(&buf->buf_elm);
+			if (buf->allocated_length >= length) {
+				spin_unlock_irq(&migf->list_lock);
+				goto found;
+			}
+			/*
+			 * Prevent holding redundant buffers. Put in a free
+			 * list and call at the end not under the spin lock
+			 * (&migf->list_lock) to mlx5vf_free_data_buffer which
+			 * might sleep.
+			 */
+			list_add(&buf->buf_elm, &free_list);
+		}
+	}
+	spin_unlock_irq(&migf->list_lock);
+	buf = mlx5vf_alloc_data_buffer(migf, length, dma_dir);
+
+found:
+	while ((temp_buf = list_first_entry_or_null(&free_list,
+				struct mlx5_vhca_data_buffer, buf_elm))) {
+		list_del(&temp_buf->buf_elm);
+		mlx5vf_free_data_buffer(temp_buf);
+	}
+
+	return buf;
 }
 
 void mlx5vf_mig_file_cleanup_cb(struct work_struct *_work)
···
 		struct mlx5vf_async_data, work);
 	struct mlx5_vf_migration_file *migf = container_of(async_data,
 		struct mlx5_vf_migration_file, async_data);
-	struct mlx5_core_dev *mdev = migf->mvdev->mdev;
 
 	mutex_lock(&migf->lock);
 	if (async_data->status) {
-		migf->is_err = true;
+		mlx5vf_put_data_buffer(async_data->buf);
+		if (async_data->header_buf)
+			mlx5vf_put_data_buffer(async_data->header_buf);
+		if (async_data->status == MLX5_CMD_STAT_BAD_RES_STATE_ERR)
+			migf->state = MLX5_MIGF_STATE_PRE_COPY_ERROR;
+		else
+			migf->state = MLX5_MIGF_STATE_ERROR;
 		wake_up_interruptible(&migf->poll_wait);
 	}
 	mutex_unlock(&migf->lock);
-
-	mlx5_core_destroy_mkey(mdev, async_data->mkey);
-	dma_unmap_sgtable(mdev->device, &migf->table.sgt, DMA_FROM_DEVICE, 0);
-	mlx5_core_dealloc_pd(mdev, async_data->pdn);
 	kvfree(async_data->out);
+	complete(&migf->save_comp);
 	fput(migf->filp);
+}
+
+static int add_buf_header(struct mlx5_vhca_data_buffer *header_buf,
+			  size_t image_size)
+{
+	struct mlx5_vf_migration_file *migf = header_buf->migf;
+	struct mlx5_vf_migration_header header = {};
+	unsigned long flags;
+	struct page *page;
+	u8 *to_buff;
+
+	header.image_size = cpu_to_le64(image_size);
+	page = mlx5vf_get_migration_page(header_buf, 0);
+	if (!page)
+		return -EINVAL;
+	to_buff = kmap_local_page(page);
+	memcpy(to_buff, &header, sizeof(header));
+	kunmap_local(to_buff);
+	header_buf->length = sizeof(header);
+	header_buf->header_image_size = image_size;
+	header_buf->start_pos = header_buf->migf->max_pos;
+	migf->max_pos += header_buf->length;
+	spin_lock_irqsave(&migf->list_lock, flags);
+	list_add_tail(&header_buf->buf_elm, &migf->buf_list);
+	spin_unlock_irqrestore(&migf->list_lock, flags);
+	return 0;
 }
 
 static void mlx5vf_save_callback(int status, struct mlx5_async_work *context)
···
 			struct mlx5_vf_migration_file, async_data);
 
 	if (!status) {
-		WRITE_ONCE(migf->total_length,
-			   MLX5_GET(save_vhca_state_out, async_data->out,
-				    actual_image_size));
+		size_t image_size;
+		unsigned long flags;
+
+		image_size = MLX5_GET(save_vhca_state_out, async_data->out,
+				      actual_image_size);
+		if (async_data->header_buf) {
+			status = add_buf_header(async_data->header_buf, image_size);
+			if (status)
+				goto err;
+		}
+		async_data->buf->length = image_size;
+		async_data->buf->start_pos = migf->max_pos;
+		migf->max_pos += async_data->buf->length;
+		spin_lock_irqsave(&migf->list_lock, flags);
+		list_add_tail(&async_data->buf->buf_elm, &migf->buf_list);
+		spin_unlock_irqrestore(&migf->list_lock, flags);
+		migf->state = async_data->last_chunk ?
+			MLX5_MIGF_STATE_COMPLETE : MLX5_MIGF_STATE_PRE_COPY;
 		wake_up_interruptible(&migf->poll_wait);
 	}
 
+err:
 	/*
 	 * The error and the cleanup flows can't run from an
 	 * interrupt context
 	 */
+	if (status == -EREMOTEIO)
+		status = MLX5_GET(save_vhca_state_out, async_data->out, status);
 	async_data->status = status;
 	queue_work(migf->mvdev->cb_wq, &async_data->work);
 }
 
 int mlx5vf_cmd_save_vhca_state(struct mlx5vf_pci_core_device *mvdev,
-			       struct mlx5_vf_migration_file *migf)
+			       struct mlx5_vf_migration_file *migf,
+			       struct mlx5_vhca_data_buffer *buf, bool inc,
+			       bool track)
 {
 	u32 out_size = MLX5_ST_SZ_BYTES(save_vhca_state_out);
 	u32 in[MLX5_ST_SZ_DW(save_vhca_state_in)] = {};
+	struct mlx5_vhca_data_buffer *header_buf = NULL;
 	struct mlx5vf_async_data *async_data;
-	struct mlx5_core_dev *mdev;
-	u32 pdn, mkey;
 	int err;
 
 	lockdep_assert_held(&mvdev->state_mutex);
 	if (mvdev->mdev_detach)
 		return -ENOTCONN;
 
-	mdev = mvdev->mdev;
-	err = mlx5_core_alloc_pd(mdev, &pdn);
+	err = wait_for_completion_interruptible(&migf->save_comp);
 	if (err)
 		return err;
 
-	err = dma_map_sgtable(mdev->device, &migf->table.sgt, DMA_FROM_DEVICE,
-			      0);
-	if (err)
-		goto err_dma_map;
-
-	err = _create_mkey(mdev, pdn, migf, NULL, &mkey);
-	if (err)
-		goto err_create_mkey;
+	if (migf->state == MLX5_MIGF_STATE_PRE_COPY_ERROR)
+		/*
+		 * In case we had a PRE_COPY error, SAVE is triggered only for
+		 * the final image, read device full image.
+		 */
+		inc = false;
 
 	MLX5_SET(save_vhca_state_in, in, opcode,
 		 MLX5_CMD_OP_SAVE_VHCA_STATE);
 	MLX5_SET(save_vhca_state_in, in, op_mod, 0);
 	MLX5_SET(save_vhca_state_in, in, vhca_id, mvdev->vhca_id);
-	MLX5_SET(save_vhca_state_in, in, mkey, mkey);
-	MLX5_SET(save_vhca_state_in, in, size, migf->total_length);
+	MLX5_SET(save_vhca_state_in, in, mkey, buf->mkey);
+	MLX5_SET(save_vhca_state_in, in, size, buf->allocated_length);
+	MLX5_SET(save_vhca_state_in, in, incremental, inc);
+	MLX5_SET(save_vhca_state_in, in, set_track, track);
 
 	async_data = &migf->async_data;
+	async_data->buf = buf;
+	async_data->last_chunk = !track;
 	async_data->out = kvzalloc(out_size, GFP_KERNEL);
 	if (!async_data->out) {
 		err = -ENOMEM;
 		goto err_out;
 	}
 
-	/* no data exists till the callback comes back */
-	migf->total_length = 0;
+	if (MLX5VF_PRE_COPY_SUPP(mvdev)) {
+		header_buf = mlx5vf_get_data_buffer(migf,
+			sizeof(struct mlx5_vf_migration_header), DMA_NONE);
+		if (IS_ERR(header_buf)) {
+			err = PTR_ERR(header_buf);
+			goto err_free;
+		}
+	}
+
+	if (async_data->last_chunk)
+		migf->state = MLX5_MIGF_STATE_SAVE_LAST;
+
+	async_data->header_buf = header_buf;
 	get_file(migf->filp);
-	async_data->mkey = mkey;
-	async_data->pdn = pdn;
 	err = mlx5_cmd_exec_cb(&migf->async_ctx, in, sizeof(in),
 			       async_data->out,
 			       out_size, mlx5vf_save_callback,
···
 	return 0;
 
 err_exec:
+	if (header_buf)
+		mlx5vf_put_data_buffer(header_buf);
 	fput(migf->filp);
+err_free:
 	kvfree(async_data->out);
 err_out:
-	mlx5_core_destroy_mkey(mdev, mkey);
-err_create_mkey:
-	dma_unmap_sgtable(mdev->device, &migf->table.sgt, DMA_FROM_DEVICE, 0);
-err_dma_map:
-	mlx5_core_dealloc_pd(mdev, pdn);
+	complete(&migf->save_comp);
 	return err;
 }
 
 int mlx5vf_cmd_load_vhca_state(struct mlx5vf_pci_core_device *mvdev,
-			       struct mlx5_vf_migration_file *migf)
+			       struct mlx5_vf_migration_file *migf,
+			       struct mlx5_vhca_data_buffer *buf)
 {
-	struct mlx5_core_dev *mdev;
-	u32 out[MLX5_ST_SZ_DW(save_vhca_state_out)] = {};
-	u32 in[MLX5_ST_SZ_DW(save_vhca_state_in)] = {};
-	u32 pdn, mkey;
+	u32 out[MLX5_ST_SZ_DW(load_vhca_state_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(load_vhca_state_in)] = {};
 	int err;
 
 	lockdep_assert_held(&mvdev->state_mutex);
 	if (mvdev->mdev_detach)
 		return -ENOTCONN;
 
-	mutex_lock(&migf->lock);
-	if (!migf->total_length) {
-		err = -EINVAL;
-		goto end;
+	if (!buf->dmaed) {
+		err = mlx5vf_dma_data_buffer(buf);
+		if (err)
+			return err;
 	}
-
-	mdev = mvdev->mdev;
-	err = mlx5_core_alloc_pd(mdev, &pdn);
-	if (err)
-		goto end;
-
-	err = dma_map_sgtable(mdev->device, &migf->table.sgt, DMA_TO_DEVICE, 0);
-	if (err)
-		goto err_reg;
-
-	err = _create_mkey(mdev, pdn, migf, NULL, &mkey);
-	if (err)
-		goto err_mkey;
 
 	MLX5_SET(load_vhca_state_in, in, opcode,
 		 MLX5_CMD_OP_LOAD_VHCA_STATE);
 	MLX5_SET(load_vhca_state_in, in, op_mod, 0);
 	MLX5_SET(load_vhca_state_in, in, vhca_id, mvdev->vhca_id);
-	MLX5_SET(load_vhca_state_in, in, mkey, mkey);
-	MLX5_SET(load_vhca_state_in, in, size, migf->total_length);
+	MLX5_SET(load_vhca_state_in, in, mkey, buf->mkey);
+	MLX5_SET(load_vhca_state_in, in, size, buf->length);
+	return mlx5_cmd_exec_inout(mvdev->mdev, load_vhca_state, in, out);
+}
 
-	err = mlx5_cmd_exec_inout(mdev, load_vhca_state, in, out);
+int mlx5vf_cmd_alloc_pd(struct mlx5_vf_migration_file *migf)
+{
+	int err;
 
-	mlx5_core_destroy_mkey(mdev, mkey);
-err_mkey:
-	dma_unmap_sgtable(mdev->device, &migf->table.sgt, DMA_TO_DEVICE, 0);
-err_reg:
-	mlx5_core_dealloc_pd(mdev, pdn);
-end:
-	mutex_unlock(&migf->lock);
+	lockdep_assert_held(&migf->mvdev->state_mutex);
+	if (migf->mvdev->mdev_detach)
+		return -ENOTCONN;
+
+	err = mlx5_core_alloc_pd(migf->mvdev->mdev, &migf->pdn);
 	return err;
+}
+
+void mlx5vf_cmd_dealloc_pd(struct mlx5_vf_migration_file *migf)
+{
+	lockdep_assert_held(&migf->mvdev->state_mutex);
+	if (migf->mvdev->mdev_detach)
+		return;
+
+	mlx5_core_dealloc_pd(migf->mvdev->mdev, migf->pdn);
+}
+
+void mlx5fv_cmd_clean_migf_resources(struct mlx5_vf_migration_file *migf)
+{
+	struct mlx5_vhca_data_buffer *entry;
+
+	lockdep_assert_held(&migf->mvdev->state_mutex);
+	WARN_ON(migf->mvdev->mdev_detach);
+
+	if (migf->buf) {
+		mlx5vf_free_data_buffer(migf->buf);
+		migf->buf = NULL;
+	}
+
+	if (migf->buf_header) {
+		mlx5vf_free_data_buffer(migf->buf_header);
+		migf->buf_header = NULL;
+	}
+
+	list_splice(&migf->avail_list, &migf->buf_list);
+
+	while ((entry = list_first_entry_or_null(&migf->buf_list,
+				struct mlx5_vhca_data_buffer, buf_elm))) {
+		list_del(&entry->buf_elm);
+		mlx5vf_free_data_buffer(entry);
+	}
+
+	mlx5vf_cmd_dealloc_pd(migf);
 }
 
 static void combine_ranges(struct rb_root_cached *root, u32 cur_nodes,

drivers/vfio/pci/mlx5/cmd.h

···
 #include <linux/mlx5/cq.h>
 #include <linux/mlx5/qp.h>
 
+#define MLX5VF_PRE_COPY_SUPP(mvdev) \
+	((mvdev)->core_device.vdev.migration_flags & VFIO_MIGRATION_PRE_COPY)
+
+enum mlx5_vf_migf_state {
+	MLX5_MIGF_STATE_ERROR = 1,
+	MLX5_MIGF_STATE_PRE_COPY_ERROR,
+	MLX5_MIGF_STATE_PRE_COPY,
+	MLX5_MIGF_STATE_SAVE_LAST,
+	MLX5_MIGF_STATE_COMPLETE,
+};
+
+enum mlx5_vf_load_state {
+	MLX5_VF_LOAD_STATE_READ_IMAGE_NO_HEADER,
+	MLX5_VF_LOAD_STATE_READ_HEADER,
+	MLX5_VF_LOAD_STATE_PREP_IMAGE,
+	MLX5_VF_LOAD_STATE_READ_IMAGE,
+	MLX5_VF_LOAD_STATE_LOAD_IMAGE,
+};
+
+struct mlx5_vf_migration_header {
+	__le64 image_size;
+	/* For future use in case we may need to change the kernel protocol */
+	__le64 flags;
+};
+
+struct mlx5_vhca_data_buffer {
+	struct sg_append_table table;
+	loff_t start_pos;
+	u64 length;
+	u64 allocated_length;
+	u64 header_image_size;
+	u32 mkey;
+	enum dma_data_direction dma_dir;
+	u8 dmaed:1;
+	struct list_head buf_elm;
+	struct mlx5_vf_migration_file *migf;
+	/* Optimize mlx5vf_get_migration_page() for sequential access */
+	struct scatterlist *last_offset_sg;
+	unsigned int sg_last_entry;
+	unsigned long last_offset;
+};
+
 struct mlx5vf_async_data {
 	struct mlx5_async_work cb_work;
 	struct work_struct work;
+	struct mlx5_vhca_data_buffer *buf;
+	struct mlx5_vhca_data_buffer *header_buf;
 	int status;
-	u32 pdn;
-	u32 mkey;
+	u8 last_chunk:1;
 	void *out;
 };
 
 struct mlx5_vf_migration_file {
 	struct file *filp;
 	struct mutex lock;
-	u8 disabled:1;
-	u8 is_err:1;
+	enum mlx5_vf_migf_state state;
 
-	struct sg_append_table table;
-	size_t total_length;
-	size_t allocated_length;
-
-	/* Optimize mlx5vf_get_migration_page() for sequential access */
-	struct scatterlist *last_offset_sg;
-	unsigned int sg_last_entry;
-	unsigned long last_offset;
+	enum mlx5_vf_load_state load_state;
+	u32 pdn;
+	loff_t max_pos;
+	struct mlx5_vhca_data_buffer *buf;
+	struct mlx5_vhca_data_buffer *buf_header;
+	spinlock_t list_lock;
+	struct list_head buf_list;
+	struct list_head avail_list;
 	struct mlx5vf_pci_core_device *mvdev;
 	wait_queue_head_t poll_wait;
+	struct completion save_comp;
 	struct mlx5_async_ctx async_ctx;
 	struct mlx5vf_async_data async_data;
 };
···
 	struct mlx5_core_dev *mdev;
 };
 
+enum {
+	MLX5VF_QUERY_INC = (1UL << 0),
+	MLX5VF_QUERY_FINAL = (1UL << 1),
+};
+
 int mlx5vf_cmd_suspend_vhca(struct mlx5vf_pci_core_device *mvdev, u16 op_mod);
 int mlx5vf_cmd_resume_vhca(struct mlx5vf_pci_core_device *mvdev, u16 op_mod);
 int mlx5vf_cmd_query_vhca_migration_state(struct mlx5vf_pci_core_device *mvdev,
-					  size_t *state_size);
+					  size_t *state_size, u8 query_flags);
 void mlx5vf_cmd_set_migratable(struct mlx5vf_pci_core_device *mvdev,
 			       const struct vfio_migration_ops *mig_ops,
 			       const struct vfio_log_ops *log_ops);
 void mlx5vf_cmd_remove_migratable(struct mlx5vf_pci_core_device *mvdev);
 void mlx5vf_cmd_close_migratable(struct mlx5vf_pci_core_device *mvdev);
 int mlx5vf_cmd_save_vhca_state(struct mlx5vf_pci_core_device *mvdev,
-			       struct mlx5_vf_migration_file *migf);
+			       struct mlx5_vf_migration_file *migf,
+			       struct mlx5_vhca_data_buffer *buf, bool inc,
+			       bool track);
 int mlx5vf_cmd_load_vhca_state(struct mlx5vf_pci_core_device *mvdev,
-			       struct mlx5_vf_migration_file *migf);
+			       struct mlx5_vf_migration_file *migf,
+			       struct mlx5_vhca_data_buffer *buf);
+int mlx5vf_cmd_alloc_pd(struct mlx5_vf_migration_file *migf);
+void mlx5vf_cmd_dealloc_pd(struct mlx5_vf_migration_file *migf);
+void mlx5fv_cmd_clean_migf_resources(struct mlx5_vf_migration_file *migf);
+struct mlx5_vhca_data_buffer *
+mlx5vf_alloc_data_buffer(struct mlx5_vf_migration_file *migf,
+			 size_t length, enum dma_data_direction dma_dir);
+void mlx5vf_free_data_buffer(struct mlx5_vhca_data_buffer *buf);
+struct mlx5_vhca_data_buffer *
+mlx5vf_get_data_buffer(struct mlx5_vf_migration_file *migf,
+		       size_t length, enum dma_data_direction dma_dir);
+void mlx5vf_put_data_buffer(struct mlx5_vhca_data_buffer *buf);
+int mlx5vf_add_migration_pages(struct mlx5_vhca_data_buffer *buf,
+			       unsigned int npages);
+struct page *mlx5vf_get_migration_page(struct mlx5_vhca_data_buffer *buf,
+				       unsigned long offset);
 void mlx5vf_state_mutex_unlock(struct mlx5vf_pci_core_device *mvdev);
 void mlx5vf_disable_fds(struct mlx5vf_pci_core_device *mvdev);
 void mlx5vf_mig_file_cleanup_cb(struct work_struct *_work);

drivers/vfio/pci/mlx5/main.c

···
 			    core_device);
 }
 
-static struct page *
-mlx5vf_get_migration_page(struct mlx5_vf_migration_file *migf,
+struct page *
+mlx5vf_get_migration_page(struct mlx5_vhca_data_buffer *buf,
 			  unsigned long offset)
 {
 	unsigned long cur_offset = 0;
···
 	unsigned int i;
 
 	/* All accesses are sequential */
-	if (offset < migf->last_offset || !migf->last_offset_sg) {
-		migf->last_offset = 0;
-		migf->last_offset_sg = migf->table.sgt.sgl;
-		migf->sg_last_entry = 0;
+	if (offset < buf->last_offset || !buf->last_offset_sg) {
+		buf->last_offset = 0;
+		buf->last_offset_sg = buf->table.sgt.sgl;
+		buf->sg_last_entry = 0;
 	}
 
-	cur_offset = migf->last_offset;
+	cur_offset = buf->last_offset;
 
-	for_each_sg(migf->last_offset_sg, sg,
-			migf->table.sgt.orig_nents - migf->sg_last_entry, i) {
+	for_each_sg(buf->last_offset_sg, sg,
+			buf->table.sgt.orig_nents - buf->sg_last_entry, i) {
 		if (offset < sg->length + cur_offset) {
-			migf->last_offset_sg = sg;
-			migf->sg_last_entry += i;
-			migf->last_offset = cur_offset;
+			buf->last_offset_sg = sg;
+			buf->sg_last_entry += i;
+			buf->last_offset = cur_offset;
 			return nth_page(sg_page(sg),
 					(offset - cur_offset) / PAGE_SIZE);
 		}
···
 	return NULL;
 }
 
-static int mlx5vf_add_migration_pages(struct mlx5_vf_migration_file *migf,
-				      unsigned int npages)
+int mlx5vf_add_migration_pages(struct mlx5_vhca_data_buffer *buf,
+			       unsigned int npages)
 {
 	unsigned int to_alloc = npages;
 	struct page **page_list;
···
 		}
 		to_alloc -= filled;
 		ret = sg_alloc_append_table_from_pages(
-			&migf->table, page_list, filled, 0,
+			&buf->table, page_list, filled, 0,
 			filled << PAGE_SHIFT, UINT_MAX, SG_MAX_SINGLE_ALLOC,
 			GFP_KERNEL);
 
 		if (ret)
 			goto err;
-		migf->allocated_length += filled * PAGE_SIZE;
+		buf->allocated_length += filled * PAGE_SIZE;
 		/* clean input for another bulk allocation */
 		memset(page_list, 0, filled * sizeof(*page_list));
 		to_fill = min_t(unsigned int, to_alloc,
···
 
 static void mlx5vf_disable_fd(struct mlx5_vf_migration_file *migf)
 {
-	struct sg_page_iter sg_iter;
-
 	mutex_lock(&migf->lock);
-	/* Undo alloc_pages_bulk_array() */
-	for_each_sgtable_page(&migf->table.sgt, &sg_iter, 0)
-		__free_page(sg_page_iter_page(&sg_iter));
-	sg_free_append_table(&migf->table);
-	migf->disabled = true;
-	migf->total_length = 0;
-	migf->allocated_length = 0;
+	migf->state = MLX5_MIGF_STATE_ERROR;
 	migf->filp->f_pos = 0;
 	mutex_unlock(&migf->lock);
 }
···
 	return 0;
 }
 
+static struct mlx5_vhca_data_buffer *
+mlx5vf_get_data_buff_from_pos(struct mlx5_vf_migration_file *migf, loff_t pos,
+			      bool *end_of_data)
+{
+	struct mlx5_vhca_data_buffer *buf;
+	bool found = false;
+
+	*end_of_data = false;
+	spin_lock_irq(&migf->list_lock);
+	if (list_empty(&migf->buf_list)) {
+		*end_of_data = true;
+		goto end;
+	}
+
+	buf = list_first_entry(&migf->buf_list, struct mlx5_vhca_data_buffer,
+			       buf_elm);
+	if (pos >= buf->start_pos &&
+	    pos < buf->start_pos + buf->length) {
+		found = true;
+		goto end;
+	}
+
+	/*
+	 * As we use a stream based FD we may expect having the data always
+	 * on first chunk
+	 */
+	migf->state = MLX5_MIGF_STATE_ERROR;
+
+end:
+	spin_unlock_irq(&migf->list_lock);
+	return found ? buf : NULL;
+}
+
+static ssize_t mlx5vf_buf_read(struct mlx5_vhca_data_buffer *vhca_buf,
+			       char __user **buf, size_t *len, loff_t *pos)
+{
+	unsigned long offset;
+	ssize_t done = 0;
+	size_t copy_len;
+
+	copy_len = min_t(size_t,
+			 vhca_buf->start_pos + vhca_buf->length - *pos, *len);
+	while (copy_len) {
+		size_t page_offset;
+		struct page *page;
+		size_t page_len;
+		u8 *from_buff;
+		int ret;
+
+		offset = *pos - vhca_buf->start_pos;
+		page_offset = offset % PAGE_SIZE;
+		offset -= page_offset;
+		page = mlx5vf_get_migration_page(vhca_buf, offset);
+		if (!page)
+			return -EINVAL;
+		page_len = min_t(size_t, copy_len, PAGE_SIZE - page_offset);
+		from_buff = kmap_local_page(page);
+		ret = copy_to_user(*buf, from_buff + page_offset, page_len);
+		kunmap_local(from_buff);
+		if (ret)
+			return -EFAULT;
+		*pos += page_len;
+		*len -= page_len;
+		*buf += page_len;
+		done += page_len;
+		copy_len -= page_len;
+	}
+
+	if (*pos >= vhca_buf->start_pos + vhca_buf->length) {
+		spin_lock_irq(&vhca_buf->migf->list_lock);
+		list_del_init(&vhca_buf->buf_elm);
+		list_add_tail(&vhca_buf->buf_elm, &vhca_buf->migf->avail_list);
+		spin_unlock_irq(&vhca_buf->migf->list_lock);
+	}
+
+	return done;
+}
+
 static ssize_t mlx5vf_save_read(struct file *filp, char __user *buf, size_t len,
 			       loff_t *pos)
 {
 	struct mlx5_vf_migration_file *migf = filp->private_data;
+	struct mlx5_vhca_data_buffer *vhca_buf;
+	bool first_loop_call = true;
+	bool end_of_data;
 	ssize_t done = 0;
 
 	if (pos)
···
 
 	if (!(filp->f_flags & O_NONBLOCK)) {
 		if (wait_event_interruptible(migf->poll_wait,
-			     READ_ONCE(migf->total_length) || migf->is_err))
+				!list_empty(&migf->buf_list) ||
+				migf->state == MLX5_MIGF_STATE_ERROR ||
+				migf->state == MLX5_MIGF_STATE_PRE_COPY_ERROR ||
+				migf->state == MLX5_MIGF_STATE_PRE_COPY ||
+				migf->state == MLX5_MIGF_STATE_COMPLETE))
 			return -ERESTARTSYS;
 	}
 
 	mutex_lock(&migf->lock);
-	if ((filp->f_flags & O_NONBLOCK) && !READ_ONCE(migf->total_length)) {
-		done = -EAGAIN;
-		goto out_unlock;
-	}
-	if (*pos > migf->total_length) {
-		done = -EINVAL;
-		goto out_unlock;
-	}
-	if (migf->disabled || migf->is_err) {
+	if (migf->state == MLX5_MIGF_STATE_ERROR) {
 		done = -ENODEV;
 		goto out_unlock;
 	}
 
-	len = min_t(size_t, migf->total_length - *pos, len);
 	while (len) {
-		size_t page_offset;
-		struct page *page;
-		size_t page_len;
-		u8 *from_buff;
-		int ret;
+		ssize_t count;
 
-		page_offset = (*pos) % PAGE_SIZE;
-		page = mlx5vf_get_migration_page(migf, *pos - page_offset);
-		if (!page) {
-			if (done == 0)
-				done = -EINVAL;
+		vhca_buf = mlx5vf_get_data_buff_from_pos(migf, *pos,
+							 &end_of_data);
+		if (first_loop_call) {
+			first_loop_call = false;
+			/* Temporary end of file as part of PRE_COPY */
+			if (end_of_data && (migf->state == MLX5_MIGF_STATE_PRE_COPY ||
+				migf->state == MLX5_MIGF_STATE_PRE_COPY_ERROR)) {
+				done = -ENOMSG;
+				goto out_unlock;
+			}
+
+			if (end_of_data && migf->state != MLX5_MIGF_STATE_COMPLETE) {
+				if (filp->f_flags & O_NONBLOCK) {
+					done = -EAGAIN;
+					goto out_unlock;
+				}
+			}
+		}
+
+		if (end_of_data)
+			goto out_unlock;
+
+		if (!vhca_buf) {
+			done = -EINVAL;
 			goto out_unlock;
 		}
 
-		page_len = min_t(size_t, len, PAGE_SIZE - page_offset);
-		from_buff = kmap_local_page(page);
-		ret = copy_to_user(buf, from_buff + page_offset, page_len);
-		kunmap_local(from_buff);
-		if (ret) {
-			done = -EFAULT;
+		count = mlx5vf_buf_read(vhca_buf, &buf, &len, pos);
+		if (count < 0) {
+			done = count;
 			goto out_unlock;
 		}
-		*pos += page_len;
-		len -= page_len;
-		done += page_len;
-		buf += page_len;
+		done += count;
 	}
 
 out_unlock:
···
 	poll_wait(filp, &migf->poll_wait, wait);
 
 	mutex_lock(&migf->lock);
-	if (migf->disabled || migf->is_err)
+	if (migf->state == MLX5_MIGF_STATE_ERROR)
 		pollflags = EPOLLIN | EPOLLRDNORM | EPOLLRDHUP;
-	else if (READ_ONCE(migf->total_length))
+	else if (!list_empty(&migf->buf_list) ||
+		 migf->state == MLX5_MIGF_STATE_COMPLETE)
 		pollflags = EPOLLIN | EPOLLRDNORM;
 	mutex_unlock(&migf->lock);
 
 	return pollflags;
 }
 
+/*
+ * FD is exposed and user can use it after receiving an error.
+ * Mark migf in error, and wake the user.
+ */
+static void mlx5vf_mark_err(struct mlx5_vf_migration_file *migf)
+{
+	migf->state = MLX5_MIGF_STATE_ERROR;
+	wake_up_interruptible(&migf->poll_wait);
+}
+
+static long mlx5vf_precopy_ioctl(struct file *filp, unsigned int cmd,
+				 unsigned long arg)
+{
+	struct mlx5_vf_migration_file *migf = filp->private_data;
+	struct mlx5vf_pci_core_device *mvdev = migf->mvdev;
+	struct mlx5_vhca_data_buffer *buf;
+	struct vfio_precopy_info info = {};
+	loff_t *pos = &filp->f_pos;
+	unsigned long minsz;
+	size_t inc_length = 0;
+	bool end_of_data;
+	int ret;
+
+	if (cmd != VFIO_MIG_GET_PRECOPY_INFO)
+		return -ENOTTY;
+
+	minsz = offsetofend(struct vfio_precopy_info, dirty_bytes);
+
+	if (copy_from_user(&info, (void __user *)arg, minsz))
+		return -EFAULT;
+
+	if (info.argsz < minsz)
+		return -EINVAL;
+
+	mutex_lock(&mvdev->state_mutex);
+	if (mvdev->mig_state != VFIO_DEVICE_STATE_PRE_COPY &&
+	    mvdev->mig_state != VFIO_DEVICE_STATE_PRE_COPY_P2P) {
+		ret = -EINVAL;
+		goto err_state_unlock;
+	}
+
+	/*
+	 * We can't issue a SAVE command when the device is suspended, so as
+	 * part of VFIO_DEVICE_STATE_PRE_COPY_P2P no reason to query for extra
+	 * bytes that can't be read.
+	 */
+	if (mvdev->mig_state == VFIO_DEVICE_STATE_PRE_COPY) {
+		/*
+		 * Once the query returns it's guaranteed that there is no
+		 * active SAVE command.
+		 * As so, the other code below is safe with the proper locks.
+		 */
+		ret = mlx5vf_cmd_query_vhca_migration_state(mvdev, &inc_length,
+							    MLX5VF_QUERY_INC);
+		if (ret)
+			goto err_state_unlock;
+	}
+
+	mutex_lock(&migf->lock);
+	if (migf->state == MLX5_MIGF_STATE_ERROR) {
+		ret = -ENODEV;
+		goto err_migf_unlock;
+	}
+
+	buf = mlx5vf_get_data_buff_from_pos(migf, *pos, &end_of_data);
+	if (buf) {
+		if (buf->start_pos == 0) {
+			info.initial_bytes = buf->header_image_size - *pos;
+		} else if (buf->start_pos ==
+				sizeof(struct mlx5_vf_migration_header)) {
+			/* First data buffer following the header */
+			info.initial_bytes = buf->start_pos +
+						buf->length - *pos;
+		} else {
+			info.dirty_bytes = buf->start_pos + buf->length - *pos;
+		}
+	} else {
+		if (!end_of_data) {
+			ret = -EINVAL;
+			goto err_migf_unlock;
+		}
+
+		info.dirty_bytes = inc_length;
+	}
+
+	if (!end_of_data || !inc_length) {
+		mutex_unlock(&migf->lock);
+		goto done;
+	}
+
+	mutex_unlock(&migf->lock);
+	/*
+	 * We finished transferring the current state and the device has a
+	 * dirty state, save a new state to be ready for.
+	 */
+	buf = mlx5vf_get_data_buffer(migf, inc_length, DMA_FROM_DEVICE);
+	if (IS_ERR(buf)) {
+		ret = PTR_ERR(buf);
+		mlx5vf_mark_err(migf);
+		goto err_state_unlock;
+	}
+
+	ret = mlx5vf_cmd_save_vhca_state(mvdev, migf, buf, true, true);
+	if (ret) {
+		mlx5vf_mark_err(migf);
+		mlx5vf_put_data_buffer(buf);
+		goto err_state_unlock;
+	}
+
+done:
+	mlx5vf_state_mutex_unlock(mvdev);
+	if (copy_to_user((void __user *)arg, &info, minsz))
+		return -EFAULT;
+	return 0;
+
+err_migf_unlock:
+	mutex_unlock(&migf->lock);
+err_state_unlock:
+	mlx5vf_state_mutex_unlock(mvdev);
+	return ret;
+}
+
 static const struct file_operations mlx5vf_save_fops = {
 	.owner = THIS_MODULE,
 	.read = mlx5vf_save_read,
 	.poll = mlx5vf_save_poll,
+	.unlocked_ioctl = mlx5vf_precopy_ioctl,
+	.compat_ioctl = compat_ptr_ioctl,
 	.release = mlx5vf_release_file,
 	.llseek = no_llseek,
 };
 
+static int mlx5vf_pci_save_device_inc_data(struct mlx5vf_pci_core_device *mvdev)
+{
+	struct mlx5_vf_migration_file *migf = mvdev->saving_migf;
+	struct mlx5_vhca_data_buffer *buf;
+	size_t length;
+	int ret;
+
+	if (migf->state == MLX5_MIGF_STATE_ERROR)
+		return -ENODEV;
+
+	ret = mlx5vf_cmd_query_vhca_migration_state(mvdev, &length,
+				MLX5VF_QUERY_INC | MLX5VF_QUERY_FINAL);
+	if (ret)
+		goto err;
+
+	buf = mlx5vf_get_data_buffer(migf, length, DMA_FROM_DEVICE);
+	if (IS_ERR(buf)) {
+		ret = PTR_ERR(buf);
+		goto err;
+	}
+
+	ret = mlx5vf_cmd_save_vhca_state(mvdev, migf, buf, true, false);
+	if (ret)
+		goto err_save;
+
+	return 0;
+
+err_save:
+	mlx5vf_put_data_buffer(buf);
+err:
+	mlx5vf_mark_err(migf);
+	return ret;
+}
+
 static struct mlx5_vf_migration_file *
-mlx5vf_pci_save_device_data(struct mlx5vf_pci_core_device *mvdev)
+mlx5vf_pci_save_device_data(struct mlx5vf_pci_core_device *mvdev, bool track)
 {
 	struct mlx5_vf_migration_file *migf;
+	struct mlx5_vhca_data_buffer *buf;
+	size_t length;
 	int ret;
 
 	migf = kzalloc(sizeof(*migf), GFP_KERNEL);
···
 	migf->filp = anon_inode_getfile("mlx5vf_mig", &mlx5vf_save_fops, migf,
 					O_RDONLY);
 	if (IS_ERR(migf->filp)) {
-		int err = PTR_ERR(migf->filp);
-
-		kfree(migf);
-		return ERR_PTR(err);
+		ret = PTR_ERR(migf->filp);
+		goto end;
 	}
+
+	migf->mvdev = mvdev;
+	ret = mlx5vf_cmd_alloc_pd(migf);
+	if (ret)
+		goto out_free;
 
 	stream_open(migf->filp->f_inode, migf->filp);
 	mutex_init(&migf->lock);
 	init_waitqueue_head(&migf->poll_wait);
+	init_completion(&migf->save_comp);
+	/*
+	 * save_comp is being used as a binary semaphore built from
+	 * a completion. A normal mutex cannot be used because the lock is
+	 * passed between kernel threads and lockdep can't model this.
+	 */
+	complete(&migf->save_comp);
 	mlx5_cmd_init_async_ctx(mvdev->mdev, &migf->async_ctx);
 	INIT_WORK(&migf->async_data.work, mlx5vf_mig_file_cleanup_cb);
-	ret = mlx5vf_cmd_query_vhca_migration_state(mvdev,
-						    &migf->total_length);
+	INIT_LIST_HEAD(&migf->buf_list);
+	INIT_LIST_HEAD(&migf->avail_list);
+	spin_lock_init(&migf->list_lock);
+	ret = mlx5vf_cmd_query_vhca_migration_state(mvdev, &length, 0);
 	if (ret)
-		goto out_free;
+		goto out_pd;
 
-	ret = mlx5vf_add_migration_pages(
-		migf, DIV_ROUND_UP_ULL(migf->total_length, PAGE_SIZE));
-	if (ret)
-		goto out_free;
+	buf = mlx5vf_alloc_data_buffer(migf, length, DMA_FROM_DEVICE);
+	if (IS_ERR(buf)) {
+		ret = PTR_ERR(buf);
+		goto out_pd;
+	}
 
-	migf->mvdev = mvdev;
-	ret = mlx5vf_cmd_save_vhca_state(mvdev, migf);
+	ret = mlx5vf_cmd_save_vhca_state(mvdev, migf, buf, false, track);
 	if (ret)
-		goto out_free;
+		goto out_save;
 	return migf;
+out_save:
+	mlx5vf_free_data_buffer(buf);
+out_pd:
+	mlx5vf_cmd_dealloc_pd(migf);
 out_free:
 	fput(migf->filp);
+end:
+	kfree(migf);
 	return ERR_PTR(ret);
+}
+
+static int
+mlx5vf_append_page_to_mig_buf(struct mlx5_vhca_data_buffer *vhca_buf,
+			      const char __user **buf, size_t *len,
+			      loff_t *pos, ssize_t *done)
+{
+	unsigned long offset;
+	size_t page_offset;
+	struct page *page;
+	size_t page_len;
+	u8 *to_buff;
+	int ret;
+
+	offset = *pos - vhca_buf->start_pos;
+	page_offset = offset % PAGE_SIZE;
+
+	page = mlx5vf_get_migration_page(vhca_buf, offset - page_offset);
+	if (!page)
+		return -EINVAL;
+	page_len = min_t(size_t, *len, PAGE_SIZE - page_offset);
+	to_buff = kmap_local_page(page);
+	ret = copy_from_user(to_buff + page_offset, *buf, page_len);
+	kunmap_local(to_buff);
+	if (ret)
+		return -EFAULT;
+
+	*pos += page_len;
+	*done += page_len;
+	*buf += page_len;
+	*len -= page_len;
+	vhca_buf->length += page_len;
+	return 0;
+}
+
+static int
+mlx5vf_resume_read_image_no_header(struct mlx5_vhca_data_buffer *vhca_buf,
+				   loff_t requested_length,
+				   const char __user **buf, size_t *len,
+				   loff_t *pos, ssize_t *done)
+{
+	int ret;
+
+	if (requested_length > MAX_MIGRATION_SIZE)
+		return -ENOMEM;
+
+	if (vhca_buf->allocated_length < requested_length) {
+		ret = mlx5vf_add_migration_pages(
+			vhca_buf,
+			DIV_ROUND_UP(requested_length - vhca_buf->allocated_length,
+				     PAGE_SIZE));
+		if (ret)
+			return ret;
+	}
+
+	while (*len) {
+		ret = mlx5vf_append_page_to_mig_buf(vhca_buf, buf, len, pos,
+						    done);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static ssize_t
+mlx5vf_resume_read_image(struct mlx5_vf_migration_file *migf,
+			 struct mlx5_vhca_data_buffer *vhca_buf,
+			 size_t image_size, const char __user **buf,
+			 size_t *len, loff_t *pos, ssize_t *done,
+			 bool *has_work)
+{
+	size_t copy_len, to_copy;
+	int ret;
+
+	to_copy = min_t(size_t, *len, image_size - vhca_buf->length);
+	copy_len = to_copy;
+	while (to_copy) {
+		ret = mlx5vf_append_page_to_mig_buf(vhca_buf, buf, &to_copy, pos,
+						    done);
+		if (ret)
+			return ret;
+	}
+
+	*len -= copy_len;
+	if (vhca_buf->length == image_size) {
+		migf->load_state = MLX5_VF_LOAD_STATE_LOAD_IMAGE;
+		migf->max_pos += image_size;
+		*has_work = true;
+	}
+
+	return 0;
+}
+
+static int
+mlx5vf_resume_read_header(struct mlx5_vf_migration_file *migf,
+			  struct mlx5_vhca_data_buffer *vhca_buf,
+			  const char __user **buf,
+			  size_t *len, loff_t *pos,
+			  ssize_t *done, bool *has_work)
+{
+	struct page *page;
+	size_t copy_len;
+	u8 *to_buff;
+	int ret;
+
+	copy_len = min_t(size_t, *len,
+		sizeof(struct mlx5_vf_migration_header) - vhca_buf->length);
+	page = mlx5vf_get_migration_page(vhca_buf, 0);
+	if (!page)
+		return -EINVAL;
+	to_buff = kmap_local_page(page);
+	ret = copy_from_user(to_buff + vhca_buf->length, *buf, copy_len);
+	if (ret) {
+		ret = -EFAULT;
+		goto end;
+	}
+
+	*buf += copy_len;
+	*pos += copy_len;
+	*done += copy_len;
+	*len -= copy_len;
+	vhca_buf->length += copy_len;
+	if (vhca_buf->length == sizeof(struct mlx5_vf_migration_header)) {
+		u64 flags;
+
+		vhca_buf->header_image_size = le64_to_cpup((__le64 *)to_buff);
+		if (vhca_buf->header_image_size > MAX_MIGRATION_SIZE) {
+			ret = -ENOMEM;
+			goto end;
+		}
+
+		flags = le64_to_cpup((__le64 *)(to_buff +
+			    offsetof(struct mlx5_vf_migration_header, flags)));
+		if (flags) {
+			ret = -EOPNOTSUPP;
+			goto end;
+		}
+
+		migf->load_state = MLX5_VF_LOAD_STATE_PREP_IMAGE;
+		migf->max_pos += vhca_buf->length;
+		*has_work = true;
+	}
+end:
+	kunmap_local(to_buff);
+	return ret;
 }
 
 static ssize_t mlx5vf_resume_write(struct file *filp, const char __user *buf,
 				   size_t len, loff_t *pos)
 {
 	struct mlx5_vf_migration_file *migf = filp->private_data;
+	struct mlx5_vhca_data_buffer *vhca_buf = migf->buf;
+	struct mlx5_vhca_data_buffer *vhca_buf_header = migf->buf_header;
 	loff_t requested_length;
+	bool has_work = false;
 	ssize_t done = 0;
+	int ret = 0;
 
 	if (pos)
 		return -ESPIPE;
···
 	    check_add_overflow((loff_t)len, *pos, &requested_length))
 		return -EINVAL;
 
-	if (requested_length > MAX_MIGRATION_SIZE)
-		return -ENOMEM;
-
+	mutex_lock(&migf->mvdev->state_mutex);
 	mutex_lock(&migf->lock);
-	if (migf->disabled) {
-		done = -ENODEV;
+	if (migf->state == MLX5_MIGF_STATE_ERROR) {
+		ret = -ENODEV;
 		goto out_unlock;
 	}
 
-	if (migf->allocated_length < requested_length) {
-		done = mlx5vf_add_migration_pages(
-			migf,
-			DIV_ROUND_UP(requested_length - migf->allocated_length,
-				     PAGE_SIZE));
-		if (done)
-			goto out_unlock;
+	while (len || has_work) {
+		has_work = false;
+		switch (migf->load_state) {
+		case MLX5_VF_LOAD_STATE_READ_HEADER:
+			ret = mlx5vf_resume_read_header(migf, vhca_buf_header,
+							&buf, &len, pos,
+							&done, &has_work);
+			if (ret)
+				goto out_unlock;
+			break;
+		case MLX5_VF_LOAD_STATE_PREP_IMAGE:
+		{
+			u64 size = vhca_buf_header->header_image_size;
+
+			if (vhca_buf->allocated_length < size) {
+				mlx5vf_free_data_buffer(vhca_buf);
+
+				migf->buf = mlx5vf_alloc_data_buffer(migf,
+							size, DMA_TO_DEVICE);
+				if (IS_ERR(migf->buf)) {
+					ret = PTR_ERR(migf->buf);
+					migf->buf = NULL;
+					goto out_unlock;
+				}
+
+				vhca_buf = migf->buf;
+			}
+
+			vhca_buf->start_pos = migf->max_pos;
+			migf->load_state = MLX5_VF_LOAD_STATE_READ_IMAGE;
+			break;
+		}
+		case MLX5_VF_LOAD_STATE_READ_IMAGE_NO_HEADER:
+			ret = mlx5vf_resume_read_image_no_header(vhca_buf,
+						requested_length,
+						&buf, &len, pos, &done);
+			if (ret)
+				goto out_unlock;
+			break;
+		case MLX5_VF_LOAD_STATE_READ_IMAGE:
+			ret = mlx5vf_resume_read_image(migf, vhca_buf,
+						vhca_buf_header->header_image_size,
+						&buf, &len, pos, &done, &has_work);
+			if (ret)
+				goto out_unlock;
+			break;
+		case MLX5_VF_LOAD_STATE_LOAD_IMAGE:
+			ret = mlx5vf_cmd_load_vhca_state(migf->mvdev, migf, vhca_buf);
+			if (ret)
+				goto out_unlock;
+			migf->load_state = MLX5_VF_LOAD_STATE_READ_HEADER;
+
+			/* prep header buf for next image */
+			vhca_buf_header->length = 0;
+			vhca_buf_header->header_image_size = 0;
+			/* prep data buf for next image */
+			vhca_buf->length = 0;
+
+			break;
+		default:
+			break;
+		}
 	}
 
-	while (len) {
-		size_t page_offset;
-		struct page *page;
-		size_t page_len;
-		u8 *to_buff;
-		int ret;
-
-		page_offset = (*pos) % PAGE_SIZE;
-		page = mlx5vf_get_migration_page(migf, *pos - page_offset);
-		if (!page) {
-			if (done == 0)
-				done = -EINVAL;
-			goto out_unlock;
-		}
-
-		page_len = min_t(size_t, len, PAGE_SIZE - page_offset);
-		to_buff = kmap_local_page(page);
-		ret = copy_from_user(to_buff + page_offset, buf, page_len);
-		kunmap_local(to_buff);
-		if (ret) {
-			done = -EFAULT;
-			goto out_unlock;
-		}
-		*pos += page_len;
-		len -= page_len;
-		done += page_len;
-		buf += page_len;
-		migf->total_length += page_len;
-	}
 out_unlock:
+	if (ret)
+		migf->state = MLX5_MIGF_STATE_ERROR;
 	mutex_unlock(&migf->lock);
-	return done;
+	mlx5vf_state_mutex_unlock(migf->mvdev);
+	return ret ? ret : done;
 }
 
 static const struct file_operations mlx5vf_resume_fops = {
···
 mlx5vf_pci_resume_device_data(struct mlx5vf_pci_core_device *mvdev)
 {
 	struct mlx5_vf_migration_file *migf;
+	struct mlx5_vhca_data_buffer *buf;
+	int ret;
 
 	migf = kzalloc(sizeof(*migf), GFP_KERNEL);
 	if (!migf)
···
 	migf->filp = anon_inode_getfile("mlx5vf_mig", &mlx5vf_resume_fops, migf,
 					O_WRONLY);
 	if (IS_ERR(migf->filp)) {
-		int err = PTR_ERR(migf->filp);
-
-		kfree(migf);
-		return ERR_PTR(err);
+		ret = PTR_ERR(migf->filp);
+		goto end;
 	}
+
+	migf->mvdev = mvdev;
+	ret = mlx5vf_cmd_alloc_pd(migf);
+	if (ret)
+		goto out_free;
+
+	buf = mlx5vf_alloc_data_buffer(migf, 0, DMA_TO_DEVICE);
+	if (IS_ERR(buf)) {
+		ret = PTR_ERR(buf);
+		goto out_pd;
+	}
+
+	migf->buf = buf;
+	if (MLX5VF_PRE_COPY_SUPP(mvdev)) {
+		buf = mlx5vf_alloc_data_buffer(migf,
+			sizeof(struct mlx5_vf_migration_header), DMA_NONE);
+		if (IS_ERR(buf)) {
+			ret = PTR_ERR(buf);
+			goto out_buf;
+		}
+
+		migf->buf_header = buf;
+		migf->load_state = MLX5_VF_LOAD_STATE_READ_HEADER;
+	} else {
+		/* Initial state will be to read the image */
+		migf->load_state = MLX5_VF_LOAD_STATE_READ_IMAGE_NO_HEADER;
+	}
+
 	stream_open(migf->filp->f_inode, migf->filp);
 	mutex_init(&migf->lock);
+	INIT_LIST_HEAD(&migf->buf_list);
+	INIT_LIST_HEAD(&migf->avail_list);
+	spin_lock_init(&migf->list_lock);
 	return migf;
+out_buf:
+	mlx5vf_free_data_buffer(migf->buf);
+out_pd:
+	mlx5vf_cmd_dealloc_pd(migf);
+out_free:
+	fput(migf->filp);
+end:
+	kfree(migf);
+	return ERR_PTR(ret);
 }
 
 void mlx5vf_disable_fds(struct mlx5vf_pci_core_device *mvdev)
 {
 	if (mvdev->resuming_migf) {
 		mlx5vf_disable_fd(mvdev->resuming_migf);
+		mlx5fv_cmd_clean_migf_resources(mvdev->resuming_migf);
 		fput(mvdev->resuming_migf->filp);
 		mvdev->resuming_migf = NULL;
 	}
···
 		mlx5_cmd_cleanup_async_ctx(&mvdev->saving_migf->async_ctx);
 		cancel_work_sync(&mvdev->saving_migf->async_data.work);
 		mlx5vf_disable_fd(mvdev->saving_migf);
+		mlx5fv_cmd_clean_migf_resources(mvdev->saving_migf);
 		fput(mvdev->saving_migf->filp);
 		mvdev->saving_migf = NULL;
 	}
···
 		return NULL;
 	}
 
-	if (cur == VFIO_DEVICE_STATE_RUNNING && new == VFIO_DEVICE_STATE_RUNNING_P2P) {
+	if ((cur == VFIO_DEVICE_STATE_RUNNING && new == VFIO_DEVICE_STATE_RUNNING_P2P) ||
+	    (cur == VFIO_DEVICE_STATE_PRE_COPY && new == VFIO_DEVICE_STATE_PRE_COPY_P2P)) {
 		ret = mlx5vf_cmd_suspend_vhca(mvdev,
 			MLX5_SUSPEND_VHCA_IN_OP_MOD_SUSPEND_INITIATOR);
 		if (ret)
···
 		return NULL;
 	}
 
-	if (cur == VFIO_DEVICE_STATE_RUNNING_P2P && new == VFIO_DEVICE_STATE_RUNNING) {
+	if ((cur == VFIO_DEVICE_STATE_RUNNING_P2P && new == VFIO_DEVICE_STATE_RUNNING) ||
+	    (cur == VFIO_DEVICE_STATE_PRE_COPY_P2P && new == VFIO_DEVICE_STATE_PRE_COPY)) {
 		ret = mlx5vf_cmd_resume_vhca(mvdev,
 			MLX5_RESUME_VHCA_IN_OP_MOD_RESUME_INITIATOR);
 		if (ret)
···
 	if (cur == VFIO_DEVICE_STATE_STOP && new == VFIO_DEVICE_STATE_STOP_COPY) {
 		struct mlx5_vf_migration_file *migf;
 
-		migf = mlx5vf_pci_save_device_data(mvdev);
+		migf = mlx5vf_pci_save_device_data(mvdev, false);
 		if (IS_ERR(migf))
 			return ERR_CAST(migf);
 		get_file(migf->filp);
···
 		return migf->filp;
 	}
 
-	if ((cur == VFIO_DEVICE_STATE_STOP_COPY && new == VFIO_DEVICE_STATE_STOP)) {
+	if ((cur == VFIO_DEVICE_STATE_STOP_COPY && new == VFIO_DEVICE_STATE_STOP) ||
+	    (cur == VFIO_DEVICE_STATE_PRE_COPY && new == VFIO_DEVICE_STATE_RUNNING) ||
+	    (cur == VFIO_DEVICE_STATE_PRE_COPY_P2P &&
+	     new == VFIO_DEVICE_STATE_RUNNING_P2P)) {
 		mlx5vf_disable_fds(mvdev);
 		return NULL;
 	}
···
 	}
 
 	if (cur == VFIO_DEVICE_STATE_RESUMING && new == VFIO_DEVICE_STATE_STOP) {
-		ret = mlx5vf_cmd_load_vhca_state(mvdev,
-						 mvdev->resuming_migf);
-		if (ret)
-			return ERR_PTR(ret);
+		if (!MLX5VF_PRE_COPY_SUPP(mvdev)) {
+			ret = mlx5vf_cmd_load_vhca_state(mvdev,
+							 mvdev->resuming_migf,
+							 mvdev->resuming_migf->buf);
+			if (ret)
+				return ERR_PTR(ret);
+		}
 		mlx5vf_disable_fds(mvdev);
 		return NULL;
+	}
+
+	if ((cur == VFIO_DEVICE_STATE_RUNNING && new == VFIO_DEVICE_STATE_PRE_COPY) ||
+	    (cur == VFIO_DEVICE_STATE_RUNNING_P2P &&
+	     new == VFIO_DEVICE_STATE_PRE_COPY_P2P)) {
+		struct mlx5_vf_migration_file *migf;
+
+		migf = mlx5vf_pci_save_device_data(mvdev, true);
+		if (IS_ERR(migf))
+			return ERR_CAST(migf);
+		get_file(migf->filp);
+		mvdev->saving_migf = migf;
+		return migf->filp;
+	}
+
+	if (cur == VFIO_DEVICE_STATE_PRE_COPY_P2P && new == VFIO_DEVICE_STATE_STOP_COPY) {
+		ret = mlx5vf_cmd_suspend_vhca(mvdev,
+			MLX5_SUSPEND_VHCA_IN_OP_MOD_SUSPEND_RESPONDER);
+		if (ret)
+			return ERR_PTR(ret);
+		ret = mlx5vf_pci_save_device_inc_data(mvdev);
+		return ret ? ERR_PTR(ret) : NULL;
 	}
 
 	/*
···
 	}
 	mlx5vf_state_mutex_unlock(mvdev);
 	return res;
+}
+
+static int mlx5vf_pci_get_data_size(struct vfio_device *vdev,
+				    unsigned long *stop_copy_length)
+{
+	struct mlx5vf_pci_core_device *mvdev = container_of(
+		vdev, struct mlx5vf_pci_core_device, core_device.vdev);
+	size_t state_size;
+	int ret;
+
+	mutex_lock(&mvdev->state_mutex);
+	ret = mlx5vf_cmd_query_vhca_migration_state(mvdev,
+						    &state_size, 0);
+	if (!ret)
+		*stop_copy_length = state_size;
+	mlx5vf_state_mutex_unlock(mvdev);
+	return ret;
 }
 
 static int mlx5vf_pci_get_device_state(struct vfio_device *vdev,
···
 static const struct vfio_migration_ops mlx5vf_pci_mig_ops = {
 	.migration_set_state = mlx5vf_pci_set_device_state,
 	.migration_get_state = mlx5vf_pci_get_device_state,
+	.migration_get_data_size = mlx5vf_pci_get_data_size,
 };
 
 static const struct vfio_log_ops mlx5vf_pci_log_ops = {
···
 	.driver_managed_dma = true,
 };
 
-static void __exit mlx5vf_pci_cleanup(void)
-{
-	pci_unregister_driver(&mlx5vf_pci_driver);
-}
-
-static int __init mlx5vf_pci_init(void)
-{
-	return pci_register_driver(&mlx5vf_pci_driver);
-}
-
-module_init(mlx5vf_pci_init);
-module_exit(mlx5vf_pci_cleanup);
+module_pci_driver(mlx5vf_pci_driver);
 
 MODULE_LICENSE("GPL");
 MODULE_AUTHOR("Max Gurtovoy <[e-mail]>");

drivers/vfio/pci/vfio_pci_core.c

···
 #include <linux/vgaarb.h>
 #include <linux/nospec.h>
 #include <linux/sched/mm.h>
+#if IS_ENABLED(CONFIG_EEH)
+#include <asm/eeh.h>
+#endif
 
 #include "vfio_pci_priv.h"
 
···
 		vdev->sriov_pf_core_dev->vf_token->users--;
 		mutex_unlock(&vdev->sriov_pf_core_dev->vf_token->lock);
 	}
-	vfio_spapr_pci_eeh_release(vdev->pdev);
+#if IS_ENABLED(CONFIG_EEH)
+	eeh_dev_release(vdev->pdev);
+#endif
 	vfio_pci_core_disable(vdev);
 
 	mutex_lock(&vdev->igate);
···
 void vfio_pci_core_finish_enable(struct vfio_pci_core_device *vdev)
 {
 	vfio_pci_probe_mmaps(vdev);
-	vfio_spapr_pci_eeh_open(vdev->pdev);
+#if IS_ENABLED(CONFIG_EEH)
+	eeh_dev_open(vdev->pdev);
+#endif
 
 	if (vdev->sriov_pf_core_dev) {
 		mutex_lock(&vdev->sriov_pf_core_dev->vf_token->lock);
···
 	mutex_destroy(&vdev->vma_lock);
 	kfree(vdev->region);
 	kfree(vdev->pm_save);
-	vfio_free_device(core_vdev);
 }
 EXPORT_SYMBOL_GPL(vfio_pci_core_release_dev);
 
···
 
 	if (vdev->vdev.mig_ops) {
 		if (!(vdev->vdev.mig_ops->migration_get_state &&
-		      vdev->vdev.mig_ops->migration_set_state) ||
+		      vdev->vdev.mig_ops->migration_set_state &&
+		      vdev->vdev.mig_ops->migration_get_data_size) ||
 		    !(vdev->vdev.migration_flags & VFIO_MIGRATION_STOP_COPY))
 			return -EINVAL;
 	}

drivers/vfio/platform/vfio_amba.c

···
 
 	vfio_platform_release_common(vdev);
 	kfree(vdev->name);
-	vfio_free_device(core_vdev);
 }
 
 static void vfio_amba_remove(struct amba_device *adev)

drivers/vfio/platform/vfio_platform.c

···
 		container_of(core_vdev, struct vfio_platform_device, vdev);
 
 	vfio_platform_release_common(vdev);
-	vfio_free_device(core_vdev);
 }
 
 static int vfio_platform_remove(struct platform_device *pdev)

drivers/vfio/platform/vfio_platform_common.c

···
 				  const char **extra_dbg)
 {
 #ifdef CONFIG_ACPI
-	struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
 	struct device *dev = vdev->device;
 	acpi_handle handle = ACPI_HANDLE(dev);
 	acpi_status acpi_ret;
 
-	acpi_ret = acpi_evaluate_object(handle, "_RST", NULL, &buffer);
+	acpi_ret = acpi_evaluate_object(handle, "_RST", NULL, NULL);
 	if (ACPI_FAILURE(acpi_ret)) {
 		if (extra_dbg)
 			*extra_dbg = acpi_format_exception(acpi_ret);

drivers/vfio/vfio.h

···
 }
 #endif
 
+#if IS_ENABLED(CONFIG_VFIO_VIRQFD)
+int __init vfio_virqfd_init(void);
+void vfio_virqfd_exit(void);
+#else
+static inline int __init vfio_virqfd_init(void)
+{
+	return 0;
+}
+static inline void vfio_virqfd_exit(void)
+{
+}
+#endif
+
 #ifdef CONFIG_VFIO_NOIOMMU
 extern bool vfio_noiommu __read_mostly;
 #else

drivers/vfio/vfio_iommu_spapr_tce.c

···
  *
  * Copyright (C) 2013 IBM Corp.  All rights reserved.
  *     Author: Alexey Kardashevskiy <[e-mail]>
+ * Copyright Gavin Shan, IBM Corporation 2014.
  *
  * Derived from original vfio_iommu_type1.c:
  * Copyright (C) 2012 Red Hat, Inc.  All rights reserved.
···
 	return ret;
 }
 
+static long vfio_spapr_ioctl_eeh_pe_op(struct iommu_group *group,
+				       unsigned long arg)
+{
+	struct eeh_pe *pe;
+	struct vfio_eeh_pe_op op;
+	unsigned long minsz;
+
+	pe = eeh_iommu_group_to_pe(group);
+	if (!pe)
+		return -ENODEV;
+
+	minsz = offsetofend(struct vfio_eeh_pe_op, op);
+	if (copy_from_user(&op, (void __user *)arg, minsz))
+		return -EFAULT;
+	if (op.argsz < minsz || op.flags)
+		return -EINVAL;
+
+	switch (op.op) {
+	case VFIO_EEH_PE_DISABLE:
+		return eeh_pe_set_option(pe, EEH_OPT_DISABLE);
+	case VFIO_EEH_PE_ENABLE:
+		return eeh_pe_set_option(pe, EEH_OPT_ENABLE);
+	case VFIO_EEH_PE_UNFREEZE_IO:
+		return eeh_pe_set_option(pe, EEH_OPT_THAW_MMIO);
+	case VFIO_EEH_PE_UNFREEZE_DMA:
+		return eeh_pe_set_option(pe, EEH_OPT_THAW_DMA);
+	case VFIO_EEH_PE_GET_STATE:
+		return eeh_pe_get_state(pe);
+		break;
+	case VFIO_EEH_PE_RESET_DEACTIVATE:
+		return eeh_pe_reset(pe, EEH_RESET_DEACTIVATE, true);
+	case VFIO_EEH_PE_RESET_HOT:
+		return eeh_pe_reset(pe, EEH_RESET_HOT, true);
+	case VFIO_EEH_PE_RESET_FUNDAMENTAL:
+		return eeh_pe_reset(pe, EEH_RESET_FUNDAMENTAL, true);
+	case VFIO_EEH_PE_CONFIGURE:
+		return eeh_pe_configure(pe);
+	case VFIO_EEH_PE_INJECT_ERR:
+		minsz = offsetofend(struct vfio_eeh_pe_op, err.mask);
+		if (op.argsz < minsz)
+			return -EINVAL;
+		if (copy_from_user(&op, (void __user *)arg, minsz))
+			return -EFAULT;
+
+		return eeh_pe_inject_err(pe, op.err.type, op.err.func,
+					 op.err.addr, op.err.mask);
+	default:
+		return -EINVAL;
+	}
+}
+
 static long tce_iommu_ioctl(void *iommu_data,
 				 unsigned int cmd, unsigned long arg)
 {
···
 		switch (arg) {
 		case VFIO_SPAPR_TCE_IOMMU:
 		case VFIO_SPAPR_TCE_v2_IOMMU:
-			ret = 1;
-			break;
+			return 1;
+		case VFIO_EEH:
+			return eeh_enabled();
 		default:
-			ret = vfio_spapr_iommu_eeh_ioctl(NULL, cmd, arg);
-			break;
+			return 0;
 		}
-
-		return (ret < 0) ? 0 : ret;
 	}
 
 	/*
···
 
 		ret = 0;
 		list_for_each_entry(tcegrp, &container->group_list, next) {
-			ret = vfio_spapr_iommu_eeh_ioctl(tcegrp->grp,
-					cmd, arg);
+			ret = vfio_spapr_ioctl_eeh_pe_op(tcegrp->grp, arg);
 			if (ret)
 				return ret;
 		}

drivers/vfio/vfio_main.c

···
 	vfio_release_device_set(device);
 	ida_free(&vfio.device_ida, device->index);
 
-	/*
-	 * kvfree() cannot be done here due to a life cycle mess in
-	 * vfio-ccw. Before the ccw part is fixed all drivers are
-	 * required to support @release and call vfio_free_device()
-	 * from there.
-	 */
-	device->ops->release(device);
+	if (device->ops->release)
+		device->ops->release(device);
+
+	kvfree(device);
 }
+
+static int vfio_init_device(struct vfio_device *device, struct device *dev,
+			    const struct vfio_device_ops *ops);
 
 /*
  * Allocate and initialize vfio_device so it can be registered to vfio
···
 
 /*
  * Initialize a vfio_device so it can be registered to vfio core.
- *
- * Only vfio-ccw driver should call this interface.
  */
-int vfio_init_device(struct vfio_device *device, struct device *dev,
-		     const struct vfio_device_ops *ops)
+static int vfio_init_device(struct vfio_device *device, struct device *dev,
+			    const struct vfio_device_ops *ops)
 {
 	int ret;
 
···
 	ida_free(&vfio.device_ida, device->index);
 	return ret;
 }
-EXPORT_SYMBOL_GPL(vfio_init_device);
-
-/*
- * The helper called by driver @release callback to free the device
- * structure. Drivers which don't have private data to clean can
- * simply use this helper as its @release.
- */
-void vfio_free_device(struct vfio_device *device)
-{
-	kvfree(device);
-}
-EXPORT_SYMBOL_GPL(vfio_free_device);
 
 static int __vfio_register_dev(struct vfio_device *device,
 			       enum vfio_group_type type)
···
 			    enum vfio_device_mig_state new_fsm,
 			    enum vfio_device_mig_state *next_fsm)
 {
-	enum { VFIO_DEVICE_NUM_STATES = VFIO_DEVICE_STATE_RUNNING_P2P + 1 };
+	enum { VFIO_DEVICE_NUM_STATES = VFIO_DEVICE_STATE_PRE_COPY_P2P + 1 };
 	/*
 	 * The coding in this table requires the driver to implement the
 	 * following FSM arcs:
···
 	 *         RUNNING_P2P -> RUNNING
 	 *         RUNNING_P2P -> STOP
 	 *         STOP -> RUNNING_P2P
-	 * Without P2P the driver must implement:
+	 *
+	 * If precopy is supported then the driver must support these additional
+	 * FSM arcs:
+	 *         RUNNING -> PRE_COPY
+	 *         PRE_COPY -> RUNNING
+	 *         PRE_COPY -> STOP_COPY
+	 * However, if precopy and P2P are supported together then the driver
+	 * must support these additional arcs beyond the P2P arcs above:
+	 *         PRE_COPY -> RUNNING
+	 *         PRE_COPY -> PRE_COPY_P2P
+	 *         PRE_COPY_P2P -> PRE_COPY
+	 *         PRE_COPY_P2P -> RUNNING_P2P
+	 *         PRE_COPY_P2P -> STOP_COPY
+	 *         RUNNING -> PRE_COPY
+	 *         RUNNING_P2P -> PRE_COPY_P2P
+	 *
+	 * Without P2P and precopy the driver must implement:
 	 *         RUNNING -> STOP
 	 *         STOP -> RUNNING
 	 *
 	 * The coding will step through multiple states for some combination
 	 * transitions; if all optional features are supported, this means the
 	 * following ones:
+	 *         PRE_COPY -> PRE_COPY_P2P -> STOP_COPY
+	 *         PRE_COPY -> RUNNING -> RUNNING_P2P
+	 *         PRE_COPY -> RUNNING -> RUNNING_P2P -> STOP
+	 *         PRE_COPY -> RUNNING -> RUNNING_P2P -> STOP -> RESUMING
+	 *         PRE_COPY_P2P -> RUNNING_P2P -> RUNNING
+	 *         PRE_COPY_P2P -> RUNNING_P2P -> STOP
+	 *         PRE_COPY_P2P -> RUNNING_P2P -> STOP -> RESUMING
 	 *         RESUMING -> STOP -> RUNNING_P2P
+	 *         RESUMING -> STOP -> RUNNING_P2P -> PRE_COPY_P2P
 	 *         RESUMING -> STOP -> RUNNING_P2P -> RUNNING
+	 *         RESUMING -> STOP -> RUNNING_P2P -> RUNNING -> PRE_COPY
 	 *         RESUMING -> STOP -> STOP_COPY
+	 *         RUNNING -> RUNNING_P2P -> PRE_COPY_P2P
 	 *         RUNNING -> RUNNING_P2P -> STOP
 	 *         RUNNING -> RUNNING_P2P -> STOP -> RESUMING
 	 *         RUNNING -> RUNNING_P2P -> STOP -> STOP_COPY
+	 *         RUNNING_P2P -> RUNNING -> PRE_COPY
 	 *         RUNNING_P2P -> STOP -> RESUMING
 	 *         RUNNING_P2P -> STOP -> STOP_COPY
+	 *         STOP -> RUNNING_P2P -> PRE_COPY_P2P
 	 *         STOP -> RUNNING_P2P -> RUNNING
+	 *         STOP -> RUNNING_P2P -> RUNNING -> PRE_COPY
 	 *         STOP_COPY -> STOP -> RESUMING
 	 *         STOP_COPY -> STOP -> RUNNING_P2P
 	 *         STOP_COPY -> STOP -> RUNNING_P2P -> RUNNING
+	 *
+	 *  The following transitions are blocked:
+	 *         STOP_COPY -> PRE_COPY
+	 *         STOP_COPY -> PRE_COPY_P2P
 	 */
 	static const u8 vfio_from_fsm_table[VFIO_DEVICE_NUM_STATES][VFIO_DEVICE_NUM_STATES] = {
 		[VFIO_DEVICE_STATE_STOP] = {
 			[VFIO_DEVICE_STATE_STOP] = VFIO_DEVICE_STATE_STOP,
 			[VFIO_DEVICE_STATE_RUNNING] = VFIO_DEVICE_STATE_RUNNING_P2P,
+			[VFIO_DEVICE_STATE_PRE_COPY] = VFIO_DEVICE_STATE_RUNNING_P2P,
+			[VFIO_DEVICE_STATE_PRE_COPY_P2P] = VFIO_DEVICE_STATE_RUNNING_P2P,
 			[VFIO_DEVICE_STATE_STOP_COPY] = VFIO_DEVICE_STATE_STOP_COPY,
 			[VFIO_DEVICE_STATE_RESUMING] = VFIO_DEVICE_STATE_RESUMING,
 			[VFIO_DEVICE_STATE_RUNNING_P2P] = VFIO_DEVICE_STATE_RUNNING_P2P,
···
 		[VFIO_DEVICE_STATE_RUNNING] = {
 			[VFIO_DEVICE_STATE_STOP] = VFIO_DEVICE_STATE_RUNNING_P2P,
 			[VFIO_DEVICE_STATE_RUNNING] = VFIO_DEVICE_STATE_RUNNING,
+			[VFIO_DEVICE_STATE_PRE_COPY] = VFIO_DEVICE_STATE_PRE_COPY,
+			[VFIO_DEVICE_STATE_PRE_COPY_P2P] = VFIO_DEVICE_STATE_RUNNING_P2P,
 			[VFIO_DEVICE_STATE_STOP_COPY] = VFIO_DEVICE_STATE_RUNNING_P2P,
+			[VFIO_DEVICE_STATE_RESUMING] = VFIO_DEVICE_STATE_RUNNING_P2P,
+			[VFIO_DEVICE_STATE_RUNNING_P2P] = VFIO_DEVICE_STATE_RUNNING_P2P,
+			[VFIO_DEVICE_STATE_ERROR] = VFIO_DEVICE_STATE_ERROR,
+		},
+		[VFIO_DEVICE_STATE_PRE_COPY] = {
+			[VFIO_DEVICE_STATE_STOP] = VFIO_DEVICE_STATE_RUNNING,
+			[VFIO_DEVICE_STATE_RUNNING] = VFIO_DEVICE_STATE_RUNNING,
+			[VFIO_DEVICE_STATE_PRE_COPY] = VFIO_DEVICE_STATE_PRE_COPY,
+			[VFIO_DEVICE_STATE_PRE_COPY_P2P] = VFIO_DEVICE_STATE_PRE_COPY_P2P,
+			[VFIO_DEVICE_STATE_STOP_COPY] = VFIO_DEVICE_STATE_PRE_COPY_P2P,
+			[VFIO_DEVICE_STATE_RESUMING] = VFIO_DEVICE_STATE_RUNNING,
+			[VFIO_DEVICE_STATE_RUNNING_P2P] = VFIO_DEVICE_STATE_RUNNING,
+			[VFIO_DEVICE_STATE_ERROR] = VFIO_DEVICE_STATE_ERROR,
+		},
+		[VFIO_DEVICE_STATE_PRE_COPY_P2P] = {
+			[VFIO_DEVICE_STATE_STOP] = VFIO_DEVICE_STATE_RUNNING_P2P,
+			[VFIO_DEVICE_STATE_RUNNING] = VFIO_DEVICE_STATE_RUNNING_P2P,
+			[VFIO_DEVICE_STATE_PRE_COPY] = VFIO_DEVICE_STATE_PRE_COPY,
+			[VFIO_DEVICE_STATE_PRE_COPY_P2P] = VFIO_DEVICE_STATE_PRE_COPY_P2P,
+			[VFIO_DEVICE_STATE_STOP_COPY] = VFIO_DEVICE_STATE_STOP_COPY,
 			[VFIO_DEVICE_STATE_RESUMING] = VFIO_DEVICE_STATE_RUNNING_P2P,
 			[VFIO_DEVICE_STATE_RUNNING_P2P] = VFIO_DEVICE_STATE_RUNNING_P2P,
 			[VFIO_DEVICE_STATE_ERROR] = VFIO_DEVICE_STATE_ERROR,
···
 		[VFIO_DEVICE_STATE_STOP_COPY] = {
 			[VFIO_DEVICE_STATE_STOP] = VFIO_DEVICE_STATE_STOP,
 			[VFIO_DEVICE_STATE_RUNNING] = VFIO_DEVICE_STATE_STOP,
+			[VFIO_DEVICE_STATE_PRE_COPY] = VFIO_DEVICE_STATE_ERROR,
+			[VFIO_DEVICE_STATE_PRE_COPY_P2P] = VFIO_DEVICE_STATE_ERROR,
 			[VFIO_DEVICE_STATE_STOP_COPY] = VFIO_DEVICE_STATE_STOP_COPY,
 			[VFIO_DEVICE_STATE_RESUMING] = VFIO_DEVICE_STATE_STOP,
 			[VFIO_DEVICE_STATE_RUNNING_P2P] = VFIO_DEVICE_STATE_STOP,
···
 		[VFIO_DEVICE_STATE_RESUMING] = {
 			[VFIO_DEVICE_STATE_STOP] = VFIO_DEVICE_STATE_STOP,
 			[VFIO_DEVICE_STATE_RUNNING] = VFIO_DEVICE_STATE_STOP,
+			[VFIO_DEVICE_STATE_PRE_COPY] = VFIO_DEVICE_STATE_STOP,
+			[VFIO_DEVICE_STATE_PRE_COPY_P2P] = VFIO_DEVICE_STATE_STOP,
 			[VFIO_DEVICE_STATE_STOP_COPY] = VFIO_DEVICE_STATE_STOP,
 			[VFIO_DEVICE_STATE_RESUMING] = VFIO_DEVICE_STATE_RESUMING,
 			[VFIO_DEVICE_STATE_RUNNING_P2P] = VFIO_DEVICE_STATE_STOP,
···
 		[VFIO_DEVICE_STATE_RUNNING_P2P] = {
 			[VFIO_DEVICE_STATE_STOP] = VFIO_DEVICE_STATE_STOP,
 			[VFIO_DEVICE_STATE_RUNNING] = VFIO_DEVICE_STATE_RUNNING,
+			[VFIO_DEVICE_STATE_PRE_COPY] = VFIO_DEVICE_STATE_RUNNING,
+			[VFIO_DEVICE_STATE_PRE_COPY_P2P] = VFIO_DEVICE_STATE_PRE_COPY_P2P,
 			[VFIO_DEVICE_STATE_STOP_COPY] = VFIO_DEVICE_STATE_STOP,
 			[VFIO_DEVICE_STATE_RESUMING] = VFIO_DEVICE_STATE_STOP,
 			[VFIO_DEVICE_STATE_RUNNING_P2P] = VFIO_DEVICE_STATE_RUNNING_P2P,
···
 		[VFIO_DEVICE_STATE_ERROR] = {
 			[VFIO_DEVICE_STATE_STOP] = VFIO_DEVICE_STATE_ERROR,
 			[VFIO_DEVICE_STATE_RUNNING] = VFIO_DEVICE_STATE_ERROR,
+			[VFIO_DEVICE_STATE_PRE_COPY] = VFIO_DEVICE_STATE_ERROR,
+			[VFIO_DEVICE_STATE_PRE_COPY_P2P] = VFIO_DEVICE_STATE_ERROR,
 			[VFIO_DEVICE_STATE_STOP_COPY] = VFIO_DEVICE_STATE_ERROR,
 			[VFIO_DEVICE_STATE_RESUMING] = VFIO_DEVICE_STATE_ERROR,
 			[VFIO_DEVICE_STATE_RUNNING_P2P] = VFIO_DEVICE_STATE_ERROR,
···
 	static const unsigned int state_flags_table[VFIO_DEVICE_NUM_STATES] = {
 		[VFIO_DEVICE_STATE_STOP] = VFIO_MIGRATION_STOP_COPY,
 		[VFIO_DEVICE_STATE_RUNNING] = VFIO_MIGRATION_STOP_COPY,
+		[VFIO_DEVICE_STATE_PRE_COPY] =
+			VFIO_MIGRATION_STOP_COPY | VFIO_MIGRATION_PRE_COPY,
+		[VFIO_DEVICE_STATE_PRE_COPY_P2P] = VFIO_MIGRATION_STOP_COPY |
+						   VFIO_MIGRATION_P2P |
+						   VFIO_MIGRATION_PRE_COPY,
 		[VFIO_DEVICE_STATE_STOP_COPY] = VFIO_MIGRATION_STOP_COPY,
 		[VFIO_DEVICE_STATE_RESUMING] = VFIO_MIGRATION_STOP_COPY,
 		[VFIO_DEVICE_STATE_RUNNING_P2P] =
···
 		return -EFAULT;
 	if (IS_ERR(filp))
 		return PTR_ERR(filp);
+	return 0;
+}
+
+static int
+vfio_ioctl_device_feature_migration_data_size(struct vfio_device *device,
+					      u32 flags, void __user *arg,
+					      size_t argsz)
+{
+	struct vfio_device_feature_mig_data_size data_size = {};
+	unsigned long stop_copy_length;
+	int ret;
+
+	if (!device->mig_ops)
+		return -ENOTTY;
+
+	ret = vfio_check_feature(flags, argsz, VFIO_DEVICE_FEATURE_GET,
+				 sizeof(data_size));
+	if (ret != 1)
+		return ret;
+
+	ret = device->mig_ops->migration_get_data_size(device, &stop_copy_length);
+	if (ret)
+		return ret;
+
+	data_size.stop_copy_length = stop_copy_length;
+	if (copy_to_user(arg, &data_size, sizeof(data_size)))
+		return -EFAULT;
+
 	return 0;
 }
 
···
 			feature.argsz - minsz);
 	case VFIO_DEVICE_FEATURE_DMA_LOGGING_REPORT:
 		return vfio_ioctl_device_feature_logging_report(
+			device, feature.flags, arg->data,
+			feature.argsz - minsz);
+	case VFIO_DEVICE_FEATURE_MIG_DATA_SIZE:
+		return vfio_ioctl_device_feature_migration_data_size(
 			device, feature.flags, arg->data,
 			feature.argsz - minsz);
 	default:
···
 	if (ret)
 		return ret;
 
+	ret = vfio_virqfd_init();
+	if (ret)
+		goto err_virqfd;
+
 	/* /sys/class/vfio-dev/vfioX */
 	vfio.device_class = class_create(THIS_MODULE, "vfio-dev");
 	if (IS_ERR(vfio.device_class)) {
···
 	return 0;
 
 err_dev_class:
+	vfio_virqfd_exit();
+err_virqfd:
 	vfio_group_cleanup();
 	return ret;
 }
···
 	ida_destroy(&vfio.device_ida);
 	class_destroy(vfio.device_class);
 	vfio.device_class = NULL;
+	vfio_virqfd_exit();
 	vfio_group_cleanup();
 	xa_destroy(&vfio_device_set_xa);
 }

drivers/vfio/vfio_spapr_eeh.c

···
-// SPDX-License-Identifier: GPL-2.0-only
-/*
- * EEH functionality support for VFIO devices. The feature is only
- * available on sPAPR compatible platforms.
- *
- * Copyright Gavin Shan, IBM Corporation 2014.
- */
-
-#include <linux/module.h>
-#include <linux/uaccess.h>
-#include <linux/vfio.h>
-#include <asm/eeh.h>
-
-#define DRIVER_VERSION	"0.1"
-#define DRIVER_AUTHOR	"Gavin Shan, IBM Corporation"
-#define DRIVER_DESC	"VFIO IOMMU SPAPR EEH"
-
-/* We might build address mapping here for "fast" path later */
-void vfio_spapr_pci_eeh_open(struct pci_dev *pdev)
-{
-	eeh_dev_open(pdev);
-}
-EXPORT_SYMBOL_GPL(vfio_spapr_pci_eeh_open);
-
-void vfio_spapr_pci_eeh_release(struct pci_dev *pdev)
-{
-	eeh_dev_release(pdev);
-}
-EXPORT_SYMBOL_GPL(vfio_spapr_pci_eeh_release);
-
-long vfio_spapr_iommu_eeh_ioctl(struct iommu_group *group,
-				unsigned int cmd, unsigned long arg)
-{
-	struct eeh_pe *pe;
-	struct vfio_eeh_pe_op op;
-	unsigned long minsz;
-	long ret = -EINVAL;
-
-	switch (cmd) {
-	case VFIO_CHECK_EXTENSION:
-		if (arg == VFIO_EEH)
-			ret = eeh_enabled() ? 1 : 0;
-		else
-			ret = 0;
-		break;
-	case VFIO_EEH_PE_OP:
-		pe = eeh_iommu_group_to_pe(group);
-		if (!pe)
-			return -ENODEV;
-
-		minsz = offsetofend(struct vfio_eeh_pe_op, op);
-		if (copy_from_user(&op, (void __user *)arg, minsz))
-			return -EFAULT;
-		if (op.argsz < minsz || op.flags)
-			return -EINVAL;
-
-		switch (op.op) {
-		case VFIO_EEH_PE_DISABLE:
-			ret = eeh_pe_set_option(pe, EEH_OPT_DISABLE);
-			break;
-		case VFIO_EEH_PE_ENABLE:
-			ret = eeh_pe_set_option(pe, EEH_OPT_ENABLE);
-			break;
-		case VFIO_EEH_PE_UNFREEZE_IO:
-			ret = eeh_pe_set_option(pe, EEH_OPT_THAW_MMIO);
-			break;
-		case VFIO_EEH_PE_UNFREEZE_DMA:
-			ret = eeh_pe_set_option(pe, EEH_OPT_THAW_DMA);
-			break;
-		case VFIO_EEH_PE_GET_STATE:
-			ret = eeh_pe_get_state(pe);
-			break;
-		case VFIO_EEH_PE_RESET_DEACTIVATE:
-			ret = eeh_pe_reset(pe, EEH_RESET_DEACTIVATE, true);
-			break;
-		case VFIO_EEH_PE_RESET_HOT:
-			ret = eeh_pe_reset(pe, EEH_RESET_HOT, true);
-			break;
-		case VFIO_EEH_PE_RESET_FUNDAMENTAL:
-			ret = eeh_pe_reset(pe, EEH_RESET_FUNDAMENTAL, true);
-			break;
-		case VFIO_EEH_PE_CONFIGURE:
-			ret = eeh_pe_configure(pe);
-			break;
-		case VFIO_EEH_PE_INJECT_ERR:
-			minsz = offsetofend(struct vfio_eeh_pe_op, err.mask);
-			if (op.argsz < minsz)
-				return -EINVAL;
-			if (copy_from_user(&op, (void __user *)arg, minsz))
-				return -EFAULT;
-
-			ret = eeh_pe_inject_err(pe, op.err.type, op.err.func,
-						op.err.addr, op.err.mask);
-			break;
-		default:
-			ret = -EINVAL;
-		}
-	}
-
-	return ret;
-}
-EXPORT_SYMBOL_GPL(vfio_spapr_iommu_eeh_ioctl);
-
-MODULE_VERSION(DRIVER_VERSION);
-MODULE_LICENSE("GPL v2");
-MODULE_AUTHOR(DRIVER_AUTHOR);
-MODULE_DESCRIPTION(DRIVER_DESC);

drivers/vfio/virqfd.c

···
 #include <linux/file.h>
 #include <linux/module.h>
 #include <linux/slab.h>
-
-#define DRIVER_VERSION  "0.1"
-#define DRIVER_AUTHOR   "Alex Williamson <[e-mail]>"
-#define DRIVER_DESC     "IRQFD support for VFIO bus drivers"
+#include "vfio.h"
 
 static struct workqueue_struct *vfio_irqfd_cleanup_wq;
 static DEFINE_SPINLOCK(virqfd_lock);
 
-static int __init vfio_virqfd_init(void)
+int __init vfio_virqfd_init(void)
 {
 	vfio_irqfd_cleanup_wq =
 		create_singlethread_workqueue("vfio-irqfd-cleanup");
···
 	return 0;
 }
 
-static void __exit vfio_virqfd_exit(void)
+void vfio_virqfd_exit(void)
 {
 	destroy_workqueue(vfio_irqfd_cleanup_wq);
 }
···
 	flush_workqueue(vfio_irqfd_cleanup_wq);
 }
 EXPORT_SYMBOL_GPL(vfio_virqfd_disable);
-
-module_init(vfio_virqfd_init);
-module_exit(vfio_virqfd_exit);
-
-MODULE_VERSION(DRIVER_VERSION);
-MODULE_LICENSE("GPL v2");
-MODULE_AUTHOR(DRIVER_AUTHOR);
-MODULE_DESCRIPTION(DRIVER_DESC);

include/linux/mlx5/mlx5_ifc.h

···
 	u8	   max_reformat_remove_size[0x8];
 	u8	   max_reformat_remove_offset[0x8];
 
-	u8	   reserved_at_c0[0xe0];
+	u8	   reserved_at_c0[0x8];
+	u8	   migration_multi_load[0x1];
+	u8	   migration_tracking_state[0x1];
+	u8	   reserved_at_ca[0x16];
+
+	u8	   reserved_at_e0[0xc0];
 
 	u8	   reserved_at_1a0[0xb];
 	u8	   log_min_mkey_entity_size[0x5];
···
 	u8         reserved_at_20[0x10];
 	u8         op_mod[0x10];
 
-	u8         reserved_at_40[0x10];
+	u8         incremental[0x1];
+	u8         reserved_at_41[0xf];
 	u8         vhca_id[0x10];
 
 	u8         reserved_at_60[0x20];
···
 	u8         reserved_at_20[0x10];
 	u8         op_mod[0x10];
 
-	u8         reserved_at_40[0x10];
+	u8         incremental[0x1];
+	u8         set_track[0x1];
+	u8         reserved_at_42[0xe];
 	u8         vhca_id[0x10];
 
 	u8         reserved_at_60[0x20];

include/linux/vfio.h

···
  * @migration_get_state: Optional callback to get the migration state for
  *         devices that support migration. It's mandatory for
  *         VFIO_DEVICE_FEATURE_MIGRATION migration support.
+ * @migration_get_data_size: Optional callback to get the estimated data
+ *          length that will be required to complete stop copy. It's mandatory for
+ *          VFIO_DEVICE_FEATURE_MIGRATION migration support.
  */
 struct vfio_migration_ops {
 	struct file *(*migration_set_state)(
···
 		enum vfio_device_mig_state new_state);
 	int (*migration_get_state)(struct vfio_device *device,
 				   enum vfio_device_mig_state *curr_state);
+	int (*migration_get_data_size)(struct vfio_device *device,
+				       unsigned long *stop_copy_length);
 };
 
 /**
···
 					dev, ops),				\
 		     struct dev_struct, member)
 
-int vfio_init_device(struct vfio_device *device, struct device *dev,
-		     const struct vfio_device_ops *ops);
-void vfio_free_device(struct vfio_device *device);
 static inline void vfio_put_device(struct vfio_device *device)
 {
 	put_device(&device->device);
···
 int vfio_set_irqs_validate_and_prepare(struct vfio_irq_set *hdr,
 				       int num_irqs, int max_irq_type,
 				       size_t *data_size);
-
-struct pci_dev;
-#if IS_ENABLED(CONFIG_VFIO_SPAPR_EEH)
-void vfio_spapr_pci_eeh_open(struct pci_dev *pdev);
-void vfio_spapr_pci_eeh_release(struct pci_dev *pdev);
-long vfio_spapr_iommu_eeh_ioctl(struct iommu_group *group, unsigned int cmd,
-				unsigned long arg);
-#else
-static inline void vfio_spapr_pci_eeh_open(struct pci_dev *pdev)
-{
-}
-
-static inline void vfio_spapr_pci_eeh_release(struct pci_dev *pdev)
-{
-}
-
-static inline long vfio_spapr_iommu_eeh_ioctl(struct iommu_group *group,
-					      unsigned int cmd,
-					      unsigned long arg)
-{
-	return -ENOTTY;
-}
-#endif /* CONFIG_VFIO_SPAPR_EEH */
 
 /*
  * IRQfd - generic

include/uapi/linux/vfio.h

···
  * VFIO_MIGRATION_STOP_COPY | VFIO_MIGRATION_P2P means that RUNNING_P2P
  * is supported in addition to the STOP_COPY states.
  *
+ * VFIO_MIGRATION_STOP_COPY | VFIO_MIGRATION_PRE_COPY means that
+ * PRE_COPY is supported in addition to the STOP_COPY states.
+ *
+ * VFIO_MIGRATION_STOP_COPY | VFIO_MIGRATION_P2P | VFIO_MIGRATION_PRE_COPY
+ * means that RUNNING_P2P, PRE_COPY and PRE_COPY_P2P are supported
+ * in addition to the STOP_COPY states.
+ *
  * Other combinations of flags have behavior to be defined in the future.
  */
 struct vfio_device_feature_migration {
 	__aligned_u64 flags;
 #define VFIO_MIGRATION_STOP_COPY	(1 << 0)
 #define VFIO_MIGRATION_P2P		(1 << 1)
+#define VFIO_MIGRATION_PRE_COPY		(1 << 2)
 };
 #define VFIO_DEVICE_FEATURE_MIGRATION 1
 
···
  *  RESUMING - The device is stopped and is loading a new internal state
  *  ERROR - The device has failed and must be reset
  *
- * And 1 optional state to support VFIO_MIGRATION_P2P:
+ * And optional states to support VFIO_MIGRATION_P2P:
  *  RUNNING_P2P - RUNNING, except the device cannot do peer to peer DMA
+ * And VFIO_MIGRATION_PRE_COPY:
+ *  PRE_COPY - The device is running normally but tracking internal state
+ *             changes
+ * And VFIO_MIGRATION_P2P | VFIO_MIGRATION_PRE_COPY:
+ *  PRE_COPY_P2P - PRE_COPY, except the device cannot do peer to peer DMA
  *
  * The FSM takes actions on the arcs between FSM states. The driver implements
  * the following behavior for the FSM arcs:
···
  *
  *   To abort a RESUMING session the device must be reset.
  *
+ * PRE_COPY -> RUNNING
  * RUNNING_P2P -> RUNNING
  *   While in RUNNING the device is fully operational, the device may generate
  *   interrupts, DMA, respond to MMIO, all vfio device regions are functional,
  *   and the device may advance its internal state.
  *
+ *   The PRE_COPY arc will terminate a data transfer session.
+ *
+ * PRE_COPY_P2P -> RUNNING_P2P
  * RUNNING -> RUNNING_P2P
  * STOP -> RUNNING_P2P
  *   While in RUNNING_P2P the device is partially running in the P2P quiescent
  *   state defined below.
  *
- * STOP -> STOP_COPY
- *   This arc begin the process of saving the device state and will return a
- *   new data_fd.
+ *   The PRE_COPY_P2P arc will terminate a data transfer session.
  *
+ * RUNNING -> PRE_COPY
+ * RUNNING_P2P -> PRE_COPY_P2P
+ * STOP -> STOP_COPY
+ *   PRE_COPY, PRE_COPY_P2P and STOP_COPY form the "saving group" of states
+ *   which share a data transfer session. Moving between these states alters
+ *   what is streamed in session, but does not terminate or otherwise affect
+ *   the associated fd.
+ *
+ *   These arcs begin the process of saving the device state and will return a
+ *   new data_fd. The migration driver may perform actions such as enabling
+ *   dirty logging of device state when entering PRE_COPY or PER_COPY_P2P.
+ *
+ *   Each arc does not change the device operation, the device remains
+ *   RUNNING, P2P quiesced or in STOP. The STOP_COPY state is described below
+ *   in PRE_COPY_P2P -> STOP_COPY.
+ *
+ * PRE_COPY -> PRE_COPY_P2P
+ *   Entering PRE_COPY_P2P continues all the behaviors of PRE_COPY above.
+ *   However, while in the PRE_COPY_P2P state, the device is partially running
+ *   in the P2P quiescent state defined below, like RUNNING_P2P.
+ *
+ * PRE_COPY_P2P -> PRE_COPY
+ *   This arc allows returning the device to a full RUNNING behavior while
+ *   continuing all the behaviors of PRE_COPY.
+ *
+ * PRE_COPY_P2P -> STOP_COPY
  *   While in the STOP_COPY state the device has the same behavior as STOP
  *   with the addition that the data transfers session continues to stream the
  *   migration state. End of stream on the FD indicates the entire device
···
  *   device state for this arc if required to prepare the device to receive the
  *   migration data.
  *
+ * STOP_COPY -> PRE_COPY
+ * STOP_COPY -> PRE_COPY_P2P
+ *   These arcs are not permitted and return error if requested. Future
+ *   revisions of this API may define behaviors for these arcs, in this case
+ *   support will be discoverable by a new flag in
+ *   VFIO_DEVICE_FEATURE_MIGRATION.
+ *
  * any -> ERROR
  *   ERROR cannot be specified as a device state, however any transition request
  *   can be failed with an errno return and may then move the device_state into
···
  * The optional peer to peer (P2P) quiescent state is intended to be a quiescent
  * state for the device for the purposes of managing multiple devices within a
  * user context where peer-to-peer DMA between devices may be active. The
- * RUNNING_P2P states must prevent the device from initiating
+ * RUNNING_P2P and PRE_COPY_P2P states must prevent the device from initiating
  * any new P2P DMA transactions. If the device can identify P2P transactions
  * then it can stop only P2P DMA, otherwise it must stop all DMA. The migration
  * driver must complete any such outstanding operations prior to completing the
···
  * above FSM arcs. As there are multiple paths through the FSM arcs the path
  * should be selected based on the following rules:
  *   - Select the shortest path.
+ *   - The path cannot have saving group states as interior arcs, only
+ *     starting/end states.
  * Refer to vfio_mig_get_next_state() for the result of the algorithm.
  *
  * The automatic transit through the FSM arcs that make up the combination
···
  * support them. The user can discover if these states are supported by using
  * VFIO_DEVICE_FEATURE_MIGRATION. By using combination transitions the user can
  * avoid knowing about these optional states if the kernel driver supports them.
+ *
+ * Arcs touching PRE_COPY and PRE_COPY_P2P are removed if support for PRE_COPY
+ * is not present.
  */
 enum vfio_device_mig_state {
 	VFIO_DEVICE_STATE_ERROR = 0,
···
 	VFIO_DEVICE_STATE_STOP_COPY = 3,
 	VFIO_DEVICE_STATE_RESUMING = 4,
 	VFIO_DEVICE_STATE_RUNNING_P2P = 5,
+	VFIO_DEVICE_STATE_PRE_COPY = 6,
+	VFIO_DEVICE_STATE_PRE_COPY_P2P = 7,
 };
+
+/**
+ * VFIO_MIG_GET_PRECOPY_INFO - _IO(VFIO_TYPE, VFIO_BASE + 21)
+ *
+ * This ioctl is used on the migration data FD in the precopy phase of the
+ * migration data transfer. It returns an estimate of the current data sizes
+ * remaining to be transferred. It allows the user to judge when it is
+ * appropriate to leave PRE_COPY for STOP_COPY.
+ *
+ * This ioctl is valid only in PRE_COPY states and kernel driver should
+ * return -EINVAL from any other migration state.
+ *
+ * The vfio_precopy_info data structure returned by this ioctl provides
+ * estimates of data available from the device during the PRE_COPY states.
+ * This estimate is split into two categories, initial_bytes and
+ * dirty_bytes.
+ *
+ * The initial_bytes field indicates the amount of initial precopy
+ * data available from the device. This field should have a non-zero initial
+ * value and decrease as migration data is read from the device.
+ * It is recommended to leave PRE_COPY for STOP_COPY only after this field
+ * reaches zero. Leaving PRE_COPY earlier might make things slower.
+ *
+ * The dirty_bytes field tracks device state changes relative to data
+ * previously retrieved.  This field starts at zero and may increase as
+ * the internal device state is modified or decrease as that modified
+ * state is read from the device.
+ *
+ * Userspace may use the combination of these fields to estimate the
+ * potential data size available during the PRE_COPY phases, as well as
+ * trends relative to the rate the device is dirtying its internal
+ * state, but these fields are not required to have any bearing relative
+ * to the data size available during the STOP_COPY phase.
+ *
+ * Drivers have a lot of flexibility in when and what they transfer during the
+ * PRE_COPY phase, and how they report this from VFIO_MIG_GET_PRECOPY_INFO.
+ *
+ * During pre-copy the migration data FD has a temporary "end of stream" that is
+ * reached when both initial_bytes and dirty_byte are zero. For instance, this
+ * may indicate that the device is idle and not currently dirtying any internal
+ * state. When read() is done on this temporary end of stream the kernel driver
+ * should return ENOMSG from read(). Userspace can wait for more data (which may
+ * never come) by using poll.
+ *
+ * Once in STOP_COPY the migration data FD has a permanent end of stream
+ * signaled in the usual way by read() always returning 0 and poll always
+ * returning readable. ENOMSG may not be returned in STOP_COPY.
+ * Support for this ioctl is mandatory if a driver claims to support
+ * VFIO_MIGRATION_PRE_COPY.
+ *
+ * Return: 0 on success, -1 and errno set on failure.
+ */
+struct vfio_precopy_info {
+	__u32 argsz;
+	__u32 flags;
+	__aligned_u64 initial_bytes;
+	__aligned_u64 dirty_bytes;
+};
+
+#define VFIO_MIG_GET_PRECOPY_INFO _IO(VFIO_TYPE, VFIO_BASE + 21)
 
 /*
  * Upon VFIO_DEVICE_FEATURE_SET, allow the device to be moved into a low power
···
 };
 
 #define VFIO_DEVICE_FEATURE_DMA_LOGGING_REPORT 8
+
+/*
+ * Upon VFIO_DEVICE_FEATURE_GET read back the estimated data length that will
+ * be required to complete stop copy.
+ *
+ * Note: Can be called on each device state.
+ */
+
+struct vfio_device_feature_mig_data_size {
+	__aligned_u64 stop_copy_length;
+};
+
+#define VFIO_DEVICE_FEATURE_MIG_DATA_SIZE 9
 
 /* -------- API for Type1 VFIO IOMMU -------- */
 

samples/vfio-mdev/mbochs.c

···
 	atomic_add(mdev_state->type->mbytes, &mbochs_avail_mbytes);
 	kfree(mdev_state->pages);
 	kfree(mdev_state->vconfig);
-	vfio_free_device(vdev);
 }
 
 static void mbochs_remove(struct mdev_device *mdev)
···
 
 	ret = device_register(&mbochs_dev);
 	if (ret)
-		goto err_class;
+		goto err_put;
 
 	ret = mdev_register_parent(&mbochs_parent, &mbochs_dev, &mbochs_driver,
 				   mbochs_mdev_types,
···
 	return 0;
 
 err_device:
-	device_unregister(&mbochs_dev);
-err_class:
+	device_del(&mbochs_dev);
+err_put:
+	put_device(&mbochs_dev);
 	class_destroy(mbochs_class);
 err_driver:
 	mdev_unregister_driver(&mbochs_driver);

samples/vfio-mdev/mdpy-fb.c

···
 
 	ret = pci_request_regions(pdev, "mdpy-fb");
 	if (ret < 0)
-		return ret;
+		goto err_disable_dev;
 
 	pci_read_config_dword(pdev, MDPY_FORMAT_OFFSET, &format);
 	pci_read_config_dword(pdev, MDPY_WIDTH_OFFSET,	&width);
···
 err_release_regions:
 	pci_release_regions(pdev);
 
+err_disable_dev:
+	pci_disable_device(pdev);
+
 	return ret;
 }
 
···
 	struct fb_info *info = pci_get_drvdata(pdev);
 
 	unregister_framebuffer(info);
+	iounmap(info->screen_base);
 	framebuffer_release(info);
+	pci_release_regions(pdev);
+	pci_disable_device(pdev);
 }
 
 static struct pci_device_id mdpy_fb_pci_table[] = {

samples/vfio-mdev/mdpy.c

···
 
 	vfree(mdev_state->memblk);
 	kfree(mdev_state->vconfig);
-	vfio_free_device(vdev);
 }
 
 static void mdpy_remove(struct mdev_device *mdev)
···
 
 	ret = device_register(&mdpy_dev);
 	if (ret)
-		goto err_class;
+		goto err_put;
 
 	ret = mdev_register_parent(&mdpy_parent, &mdpy_dev, &mdpy_driver,
 				   mdpy_mdev_types,
···
 	return 0;
 
 err_device:
-	device_unregister(&mdpy_dev);
-err_class:
+	device_del(&mdpy_dev);
+err_put:
+	put_device(&mdpy_dev);
 	class_destroy(mdpy_class);
 err_driver:
 	mdev_unregister_driver(&mdpy_driver);

samples/vfio-mdev/mtty.c

···
 
 	atomic_add(mdev_state->nr_ports, &mdev_avail_ports);
 	kfree(mdev_state->vconfig);
-	vfio_free_device(vdev);
 }
 
 static void mtty_remove(struct mdev_device *mdev)
···
 
 	ret = device_register(&mtty_dev.dev);
 	if (ret)
-		goto err_class;
+		goto err_put;
 
 	ret = mdev_register_parent(&mtty_dev.parent, &mtty_dev.dev,
 				   &mtty_driver, mtty_mdev_types,
···
 	return 0;
 
 err_device:
-	device_unregister(&mtty_dev.dev);
-err_class:
+	device_del(&mtty_dev.dev);
+err_put:
+	put_device(&mtty_dev.dev);
 	class_destroy(mtty_dev.vd_class);
 err_driver:
 	mdev_unregister_driver(&mtty_driver);