Merge tag 'vfio-v6.7-rc1' of https://github.com/awilliam/linux-vfio

Pull VFIO updates from Alex Williamson:

 - Add support for "chunk mode" in the mlx5-vfio-pci variant driver,
   which allows both larger device image sizes for migration, beyond the
   previous 4GB limit, and also read-ahead support for improved
   migration performance (Yishai Hadas)

 - A new bus master control interface for the CDX bus driver where there
   is no in-band mechanism to toggle device DMA as there is through
   config space on PCI devices (Nipun Gupta)

 - Add explicit alignment directives to vfio data structures to reduce
   the chance of breaking 32-bit userspace. In most cases this is
   transparent and the remaining cases where data structures are padded
   work within the existing rules for extending data structures within
   vfio (Stefan Hajnoczi)

 - Resolve a bug in the cdx bus driver noted when compiled with clang
   where missing parenthesis result in the wrong operation (Nathan
   Chancellor)

 - Resolve errors reported by smatch for a function when dealing with
   invalid inputs (Alex Williamson)

 - Add migration support to the mtty vfio/mdev sample driver for testing
   and integration purposes, allowing CI of migration without specific
   hardware requirements. Also resolve many of the short- comings of
   this driver relative to implementation of the vfio interrupt ioctl
   along the way (Alex Williamson)

* tag 'vfio-v6.7-rc1' of https://github.com/awilliam/linux-vfio:
  vfio/mtty: Enable migration support
  vfio/mtty: Overhaul mtty interrupt handling
  vfio: Fix smatch errors in vfio_combine_iova_ranges()
  vfio/cdx: Add parentheses between bitwise AND expression and logical NOT
  vfio/mlx5: Activate the chunk mode functionality
  vfio/mlx5: Add support for READING in chunk mode
  vfio/mlx5: Add support for SAVING in chunk mode
  vfio/mlx5: Pre-allocate chunks for the STOP_COPY phase
  vfio/mlx5: Rename some stuff to match chunk mode
  vfio/mlx5: Enable querying state size which is > 4GB
  vfio/mlx5: Refactor the SAVE callback to activate a work only upon an error
  vfio/mlx5: Wake up the reader post of disabling the SAVING migration file
  vfio: use __aligned_u64 in struct vfio_device_ioeventfd
  vfio: use __aligned_u64 in struct vfio_device_gfx_plane_info
  vfio: trivially use __aligned_u64 for ioctl structs
  vfio-cdx: add bus mastering device feature support
  vfio: add bus master feature to device feature ioctl
  cdx: add support for bus mastering

drivers/cdx/cdx.c

@@ -182,6 +182,38 @@ cdx_match_id(const struct cdx_device_id *ids, struct cdx_device *dev)
 	return NULL;
 }
 
+int cdx_set_master(struct cdx_device *cdx_dev)
+{
+	struct cdx_controller *cdx = cdx_dev->cdx;
+	struct cdx_device_config dev_config;
+	int ret = -EOPNOTSUPP;
+
+	dev_config.type = CDX_DEV_BUS_MASTER_CONF;
+	dev_config.bus_master_enable = true;
+	if (cdx->ops->dev_configure)
+		ret = cdx->ops->dev_configure(cdx, cdx_dev->bus_num,
+					      cdx_dev->dev_num, &dev_config);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(cdx_set_master);
+
+int cdx_clear_master(struct cdx_device *cdx_dev)
+{
+	struct cdx_controller *cdx = cdx_dev->cdx;
+	struct cdx_device_config dev_config;
+	int ret = -EOPNOTSUPP;
+
+	dev_config.type = CDX_DEV_BUS_MASTER_CONF;
+	dev_config.bus_master_enable = false;
+	if (cdx->ops->dev_configure)
+		ret = cdx->ops->dev_configure(cdx, cdx_dev->bus_num,
+					      cdx_dev->dev_num, &dev_config);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(cdx_clear_master);
+
 /**
  * cdx_bus_match - device to driver matching callback
  * @dev: the cdx device to match against

drivers/cdx/controller/cdx_controller.c

@@ -56,6 +56,10 @@ static int cdx_configure_device(struct cdx_controller *cdx,
 	case CDX_DEV_RESET_CONF:
 		ret = cdx_mcdi_reset_device(cdx->priv, bus_num, dev_num);
 		break;
+	case CDX_DEV_BUS_MASTER_CONF:
+		ret = cdx_mcdi_bus_master_enable(cdx->priv, bus_num, dev_num,
+						 dev_config->bus_master_enable);
+		break;
 	default:
 		ret = -EINVAL;
 	}

drivers/cdx/controller/mcdi_functions.c

@@ -137,3 +137,61 @@ int cdx_mcdi_reset_device(struct cdx_mcdi *cdx, u8 bus_num, u8 dev_num)
 
 	return ret;
 }
+
+static int cdx_mcdi_ctrl_flag_get(struct cdx_mcdi *cdx, u8 bus_num,
+				  u8 dev_num, u32 *flags)
+{
+	MCDI_DECLARE_BUF(inbuf, MC_CMD_CDX_DEVICE_CONTROL_GET_IN_LEN);
+	MCDI_DECLARE_BUF(outbuf, MC_CMD_CDX_DEVICE_CONTROL_GET_OUT_LEN);
+	size_t outlen;
+	int ret;
+
+	MCDI_SET_DWORD(inbuf, CDX_DEVICE_CONTROL_GET_IN_BUS, bus_num);
+	MCDI_SET_DWORD(inbuf, CDX_DEVICE_CONTROL_GET_IN_DEVICE, dev_num);
+	ret = cdx_mcdi_rpc(cdx, MC_CMD_CDX_DEVICE_CONTROL_GET, inbuf,
+			   sizeof(inbuf), outbuf, sizeof(outbuf), &outlen);
+	if (ret)
+		return ret;
+
+	if (outlen != MC_CMD_CDX_DEVICE_CONTROL_GET_OUT_LEN)
+		return -EIO;
+
+	*flags = MCDI_DWORD(outbuf, CDX_DEVICE_CONTROL_GET_OUT_FLAGS);
+
+	return 0;
+}
+
+static int cdx_mcdi_ctrl_flag_set(struct cdx_mcdi *cdx, u8 bus_num,
+				  u8 dev_num, bool enable, int bit_pos)
+{
+	MCDI_DECLARE_BUF(inbuf, MC_CMD_CDX_DEVICE_CONTROL_SET_IN_LEN);
+	u32 flags;
+	int ret;
+
+	/*
+	 * Get flags and then set/reset bit at bit_pos according to
+	 * the input params.
+	 */
+	ret = cdx_mcdi_ctrl_flag_get(cdx, bus_num, dev_num, &flags);
+	if (ret)
+		return ret;
+
+	flags = flags & (u32)(~(BIT(bit_pos)));
+	if (enable)
+		flags |= (1 << bit_pos);
+
+	MCDI_SET_DWORD(inbuf, CDX_DEVICE_CONTROL_SET_IN_BUS, bus_num);
+	MCDI_SET_DWORD(inbuf, CDX_DEVICE_CONTROL_SET_IN_DEVICE, dev_num);
+	MCDI_SET_DWORD(inbuf, CDX_DEVICE_CONTROL_SET_IN_FLAGS, flags);
+	ret = cdx_mcdi_rpc(cdx, MC_CMD_CDX_DEVICE_CONTROL_SET, inbuf,
+			   sizeof(inbuf), NULL, 0, NULL);
+
+	return ret;
+}
+
+int cdx_mcdi_bus_master_enable(struct cdx_mcdi *cdx, u8 bus_num,
+			       u8 dev_num, bool enable)
+{
+	return cdx_mcdi_ctrl_flag_set(cdx, bus_num, dev_num, enable,
+			MC_CMD_CDX_DEVICE_CONTROL_SET_IN_BUS_MASTER_ENABLE_LBN);
+}

drivers/cdx/controller/mcdi_functions.h

@@ -58,4 +58,17 @@ int cdx_mcdi_get_dev_config(struct cdx_mcdi *cdx,
 int cdx_mcdi_reset_device(struct cdx_mcdi *cdx,
 			  u8 bus_num, u8 dev_num);
 
+/**
+ * cdx_mcdi_bus_master_enable - Set/Reset bus mastering for cdx device
+ *				represented by bus_num:dev_num
+ * @cdx: pointer to MCDI interface.
+ * @bus_num: Bus number.
+ * @dev_num: Device number.
+ * @enable: Enable bus mastering if set, disable otherwise.
+ *
+ * Return: 0 on success, <0 on failure
+ */
+int cdx_mcdi_bus_master_enable(struct cdx_mcdi *cdx, u8 bus_num,
+			       u8 dev_num, bool enable);
+
 #endif /* CDX_MCDI_FUNCTIONS_H */

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

@@ -1379,7 +1379,7 @@ static long intel_vgpu_ioctl(struct vfio_device *vfio_dev, unsigned int cmd,
 		intel_gvt_reset_vgpu(vgpu);
 		return 0;
 	} else if (cmd == VFIO_DEVICE_QUERY_GFX_PLANE) {
-		struct vfio_device_gfx_plane_info dmabuf;
+		struct vfio_device_gfx_plane_info dmabuf = {};
 		int ret = 0;
 
 		minsz = offsetofend(struct vfio_device_gfx_plane_info,

drivers/vfio/cdx/main.c

@@ -14,7 +14,7 @@ static int vfio_cdx_open_device(struct vfio_device *core_vdev)
 		container_of(core_vdev, struct vfio_cdx_device, vdev);
 	struct cdx_device *cdx_dev = to_cdx_device(core_vdev->dev);
 	int count = cdx_dev->res_count;
-	int i;
+	int i, ret;
 
 	vdev->regions = kcalloc(count, sizeof(struct vfio_cdx_region),
 				GFP_KERNEL_ACCOUNT);
@@ -39,6 +39,17 @@ static int vfio_cdx_open_device(struct vfio_device *core_vdev)
 		if (!(cdx_dev->res[i].flags & IORESOURCE_READONLY))
 			vdev->regions[i].flags |= VFIO_REGION_INFO_FLAG_WRITE;
 	}
+	ret = cdx_dev_reset(core_vdev->dev);
+	if (ret) {
+		kfree(vdev->regions);
+		vdev->regions = NULL;
+		return ret;
+	}
+	ret = cdx_clear_master(cdx_dev);
+	if (ret)
+		vdev->flags &= ~BME_SUPPORT;
+	else
+		vdev->flags |= BME_SUPPORT;
 
 	return 0;
 }
@@ -52,6 +63,49 @@ static void vfio_cdx_close_device(struct vfio_device *core_vdev)
 	cdx_dev_reset(core_vdev->dev);
 }
 
+static int vfio_cdx_bm_ctrl(struct vfio_device *core_vdev, u32 flags,
+			    void __user *arg, size_t argsz)
+{
+	size_t minsz =
+		offsetofend(struct vfio_device_feature_bus_master, op);
+	struct vfio_cdx_device *vdev =
+		container_of(core_vdev, struct vfio_cdx_device, vdev);
+	struct cdx_device *cdx_dev = to_cdx_device(core_vdev->dev);
+	struct vfio_device_feature_bus_master ops;
+	int ret;
+
+	if (!(vdev->flags & BME_SUPPORT))
+		return -ENOTTY;
+
+	ret = vfio_check_feature(flags, argsz, VFIO_DEVICE_FEATURE_SET,
+				 sizeof(ops));
+	if (ret != 1)
+		return ret;
+
+	if (copy_from_user(&ops, arg, minsz))
+		return -EFAULT;
+
+	switch (ops.op) {
+	case VFIO_DEVICE_FEATURE_CLEAR_MASTER:
+		return cdx_clear_master(cdx_dev);
+	case VFIO_DEVICE_FEATURE_SET_MASTER:
+		return cdx_set_master(cdx_dev);
+	default:
+		return -EINVAL;
+	}
+}
+
+static int vfio_cdx_ioctl_feature(struct vfio_device *device, u32 flags,
+				  void __user *arg, size_t argsz)
+{
+	switch (flags & VFIO_DEVICE_FEATURE_MASK) {
+	case VFIO_DEVICE_FEATURE_BUS_MASTER:
+		return vfio_cdx_bm_ctrl(device, flags, arg, argsz);
+	default:
+		return -ENOTTY;
+	}
+}
+
 static int vfio_cdx_ioctl_get_info(struct vfio_cdx_device *vdev,
 				   struct vfio_device_info __user *arg)
 {
@@ -169,6 +223,7 @@ static const struct vfio_device_ops vfio_cdx_ops = {
 	.open_device	= vfio_cdx_open_device,
 	.close_device	= vfio_cdx_close_device,
 	.ioctl		= vfio_cdx_ioctl,
+	.device_feature = vfio_cdx_ioctl_feature,
 	.mmap		= vfio_cdx_mmap,
 	.bind_iommufd	= vfio_iommufd_physical_bind,
 	.unbind_iommufd	= vfio_iommufd_physical_unbind,

drivers/vfio/cdx/private.h

@@ -23,6 +23,8 @@ struct vfio_cdx_region {
 struct vfio_cdx_device {
 	struct vfio_device	vdev;
 	struct vfio_cdx_region	*regions;
+	u32			flags;
+#define BME_SUPPORT BIT(0)
 };
 
 #endif /* VFIO_CDX_PRIVATE_H */

drivers/vfio/pci/mlx5/cmd.c

@@ -86,7 +86,8 @@ 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, u8 query_flags)
+					  size_t *state_size, u64 *total_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)] = {};
@@ -128,6 +129,7 @@ int mlx5vf_cmd_query_vhca_migration_state(struct mlx5vf_pci_core_device *mvdev,
 	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);
+	MLX5_SET(query_vhca_migration_state_in, in, chunk, mvdev->chunk_mode);
 
 	ret = mlx5_cmd_exec_inout(mvdev->mdev, query_vhca_migration_state, in,
 				  out);
@@ -139,6 +141,11 @@ int mlx5vf_cmd_query_vhca_migration_state(struct mlx5vf_pci_core_device *mvdev,
 
 	*state_size = MLX5_GET(query_vhca_migration_state_out, out,
 			       required_umem_size);
+	if (total_size)
+		*total_size = mvdev->chunk_mode ?
+			MLX5_GET64(query_vhca_migration_state_out, out,
+				   remaining_total_size) : *state_size;
+
 	return 0;
 }
 
@@ -254,6 +261,9 @@ void mlx5vf_cmd_set_migratable(struct mlx5vf_pci_core_device *mvdev,
 		mvdev->core_device.vdev.migration_flags |=
 			VFIO_MIGRATION_PRE_COPY;
 
+	if (MLX5_CAP_GEN_2(mvdev->mdev, migration_in_chunks))
+		mvdev->chunk_mode = 1;
+
 end:
 	mlx5_vf_put_core_dev(mvdev->mdev);
 }
@@ -428,6 +438,7 @@ mlx5vf_alloc_data_buffer(struct mlx5_vf_migration_file *migf,
 void mlx5vf_put_data_buffer(struct mlx5_vhca_data_buffer *buf)
 {
 	spin_lock_irq(&buf->migf->list_lock);
+	buf->stop_copy_chunk_num = 0;
 	list_add_tail(&buf->buf_elm, &buf->migf->avail_list);
 	spin_unlock_irq(&buf->migf->list_lock);
 }
@@ -475,6 +486,15 @@ mlx5vf_get_data_buffer(struct mlx5_vf_migration_file *migf,
 	return buf;
 }
 
+static void
+mlx5vf_save_callback_complete(struct mlx5_vf_migration_file *migf,
+			      struct mlx5vf_async_data *async_data)
+{
+	kvfree(async_data->out);
+	complete(&migf->save_comp);
+	fput(migf->filp);
+}
+
 void mlx5vf_mig_file_cleanup_cb(struct work_struct *_work)
 {
 	struct mlx5vf_async_data *async_data = container_of(_work,
@@ -487,16 +507,15 @@ void mlx5vf_mig_file_cleanup_cb(struct work_struct *_work)
 		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)
+		if (!async_data->stop_copy_chunk &&
+		    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);
-	kvfree(async_data->out);
-	complete(&migf->save_comp);
-	fput(migf->filp);
+	mlx5vf_save_callback_complete(migf, async_data);
 }
 
 static int add_buf_header(struct mlx5_vhca_data_buffer *header_buf,
@@ -536,13 +555,20 @@ static void mlx5vf_save_callback(int status, struct mlx5_async_work *context)
 			struct mlx5_vf_migration_file, async_data);
 
 	if (!status) {
+		size_t next_required_umem_size = 0;
+		bool stop_copy_last_chunk;
 		size_t image_size;
 		unsigned long flags;
 		bool initial_pre_copy = migf->state != MLX5_MIGF_STATE_PRE_COPY &&
-				!async_data->last_chunk;
+				!async_data->stop_copy_chunk;
 
 		image_size = MLX5_GET(save_vhca_state_out, async_data->out,
 				      actual_image_size);
+		if (async_data->buf->stop_copy_chunk_num)
+			next_required_umem_size = MLX5_GET(save_vhca_state_out,
+					async_data->out, next_required_umem_size);
+		stop_copy_last_chunk = async_data->stop_copy_chunk &&
+				!next_required_umem_size;
 		if (async_data->header_buf) {
 			status = add_buf_header(async_data->header_buf, image_size,
 						initial_pre_copy);
@@ -554,19 +580,34 @@ static void mlx5vf_save_callback(int status, struct mlx5_async_work *context)
 		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);
+		if (async_data->buf->stop_copy_chunk_num) {
+			migf->num_ready_chunks++;
+			if (next_required_umem_size &&
+			    migf->num_ready_chunks >= MAX_NUM_CHUNKS) {
+				/* Delay the next SAVE till one chunk be consumed */
+				migf->next_required_umem_size = next_required_umem_size;
+				next_required_umem_size = 0;
+			}
+		}
 		spin_unlock_irqrestore(&migf->list_lock, flags);
-		if (initial_pre_copy)
+		if (initial_pre_copy) {
 			migf->pre_copy_initial_bytes += image_size;
-		migf->state = async_data->last_chunk ?
-			MLX5_MIGF_STATE_COMPLETE : MLX5_MIGF_STATE_PRE_COPY;
+			migf->state = MLX5_MIGF_STATE_PRE_COPY;
+		}
+		if (stop_copy_last_chunk)
+			migf->state = MLX5_MIGF_STATE_COMPLETE;
 		wake_up_interruptible(&migf->poll_wait);
+		if (next_required_umem_size)
+			mlx5vf_mig_file_set_save_work(migf,
+				/* Picking up the next chunk num */
+				(async_data->buf->stop_copy_chunk_num % MAX_NUM_CHUNKS) + 1,
+				next_required_umem_size);
+		mlx5vf_save_callback_complete(migf, async_data);
+		return;
 	}
 
 err:
-	/*
-	 * The error and the cleanup flows can't run from an
-	 * interrupt context
-	 */
+	/* The error flow 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;
@@ -610,7 +651,7 @@ int mlx5vf_cmd_save_vhca_state(struct mlx5vf_pci_core_device *mvdev,
 
 	async_data = &migf->async_data;
 	async_data->buf = buf;
-	async_data->last_chunk = !track;
+	async_data->stop_copy_chunk = !track;
 	async_data->out = kvzalloc(out_size, GFP_KERNEL);
 	if (!async_data->out) {
 		err = -ENOMEM;
@@ -618,10 +659,15 @@ int mlx5vf_cmd_save_vhca_state(struct mlx5vf_pci_core_device *mvdev,
 	}
 
 	if (MLX5VF_PRE_COPY_SUPP(mvdev)) {
-		if (async_data->last_chunk && migf->buf_header) {
-			header_buf = migf->buf_header;
-			migf->buf_header = NULL;
-		} else {
+		if (async_data->stop_copy_chunk) {
+			u8 header_idx = buf->stop_copy_chunk_num ?
+				buf->stop_copy_chunk_num - 1 : 0;
+
+			header_buf = migf->buf_header[header_idx];
+			migf->buf_header[header_idx] = NULL;
+		}
+
+		if (!header_buf) {
 			header_buf = mlx5vf_get_data_buffer(migf,
 				sizeof(struct mlx5_vf_migration_header), DMA_NONE);
 			if (IS_ERR(header_buf)) {
@@ -631,8 +677,8 @@ int mlx5vf_cmd_save_vhca_state(struct mlx5vf_pci_core_device *mvdev,
 		}
 	}
 
-	if (async_data->last_chunk)
-		migf->state = MLX5_MIGF_STATE_SAVE_LAST;
+	if (async_data->stop_copy_chunk)
+		migf->state = MLX5_MIGF_STATE_SAVE_STOP_COPY_CHUNK;
 
 	async_data->header_buf = header_buf;
 	get_file(migf->filp);
@@ -707,18 +753,21 @@ 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 *entry;
+	int i;
 
 	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;
-	}
+	for (i = 0; i < MAX_NUM_CHUNKS; i++) {
+		if (migf->buf[i]) {
+			mlx5vf_free_data_buffer(migf->buf[i]);
+			migf->buf[i] = NULL;
+		}
 
-	if (migf->buf_header) {
-		mlx5vf_free_data_buffer(migf->buf_header);
-		migf->buf_header = NULL;
+		if (migf->buf_header[i]) {
+			mlx5vf_free_data_buffer(migf->buf_header[i]);
+			migf->buf_header[i] = NULL;
+		}
 	}
 
 	list_splice(&migf->avail_list, &migf->buf_list);

drivers/vfio/pci/mlx5/cmd.h

@@ -20,7 +20,7 @@ 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_SAVE_STOP_COPY_CHUNK,
 	MLX5_MIGF_STATE_COMPLETE,
 };
 
@@ -64,6 +64,7 @@ struct mlx5_vhca_data_buffer {
 	u32 mkey;
 	enum dma_data_direction dma_dir;
 	u8 dmaed:1;
+	u8 stop_copy_chunk_num;
 	struct list_head buf_elm;
 	struct mlx5_vf_migration_file *migf;
 	/* Optimize mlx5vf_get_migration_page() for sequential access */
@@ -78,10 +79,19 @@ struct mlx5vf_async_data {
 	struct mlx5_vhca_data_buffer *buf;
 	struct mlx5_vhca_data_buffer *header_buf;
 	int status;
-	u8 last_chunk:1;
+	u8 stop_copy_chunk:1;
 	void *out;
 };
 
+struct mlx5vf_save_work_data {
+	struct mlx5_vf_migration_file *migf;
+	size_t next_required_umem_size;
+	struct work_struct work;
+	u8 chunk_num;
+};
+
+#define MAX_NUM_CHUNKS 2
+
 struct mlx5_vf_migration_file {
 	struct file *filp;
 	struct mutex lock;
@@ -94,8 +104,12 @@ struct mlx5_vf_migration_file {
 	u32 record_tag;
 	u64 stop_copy_prep_size;
 	u64 pre_copy_initial_bytes;
-	struct mlx5_vhca_data_buffer *buf;
-	struct mlx5_vhca_data_buffer *buf_header;
+	size_t next_required_umem_size;
+	u8 num_ready_chunks;
+	/* Upon chunk mode preserve another set of buffers for stop_copy phase */
+	struct mlx5_vhca_data_buffer *buf[MAX_NUM_CHUNKS];
+	struct mlx5_vhca_data_buffer *buf_header[MAX_NUM_CHUNKS];
+	struct mlx5vf_save_work_data save_data[MAX_NUM_CHUNKS];
 	spinlock_t list_lock;
 	struct list_head buf_list;
 	struct list_head avail_list;
@@ -164,6 +178,7 @@ struct mlx5vf_pci_core_device {
 	u8 deferred_reset:1;
 	u8 mdev_detach:1;
 	u8 log_active:1;
+	u8 chunk_mode:1;
 	struct completion tracker_comp;
 	/* protect migration state */
 	struct mutex state_mutex;
@@ -186,7 +201,8 @@ enum {
 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, u8 query_flags);
+					  size_t *state_size, u64 *total_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);
@@ -217,6 +233,8 @@ struct page *mlx5vf_get_migration_page(struct mlx5_vhca_data_buffer *buf,
 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);
+void mlx5vf_mig_file_set_save_work(struct mlx5_vf_migration_file *migf,
+				   u8 chunk_num, size_t next_required_umem_size);
 int mlx5vf_start_page_tracker(struct vfio_device *vdev,
 		struct rb_root_cached *ranges, u32 nnodes, u64 *page_size);
 int mlx5vf_stop_page_tracker(struct vfio_device *vdev);

drivers/vfio/pci/mlx5/main.c

@@ -24,6 +24,8 @@
 /* Device specification max LOAD size */
 #define MAX_LOAD_SIZE (BIT_ULL(__mlx5_bit_sz(load_vhca_state_in, size)) - 1)
 
+#define MAX_CHUNK_SIZE SZ_8M
+
 static struct mlx5vf_pci_core_device *mlx5vf_drvdata(struct pci_dev *pdev)
 {
 	struct vfio_pci_core_device *core_device = dev_get_drvdata(&pdev->dev);
@@ -158,6 +160,41 @@ mlx5vf_get_data_buff_from_pos(struct mlx5_vf_migration_file *migf, loff_t pos,
 	return found ? buf : NULL;
 }
 
+static void mlx5vf_buf_read_done(struct mlx5_vhca_data_buffer *vhca_buf)
+{
+	struct mlx5_vf_migration_file *migf = vhca_buf->migf;
+
+	if (vhca_buf->stop_copy_chunk_num) {
+		bool is_header = vhca_buf->dma_dir == DMA_NONE;
+		u8 chunk_num = vhca_buf->stop_copy_chunk_num;
+		size_t next_required_umem_size = 0;
+
+		if (is_header)
+			migf->buf_header[chunk_num - 1] = vhca_buf;
+		else
+			migf->buf[chunk_num - 1] = vhca_buf;
+
+		spin_lock_irq(&migf->list_lock);
+		list_del_init(&vhca_buf->buf_elm);
+		if (!is_header) {
+			next_required_umem_size =
+				migf->next_required_umem_size;
+			migf->next_required_umem_size = 0;
+			migf->num_ready_chunks--;
+		}
+		spin_unlock_irq(&migf->list_lock);
+		if (next_required_umem_size)
+			mlx5vf_mig_file_set_save_work(migf, chunk_num,
+						      next_required_umem_size);
+		return;
+	}
+
+	spin_lock_irq(&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(&migf->list_lock);
+}
+
 static ssize_t mlx5vf_buf_read(struct mlx5_vhca_data_buffer *vhca_buf,
 			       char __user **buf, size_t *len, loff_t *pos)
 {
@@ -193,12 +230,8 @@ static ssize_t mlx5vf_buf_read(struct mlx5_vhca_data_buffer *vhca_buf,
 		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);
-	}
+	if (*pos >= vhca_buf->start_pos + vhca_buf->length)
+		mlx5vf_buf_read_done(vhca_buf);
 
 	return done;
 }
@@ -304,7 +337,75 @@ static void mlx5vf_mark_err(struct mlx5_vf_migration_file *migf)
 	wake_up_interruptible(&migf->poll_wait);
 }
 
-static int mlx5vf_add_stop_copy_header(struct mlx5_vf_migration_file *migf)
+void mlx5vf_mig_file_set_save_work(struct mlx5_vf_migration_file *migf,
+				   u8 chunk_num, size_t next_required_umem_size)
+{
+	migf->save_data[chunk_num - 1].next_required_umem_size =
+			next_required_umem_size;
+	migf->save_data[chunk_num - 1].migf = migf;
+	get_file(migf->filp);
+	queue_work(migf->mvdev->cb_wq,
+		   &migf->save_data[chunk_num - 1].work);
+}
+
+static struct mlx5_vhca_data_buffer *
+mlx5vf_mig_file_get_stop_copy_buf(struct mlx5_vf_migration_file *migf,
+				  u8 index, size_t required_length)
+{
+	struct mlx5_vhca_data_buffer *buf = migf->buf[index];
+	u8 chunk_num;
+
+	WARN_ON(!buf);
+	chunk_num = buf->stop_copy_chunk_num;
+	buf->migf->buf[index] = NULL;
+	/* Checking whether the pre-allocated buffer can fit */
+	if (buf->allocated_length >= required_length)
+		return buf;
+
+	mlx5vf_put_data_buffer(buf);
+	buf = mlx5vf_get_data_buffer(buf->migf, required_length,
+				     DMA_FROM_DEVICE);
+	if (IS_ERR(buf))
+		return buf;
+
+	buf->stop_copy_chunk_num = chunk_num;
+	return buf;
+}
+
+static void mlx5vf_mig_file_save_work(struct work_struct *_work)
+{
+	struct mlx5vf_save_work_data *save_data = container_of(_work,
+		struct mlx5vf_save_work_data, work);
+	struct mlx5_vf_migration_file *migf = save_data->migf;
+	struct mlx5vf_pci_core_device *mvdev = migf->mvdev;
+	struct mlx5_vhca_data_buffer *buf;
+
+	mutex_lock(&mvdev->state_mutex);
+	if (migf->state == MLX5_MIGF_STATE_ERROR)
+		goto end;
+
+	buf = mlx5vf_mig_file_get_stop_copy_buf(migf,
+				save_data->chunk_num - 1,
+				save_data->next_required_umem_size);
+	if (IS_ERR(buf))
+		goto err;
+
+	if (mlx5vf_cmd_save_vhca_state(mvdev, migf, buf, true, false))
+		goto err_save;
+
+	goto end;
+
+err_save:
+	mlx5vf_put_data_buffer(buf);
+err:
+	mlx5vf_mark_err(migf);
+end:
+	mlx5vf_state_mutex_unlock(mvdev);
+	fput(migf->filp);
+}
+
+static int mlx5vf_add_stop_copy_header(struct mlx5_vf_migration_file *migf,
+				       bool track)
 {
 	size_t size = sizeof(struct mlx5_vf_migration_header) +
 		sizeof(struct mlx5_vf_migration_tag_stop_copy_data);
@@ -331,7 +432,7 @@ static int mlx5vf_add_stop_copy_header(struct mlx5_vf_migration_file *migf)
 	to_buff = kmap_local_page(page);
 	memcpy(to_buff, &header, sizeof(header));
 	header_buf->length = sizeof(header);
-	data.stop_copy_size = cpu_to_le64(migf->buf->allocated_length);
+	data.stop_copy_size = cpu_to_le64(migf->buf[0]->allocated_length);
 	memcpy(to_buff + sizeof(header), &data, sizeof(data));
 	header_buf->length += sizeof(data);
 	kunmap_local(to_buff);
@@ -340,48 +441,86 @@ static int mlx5vf_add_stop_copy_header(struct mlx5_vf_migration_file *migf)
 	spin_lock_irqsave(&migf->list_lock, flags);
 	list_add_tail(&header_buf->buf_elm, &migf->buf_list);
 	spin_unlock_irqrestore(&migf->list_lock, flags);
-	migf->pre_copy_initial_bytes = size;
+	if (track)
+		migf->pre_copy_initial_bytes = size;
 	return 0;
 err:
 	mlx5vf_put_data_buffer(header_buf);
 	return ret;
 }
 
-static int mlx5vf_prep_stop_copy(struct mlx5_vf_migration_file *migf,
-				 size_t state_size)
+static int mlx5vf_prep_stop_copy(struct mlx5vf_pci_core_device *mvdev,
+				 struct mlx5_vf_migration_file *migf,
+				 size_t state_size, u64 full_size,
+				 bool track)
 {
 	struct mlx5_vhca_data_buffer *buf;
 	size_t inc_state_size;
+	int num_chunks;
 	int ret;
+	int i;
 
-	/* let's be ready for stop_copy size that might grow by 10 percents */
-	if (check_add_overflow(state_size, state_size / 10, &inc_state_size))
-		inc_state_size = state_size;
+	if (mvdev->chunk_mode) {
+		size_t chunk_size = min_t(size_t, MAX_CHUNK_SIZE, full_size);
 
-	buf = mlx5vf_get_data_buffer(migf, inc_state_size, DMA_FROM_DEVICE);
-	if (IS_ERR(buf))
-		return PTR_ERR(buf);
+		/* from firmware perspective at least 'state_size' buffer should be set */
+		inc_state_size = max(state_size, chunk_size);
+	} else {
+		if (track) {
+			/* let's be ready for stop_copy size that might grow by 10 percents */
+			if (check_add_overflow(state_size, state_size / 10, &inc_state_size))
+				inc_state_size = state_size;
+		} else {
+			inc_state_size = state_size;
+		}
+	}
 
-	migf->buf = buf;
-	buf = mlx5vf_get_data_buffer(migf,
-			sizeof(struct mlx5_vf_migration_header), DMA_NONE);
-	if (IS_ERR(buf)) {
-		ret = PTR_ERR(buf);
-		goto err;
+	/* let's not overflow the device specification max SAVE size */
+	inc_state_size = min_t(size_t, inc_state_size,
+		(BIT_ULL(__mlx5_bit_sz(save_vhca_state_in, size)) - PAGE_SIZE));
+
+	num_chunks = mvdev->chunk_mode ? MAX_NUM_CHUNKS : 1;
+	for (i = 0; i < num_chunks; i++) {
+		buf = mlx5vf_get_data_buffer(migf, inc_state_size, DMA_FROM_DEVICE);
+		if (IS_ERR(buf)) {
+			ret = PTR_ERR(buf);
+			goto err;
+		}
+
+		migf->buf[i] = buf;
+		buf = mlx5vf_get_data_buffer(migf,
+				sizeof(struct mlx5_vf_migration_header), DMA_NONE);
+		if (IS_ERR(buf)) {
+			ret = PTR_ERR(buf);
+			goto err;
+		}
+		migf->buf_header[i] = buf;
+		if (mvdev->chunk_mode) {
+			migf->buf[i]->stop_copy_chunk_num = i + 1;
+			migf->buf_header[i]->stop_copy_chunk_num = i + 1;
+			INIT_WORK(&migf->save_data[i].work,
+				  mlx5vf_mig_file_save_work);
+			migf->save_data[i].chunk_num = i + 1;
+		}
 	}
 
-	migf->buf_header = buf;
-	ret = mlx5vf_add_stop_copy_header(migf);
+	ret = mlx5vf_add_stop_copy_header(migf, track);
 	if (ret)
-		goto err_header;
+		goto err;
 	return 0;
 
-err_header:
-	mlx5vf_put_data_buffer(migf->buf_header);
-	migf->buf_header = NULL;
 err:
-	mlx5vf_put_data_buffer(migf->buf);
-	migf->buf = NULL;
+	for (i = 0; i < num_chunks; i++) {
+		if (migf->buf[i]) {
+			mlx5vf_put_data_buffer(migf->buf[i]);
+			migf->buf[i] = NULL;
+		}
+		if (migf->buf_header[i]) {
+			mlx5vf_put_data_buffer(migf->buf_header[i]);
+			migf->buf_header[i] = NULL;
+		}
+	}
+
 	return ret;
 }
 
@@ -428,7 +567,7 @@ static long mlx5vf_precopy_ioctl(struct file *filp, unsigned int cmd,
 		 * 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);
+							    NULL, MLX5VF_QUERY_INC);
 		if (ret)
 			goto err_state_unlock;
 	}
@@ -505,21 +644,15 @@ static int mlx5vf_pci_save_device_inc_data(struct mlx5vf_pci_core_device *mvdev)
 	if (migf->state == MLX5_MIGF_STATE_ERROR)
 		return -ENODEV;
 
-	ret = mlx5vf_cmd_query_vhca_migration_state(mvdev, &length,
+	ret = mlx5vf_cmd_query_vhca_migration_state(mvdev, &length, NULL,
 				MLX5VF_QUERY_INC | MLX5VF_QUERY_FINAL);
 	if (ret)
 		goto err;
 
-	/* Checking whether we have a matching pre-allocated buffer that can fit */
-	if (migf->buf && migf->buf->allocated_length >= length) {
-		buf = migf->buf;
-		migf->buf = NULL;
-	} else {
-		buf = mlx5vf_get_data_buffer(migf, length, DMA_FROM_DEVICE);
-		if (IS_ERR(buf)) {
-			ret = PTR_ERR(buf);
-			goto err;
-		}
+	buf = mlx5vf_mig_file_get_stop_copy_buf(migf, 0, length);
+	if (IS_ERR(buf)) {
+		ret = PTR_ERR(buf);
+		goto err;
 	}
 
 	ret = mlx5vf_cmd_save_vhca_state(mvdev, migf, buf, true, false);
@@ -541,6 +674,7 @@ 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;
+	u64 full_size;
 	int ret;
 
 	migf = kzalloc(sizeof(*migf), GFP_KERNEL_ACCOUNT);
@@ -574,20 +708,25 @@ mlx5vf_pci_save_device_data(struct mlx5vf_pci_core_device *mvdev, bool track)
 	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);
+	ret = mlx5vf_cmd_query_vhca_migration_state(mvdev, &length, &full_size, 0);
+	if (ret)
+		goto out_pd;
+
+	ret = mlx5vf_prep_stop_copy(mvdev, migf, length, full_size, track);
 	if (ret)
 		goto out_pd;
 
 	if (track) {
-		ret = mlx5vf_prep_stop_copy(migf, length);
-		if (ret)
+		/* leave the allocated buffer ready for the stop-copy phase */
+		buf = mlx5vf_alloc_data_buffer(migf,
+			migf->buf[0]->allocated_length, DMA_FROM_DEVICE);
+		if (IS_ERR(buf)) {
+			ret = PTR_ERR(buf);
 			goto out_pd;
-	}
-
-	buf = mlx5vf_alloc_data_buffer(migf, length, DMA_FROM_DEVICE);
-	if (IS_ERR(buf)) {
-		ret = PTR_ERR(buf);
-		goto out_pd;
+		}
+	} else {
+		buf = migf->buf[0];
+		migf->buf[0] = NULL;
 	}
 
 	ret = mlx5vf_cmd_save_vhca_state(mvdev, migf, buf, false, track);
@@ -820,8 +959,8 @@ 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;
+	struct mlx5_vhca_data_buffer *vhca_buf = migf->buf[0];
+	struct mlx5_vhca_data_buffer *vhca_buf_header = migf->buf_header[0];
 	loff_t requested_length;
 	bool has_work = false;
 	ssize_t done = 0;
@@ -856,15 +995,15 @@ static ssize_t mlx5vf_resume_write(struct file *filp, const char __user *buf,
 			if (vhca_buf_header->allocated_length < migf->record_size) {
 				mlx5vf_free_data_buffer(vhca_buf_header);
 
-				migf->buf_header = mlx5vf_alloc_data_buffer(migf,
+				migf->buf_header[0] = mlx5vf_alloc_data_buffer(migf,
 						migf->record_size, DMA_NONE);
-				if (IS_ERR(migf->buf_header)) {
-					ret = PTR_ERR(migf->buf_header);
-					migf->buf_header = NULL;
+				if (IS_ERR(migf->buf_header[0])) {
+					ret = PTR_ERR(migf->buf_header[0]);
+					migf->buf_header[0] = NULL;
 					goto out_unlock;
 				}
 
-				vhca_buf_header = migf->buf_header;
+				vhca_buf_header = migf->buf_header[0];
 			}
 
 			vhca_buf_header->start_pos = migf->max_pos;
@@ -884,15 +1023,15 @@ static ssize_t mlx5vf_resume_write(struct file *filp, const char __user *buf,
 			if (vhca_buf->allocated_length < size) {
 				mlx5vf_free_data_buffer(vhca_buf);
 
-				migf->buf = mlx5vf_alloc_data_buffer(migf,
+				migf->buf[0] = mlx5vf_alloc_data_buffer(migf,
 							size, DMA_TO_DEVICE);
-				if (IS_ERR(migf->buf)) {
-					ret = PTR_ERR(migf->buf);
-					migf->buf = NULL;
+				if (IS_ERR(migf->buf[0])) {
+					ret = PTR_ERR(migf->buf[0]);
+					migf->buf[0] = NULL;
 					goto out_unlock;
 				}
 
-				vhca_buf = migf->buf;
+				vhca_buf = migf->buf[0];
 			}
 
 			vhca_buf->start_pos = migf->max_pos;
@@ -974,7 +1113,7 @@ mlx5vf_pci_resume_device_data(struct mlx5vf_pci_core_device *mvdev)
 		goto out_pd;
 	}
 
-	migf->buf = buf;
+	migf->buf[0] = buf;
 	if (MLX5VF_PRE_COPY_SUPP(mvdev)) {
 		buf = mlx5vf_alloc_data_buffer(migf,
 			sizeof(struct mlx5_vf_migration_header), DMA_NONE);
@@ -983,7 +1122,7 @@ mlx5vf_pci_resume_device_data(struct mlx5vf_pci_core_device *mvdev)
 			goto out_buf;
 		}
 
-		migf->buf_header = buf;
+		migf->buf_header[0] = buf;
 		migf->load_state = MLX5_VF_LOAD_STATE_READ_HEADER;
 	} else {
 		/* Initial state will be to read the image */
@@ -997,7 +1136,7 @@ mlx5vf_pci_resume_device_data(struct mlx5vf_pci_core_device *mvdev)
 	spin_lock_init(&migf->list_lock);
 	return migf;
 out_buf:
-	mlx5vf_free_data_buffer(migf->buf);
+	mlx5vf_free_data_buffer(migf->buf[0]);
 out_pd:
 	mlx5vf_cmd_dealloc_pd(migf);
 out_free:
@@ -1019,6 +1158,7 @@ void mlx5vf_disable_fds(struct mlx5vf_pci_core_device *mvdev)
 		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);
+		wake_up_interruptible(&mvdev->saving_migf->poll_wait);
 		mlx5fv_cmd_clean_migf_resources(mvdev->saving_migf);
 		fput(mvdev->saving_migf->filp);
 		mvdev->saving_migf = NULL;
@@ -1100,7 +1240,7 @@ mlx5vf_pci_step_device_state_locked(struct mlx5vf_pci_core_device *mvdev,
 		if (!MLX5VF_PRE_COPY_SUPP(mvdev)) {
 			ret = mlx5vf_cmd_load_vhca_state(mvdev,
 							 mvdev->resuming_migf,
-							 mvdev->resuming_migf->buf);
+							 mvdev->resuming_migf->buf[0]);
 			if (ret)
 				return ERR_PTR(ret);
 		}
@@ -1194,13 +1334,14 @@ static int mlx5vf_pci_get_data_size(struct vfio_device *vdev,
 	struct mlx5vf_pci_core_device *mvdev = container_of(
 		vdev, struct mlx5vf_pci_core_device, core_device.vdev);
 	size_t state_size;
+	u64 total_size;
 	int ret;
 
 	mutex_lock(&mvdev->state_mutex);
-	ret = mlx5vf_cmd_query_vhca_migration_state(mvdev,
-						    &state_size, 0);
+	ret = mlx5vf_cmd_query_vhca_migration_state(mvdev, &state_size,
+						    &total_size, 0);
 	if (!ret)
-		*stop_copy_length = state_size;
+		*stop_copy_length = total_size;
 	mlx5vf_state_mutex_unlock(mvdev);
 	return ret;
 }

drivers/vfio/vfio_main.c

@@ -946,6 +946,11 @@ void vfio_combine_iova_ranges(struct rb_root_cached *root, u32 cur_nodes,
 		unsigned long last;
 
 		comb_start = interval_tree_iter_first(root, 0, ULONG_MAX);
+
+		/* Empty list */
+		if (WARN_ON_ONCE(!comb_start))
+			return;
+
 		curr = comb_start;
 		while (curr) {
 			last = curr->last;
@@ -975,6 +980,11 @@ void vfio_combine_iova_ranges(struct rb_root_cached *root, u32 cur_nodes,
 			prev = curr;
 			curr = interval_tree_iter_next(curr, 0, ULONG_MAX);
 		}
+
+		/* Empty list or no nodes to combine */
+		if (WARN_ON_ONCE(min_gap == ULONG_MAX))
+			break;
+
 		comb_start->last = comb_end->last;
 		interval_tree_remove(comb_end, root);
 		cur_nodes--;

include/linux/cdx/cdx_bus.h

@@ -21,11 +21,13 @@
 struct cdx_controller;
 
 enum {
+	CDX_DEV_BUS_MASTER_CONF,
 	CDX_DEV_RESET_CONF,
 };
 
 struct cdx_device_config {
 	u8 type;
+	bool bus_master_enable;
 };
 
 typedef int (*cdx_scan_cb)(struct cdx_controller *cdx);
@@ -170,4 +172,20 @@ extern struct bus_type cdx_bus_type;
  */
 int cdx_dev_reset(struct device *dev);
 
+/**
+ * cdx_set_master - enables bus-mastering for CDX device
+ * @cdx_dev: the CDX device to enable
+ *
+ * Return: 0 for success, -errno on failure
+ */
+int cdx_set_master(struct cdx_device *cdx_dev);
+
+/**
+ * cdx_clear_master - disables bus-mastering for CDX device
+ * @cdx_dev: the CDX device to disable
+ *
+ * Return: 0 for success, -errno on failure
+ */
+int cdx_clear_master(struct cdx_device *cdx_dev);
+
 #endif /* _CDX_BUS_H_ */

include/uapi/linux/vfio.h

@@ -277,8 +277,8 @@ struct vfio_region_info {
 #define VFIO_REGION_INFO_FLAG_CAPS	(1 << 3) /* Info supports caps */
 	__u32	index;		/* Region index */
 	__u32	cap_offset;	/* Offset within info struct of first cap */
-	__u64	size;		/* Region size (bytes) */
-	__u64	offset;		/* Region offset from start of device fd */
+	__aligned_u64	size;	/* Region size (bytes) */
+	__aligned_u64	offset;	/* Region offset from start of device fd */
 };
 #define VFIO_DEVICE_GET_REGION_INFO	_IO(VFIO_TYPE, VFIO_BASE + 8)
 
@@ -294,8 +294,8 @@ struct vfio_region_info {
 #define VFIO_REGION_INFO_CAP_SPARSE_MMAP	1
 
 struct vfio_region_sparse_mmap_area {
-	__u64	offset;	/* Offset of mmap'able area within region */
-	__u64	size;	/* Size of mmap'able area */
+	__aligned_u64	offset;	/* Offset of mmap'able area within region */
+	__aligned_u64	size;	/* Size of mmap'able area */
 };
 
 struct vfio_region_info_cap_sparse_mmap {
@@ -450,9 +450,9 @@ struct vfio_device_migration_info {
 					     VFIO_DEVICE_STATE_V1_RESUMING)
 
 	__u32 reserved;
-	__u64 pending_bytes;
-	__u64 data_offset;
-	__u64 data_size;
+	__aligned_u64 pending_bytes;
+	__aligned_u64 data_offset;
+	__aligned_u64 data_size;
 };
 
 /*
@@ -476,7 +476,7 @@ struct vfio_device_migration_info {
 
 struct vfio_region_info_cap_nvlink2_ssatgt {
 	struct vfio_info_cap_header header;
-	__u64 tgt;
+	__aligned_u64 tgt;
 };
 
 /*
@@ -816,7 +816,7 @@ struct vfio_device_gfx_plane_info {
 	__u32 drm_plane_type;	/* type of plane: DRM_PLANE_TYPE_* */
 	/* out */
 	__u32 drm_format;	/* drm format of plane */
-	__u64 drm_format_mod;   /* tiled mode */
+	__aligned_u64 drm_format_mod;   /* tiled mode */
 	__u32 width;	/* width of plane */
 	__u32 height;	/* height of plane */
 	__u32 stride;	/* stride of plane */
@@ -829,6 +829,7 @@ struct vfio_device_gfx_plane_info {
 		__u32 region_index;	/* region index */
 		__u32 dmabuf_id;	/* dma-buf id */
 	};
+	__u32 reserved;
 };
 
 #define VFIO_DEVICE_QUERY_GFX_PLANE _IO(VFIO_TYPE, VFIO_BASE + 14)
@@ -863,9 +864,10 @@ struct vfio_device_ioeventfd {
 #define VFIO_DEVICE_IOEVENTFD_32	(1 << 2) /* 4-byte write */
 #define VFIO_DEVICE_IOEVENTFD_64	(1 << 3) /* 8-byte write */
 #define VFIO_DEVICE_IOEVENTFD_SIZE_MASK	(0xf)
-	__u64	offset;			/* device fd offset of write */
-	__u64	data;			/* data to be written */
+	__aligned_u64	offset;		/* device fd offset of write */
+	__aligned_u64	data;		/* data to be written */
 	__s32	fd;			/* -1 for de-assignment */
+	__u32	reserved;
 };
 
 #define VFIO_DEVICE_IOEVENTFD		_IO(VFIO_TYPE, VFIO_BASE + 16)
@@ -1434,6 +1436,27 @@ struct vfio_device_feature_mig_data_size {
 
 #define VFIO_DEVICE_FEATURE_MIG_DATA_SIZE 9
 
+/**
+ * Upon VFIO_DEVICE_FEATURE_SET, set or clear the BUS mastering for the device
+ * based on the operation specified in op flag.
+ *
+ * The functionality is incorporated for devices that needs bus master control,
+ * but the in-band device interface lacks the support. Consequently, it is not
+ * applicable to PCI devices, as bus master control for PCI devices is managed
+ * in-band through the configuration space. At present, this feature is supported
+ * only for CDX devices.
+ * When the device's BUS MASTER setting is configured as CLEAR, it will result in
+ * blocking all incoming DMA requests from the device. On the other hand, configuring
+ * the device's BUS MASTER setting as SET (enable) will grant the device the
+ * capability to perform DMA to the host memory.
+ */
+struct vfio_device_feature_bus_master {
+	__u32 op;
+#define		VFIO_DEVICE_FEATURE_CLEAR_MASTER	0	/* Clear Bus Master */
+#define		VFIO_DEVICE_FEATURE_SET_MASTER		1	/* Set Bus Master */
+};
+#define VFIO_DEVICE_FEATURE_BUS_MASTER 10
+
 /* -------- API for Type1 VFIO IOMMU -------- */
 
 /**
@@ -1449,7 +1472,7 @@ struct vfio_iommu_type1_info {
 	__u32	flags;
 #define VFIO_IOMMU_INFO_PGSIZES (1 << 0)	/* supported page sizes info */
 #define VFIO_IOMMU_INFO_CAPS	(1 << 1)	/* Info supports caps */
-	__u64	iova_pgsizes;	/* Bitmap of supported page sizes */
+	__aligned_u64	iova_pgsizes;		/* Bitmap of supported page sizes */
 	__u32   cap_offset;	/* Offset within info struct of first cap */
 	__u32   pad;
 };

samples/vfio-mdev/mbochs.c

@@ -1262,7 +1262,7 @@ static long mbochs_ioctl(struct vfio_device *vdev, unsigned int cmd,
 
 	case VFIO_DEVICE_QUERY_GFX_PLANE:
 	{
-		struct vfio_device_gfx_plane_info plane;
+		struct vfio_device_gfx_plane_info plane = {};
 
 		minsz = offsetofend(struct vfio_device_gfx_plane_info,
 				    region_index);

samples/vfio-mdev/mdpy.c

@@ -591,7 +591,7 @@ static long mdpy_ioctl(struct vfio_device *vdev, unsigned int cmd,
 
 	case VFIO_DEVICE_QUERY_GFX_PLANE:
 	{
-		struct vfio_device_gfx_plane_info plane;
+		struct vfio_device_gfx_plane_info plane = {};
 
 		minsz = offsetofend(struct vfio_device_gfx_plane_info,
 				    region_index);

samples/vfio-mdev/mtty.c

@@ -29,6 +29,8 @@
 #include <linux/serial.h>
 #include <uapi/linux/serial_reg.h>
 #include <linux/eventfd.h>
+#include <linux/anon_inodes.h>
+
 /*
  * #defines
  */
@@ -124,10 +126,32 @@ struct serial_port {
 	u8 intr_trigger_level;  /* interrupt trigger level */
 };
 
+struct mtty_data {
+	u64 magic;
+#define MTTY_MAGIC 0x7e9d09898c3e2c4e /* Nothing clever, just random */
+	u32 major_ver;
+#define MTTY_MAJOR_VER 1
+	u32 minor_ver;
+#define MTTY_MINOR_VER 0
+	u32 nr_ports;
+	u32 flags;
+	struct serial_port ports[2];
+};
+
+struct mdev_state;
+
+struct mtty_migration_file {
+	struct file *filp;
+	struct mutex lock;
+	struct mdev_state *mdev_state;
+	struct mtty_data data;
+	ssize_t filled_size;
+	u8 disabled:1;
+};
+
 /* State of each mdev device */
 struct mdev_state {
 	struct vfio_device vdev;
-	int irq_fd;
 	struct eventfd_ctx *intx_evtfd;
 	struct eventfd_ctx *msi_evtfd;
 	int irq_index;
@@ -141,6 +165,13 @@ struct mdev_state {
 	struct mutex rxtx_lock;
 	struct vfio_device_info dev_info;
 	int nr_ports;
+	enum vfio_device_mig_state state;
+	struct mutex state_mutex;
+	struct mutex reset_mutex;
+	struct mtty_migration_file *saving_migf;
+	struct mtty_migration_file *resuming_migf;
+	u8 deferred_reset:1;
+	u8 intx_mask:1;
 };
 
 static struct mtty_type {
@@ -166,10 +197,6 @@ static const struct file_operations vd_fops = {
 
 static const struct vfio_device_ops mtty_dev_ops;
 
-/* function prototypes */
-
-static int mtty_trigger_interrupt(struct mdev_state *mdev_state);
-
 /* Helper functions */
 
 static void dump_buffer(u8 *buf, uint32_t count)
@@ -186,6 +213,36 @@ static void dump_buffer(u8 *buf, uint32_t count)
 #endif
 }
 
+static bool is_intx(struct mdev_state *mdev_state)
+{
+	return mdev_state->irq_index == VFIO_PCI_INTX_IRQ_INDEX;
+}
+
+static bool is_msi(struct mdev_state *mdev_state)
+{
+	return mdev_state->irq_index == VFIO_PCI_MSI_IRQ_INDEX;
+}
+
+static bool is_noirq(struct mdev_state *mdev_state)
+{
+	return !is_intx(mdev_state) && !is_msi(mdev_state);
+}
+
+static void mtty_trigger_interrupt(struct mdev_state *mdev_state)
+{
+	lockdep_assert_held(&mdev_state->ops_lock);
+
+	if (is_msi(mdev_state)) {
+		if (mdev_state->msi_evtfd)
+			eventfd_signal(mdev_state->msi_evtfd, 1);
+	} else if (is_intx(mdev_state)) {
+		if (mdev_state->intx_evtfd && !mdev_state->intx_mask) {
+			eventfd_signal(mdev_state->intx_evtfd, 1);
+			mdev_state->intx_mask = true;
+		}
+	}
+}
+
 static void mtty_create_config_space(struct mdev_state *mdev_state)
 {
 	/* PCI dev ID */
@@ -717,6 +774,543 @@ static ssize_t mdev_access(struct mdev_state *mdev_state, u8 *buf, size_t count,
 	return ret;
 }
 
+static size_t mtty_data_size(struct mdev_state *mdev_state)
+{
+	return offsetof(struct mtty_data, ports) +
+		(mdev_state->nr_ports * sizeof(struct serial_port));
+}
+
+static void mtty_disable_file(struct mtty_migration_file *migf)
+{
+	mutex_lock(&migf->lock);
+	migf->disabled = true;
+	migf->filled_size = 0;
+	migf->filp->f_pos = 0;
+	mutex_unlock(&migf->lock);
+}
+
+static void mtty_disable_files(struct mdev_state *mdev_state)
+{
+	if (mdev_state->saving_migf) {
+		mtty_disable_file(mdev_state->saving_migf);
+		fput(mdev_state->saving_migf->filp);
+		mdev_state->saving_migf = NULL;
+	}
+
+	if (mdev_state->resuming_migf) {
+		mtty_disable_file(mdev_state->resuming_migf);
+		fput(mdev_state->resuming_migf->filp);
+		mdev_state->resuming_migf = NULL;
+	}
+}
+
+static void mtty_state_mutex_unlock(struct mdev_state *mdev_state)
+{
+again:
+	mutex_lock(&mdev_state->reset_mutex);
+	if (mdev_state->deferred_reset) {
+		mdev_state->deferred_reset = false;
+		mutex_unlock(&mdev_state->reset_mutex);
+		mdev_state->state = VFIO_DEVICE_STATE_RUNNING;
+		mtty_disable_files(mdev_state);
+		goto again;
+	}
+	mutex_unlock(&mdev_state->state_mutex);
+	mutex_unlock(&mdev_state->reset_mutex);
+}
+
+static int mtty_release_migf(struct inode *inode, struct file *filp)
+{
+	struct mtty_migration_file *migf = filp->private_data;
+
+	mtty_disable_file(migf);
+	mutex_destroy(&migf->lock);
+	kfree(migf);
+
+	return 0;
+}
+
+static long mtty_precopy_ioctl(struct file *filp, unsigned int cmd,
+			       unsigned long arg)
+{
+	struct mtty_migration_file *migf = filp->private_data;
+	struct mdev_state *mdev_state = migf->mdev_state;
+	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(&mdev_state->state_mutex);
+	if (mdev_state->state != VFIO_DEVICE_STATE_PRE_COPY &&
+	    mdev_state->state != VFIO_DEVICE_STATE_PRE_COPY_P2P) {
+		ret = -EINVAL;
+		goto unlock;
+	}
+
+	mutex_lock(&migf->lock);
+
+	if (migf->disabled) {
+		mutex_unlock(&migf->lock);
+		ret = -ENODEV;
+		goto unlock;
+	}
+
+	if (*pos > migf->filled_size) {
+		mutex_unlock(&migf->lock);
+		ret = -EINVAL;
+		goto unlock;
+	}
+
+	info.dirty_bytes = 0;
+	info.initial_bytes = migf->filled_size - *pos;
+	mutex_unlock(&migf->lock);
+
+	ret = copy_to_user((void __user *)arg, &info, minsz) ? -EFAULT : 0;
+unlock:
+	mtty_state_mutex_unlock(mdev_state);
+	return ret;
+}
+
+static ssize_t mtty_save_read(struct file *filp, char __user *buf,
+			      size_t len, loff_t *pos)
+{
+	struct mtty_migration_file *migf = filp->private_data;
+	ssize_t ret = 0;
+
+	if (pos)
+		return -ESPIPE;
+
+	pos = &filp->f_pos;
+
+	mutex_lock(&migf->lock);
+
+	dev_dbg(migf->mdev_state->vdev.dev, "%s ask %zu\n", __func__, len);
+
+	if (migf->disabled) {
+		ret = -ENODEV;
+		goto out_unlock;
+	}
+
+	if (*pos > migf->filled_size) {
+		ret = -EINVAL;
+		goto out_unlock;
+	}
+
+	len = min_t(size_t, migf->filled_size - *pos, len);
+	if (len) {
+		if (copy_to_user(buf, (void *)&migf->data + *pos, len)) {
+			ret = -EFAULT;
+			goto out_unlock;
+		}
+		*pos += len;
+		ret = len;
+	}
+out_unlock:
+	dev_dbg(migf->mdev_state->vdev.dev, "%s read %zu\n", __func__, ret);
+	mutex_unlock(&migf->lock);
+	return ret;
+}
+
+static const struct file_operations mtty_save_fops = {
+	.owner = THIS_MODULE,
+	.read = mtty_save_read,
+	.unlocked_ioctl = mtty_precopy_ioctl,
+	.compat_ioctl = compat_ptr_ioctl,
+	.release = mtty_release_migf,
+	.llseek = no_llseek,
+};
+
+static void mtty_save_state(struct mdev_state *mdev_state)
+{
+	struct mtty_migration_file *migf = mdev_state->saving_migf;
+	int i;
+
+	mutex_lock(&migf->lock);
+	for (i = 0; i < mdev_state->nr_ports; i++) {
+		memcpy(&migf->data.ports[i],
+			&mdev_state->s[i], sizeof(struct serial_port));
+		migf->filled_size += sizeof(struct serial_port);
+	}
+	dev_dbg(mdev_state->vdev.dev,
+		"%s filled to %zu\n", __func__, migf->filled_size);
+	mutex_unlock(&migf->lock);
+}
+
+static int mtty_load_state(struct mdev_state *mdev_state)
+{
+	struct mtty_migration_file *migf = mdev_state->resuming_migf;
+	int i;
+
+	mutex_lock(&migf->lock);
+	/* magic and version already tested by resume write fn */
+	if (migf->filled_size < mtty_data_size(mdev_state)) {
+		dev_dbg(mdev_state->vdev.dev, "%s expected %zu, got %zu\n",
+			__func__, mtty_data_size(mdev_state),
+			migf->filled_size);
+		mutex_unlock(&migf->lock);
+		return -EINVAL;
+	}
+
+	for (i = 0; i < mdev_state->nr_ports; i++)
+		memcpy(&mdev_state->s[i],
+		       &migf->data.ports[i], sizeof(struct serial_port));
+
+	mutex_unlock(&migf->lock);
+	return 0;
+}
+
+static struct mtty_migration_file *
+mtty_save_device_data(struct mdev_state *mdev_state,
+		      enum vfio_device_mig_state state)
+{
+	struct mtty_migration_file *migf = mdev_state->saving_migf;
+	struct mtty_migration_file *ret = NULL;
+
+	if (migf) {
+		if (state == VFIO_DEVICE_STATE_STOP_COPY)
+			goto fill_data;
+		return ret;
+	}
+
+	migf = kzalloc(sizeof(*migf), GFP_KERNEL_ACCOUNT);
+	if (!migf)
+		return ERR_PTR(-ENOMEM);
+
+	migf->filp = anon_inode_getfile("mtty_mig", &mtty_save_fops,
+					migf, O_RDONLY);
+	if (IS_ERR(migf->filp)) {
+		int rc = PTR_ERR(migf->filp);
+
+		kfree(migf);
+		return ERR_PTR(rc);
+	}
+
+	stream_open(migf->filp->f_inode, migf->filp);
+	mutex_init(&migf->lock);
+	migf->mdev_state = mdev_state;
+
+	migf->data.magic = MTTY_MAGIC;
+	migf->data.major_ver = MTTY_MAJOR_VER;
+	migf->data.minor_ver = MTTY_MINOR_VER;
+	migf->data.nr_ports = mdev_state->nr_ports;
+
+	migf->filled_size = offsetof(struct mtty_data, ports);
+
+	dev_dbg(mdev_state->vdev.dev, "%s filled header to %zu\n",
+		__func__, migf->filled_size);
+
+	ret = mdev_state->saving_migf = migf;
+
+fill_data:
+	if (state == VFIO_DEVICE_STATE_STOP_COPY)
+		mtty_save_state(mdev_state);
+
+	return ret;
+}
+
+static ssize_t mtty_resume_write(struct file *filp, const char __user *buf,
+				 size_t len, loff_t *pos)
+{
+	struct mtty_migration_file *migf = filp->private_data;
+	struct mdev_state *mdev_state = migf->mdev_state;
+	loff_t requested_length;
+	ssize_t ret = 0;
+
+	if (pos)
+		return -ESPIPE;
+
+	pos = &filp->f_pos;
+
+	if (*pos < 0 ||
+	    check_add_overflow((loff_t)len, *pos, &requested_length))
+		return -EINVAL;
+
+	if (requested_length > mtty_data_size(mdev_state))
+		return -ENOMEM;
+
+	mutex_lock(&migf->lock);
+
+	if (migf->disabled) {
+		ret = -ENODEV;
+		goto out_unlock;
+	}
+
+	if (copy_from_user((void *)&migf->data + *pos, buf, len)) {
+		ret = -EFAULT;
+		goto out_unlock;
+	}
+
+	*pos += len;
+	ret = len;
+
+	dev_dbg(migf->mdev_state->vdev.dev, "%s received %zu, total %zu\n",
+		__func__, len, migf->filled_size + len);
+
+	if (migf->filled_size < offsetof(struct mtty_data, ports) &&
+	    migf->filled_size + len >= offsetof(struct mtty_data, ports)) {
+		if (migf->data.magic != MTTY_MAGIC || migf->data.flags ||
+		    migf->data.major_ver != MTTY_MAJOR_VER ||
+		    migf->data.minor_ver != MTTY_MINOR_VER ||
+		    migf->data.nr_ports != mdev_state->nr_ports) {
+			dev_dbg(migf->mdev_state->vdev.dev,
+				"%s failed validation\n", __func__);
+			ret = -EFAULT;
+		} else {
+			dev_dbg(migf->mdev_state->vdev.dev,
+				"%s header validated\n", __func__);
+		}
+	}
+
+	migf->filled_size += len;
+
+out_unlock:
+	mutex_unlock(&migf->lock);
+	return ret;
+}
+
+static const struct file_operations mtty_resume_fops = {
+	.owner = THIS_MODULE,
+	.write = mtty_resume_write,
+	.release = mtty_release_migf,
+	.llseek = no_llseek,
+};
+
+static struct mtty_migration_file *
+mtty_resume_device_data(struct mdev_state *mdev_state)
+{
+	struct mtty_migration_file *migf;
+	int ret;
+
+	migf = kzalloc(sizeof(*migf), GFP_KERNEL_ACCOUNT);
+	if (!migf)
+		return ERR_PTR(-ENOMEM);
+
+	migf->filp = anon_inode_getfile("mtty_mig", &mtty_resume_fops,
+					migf, O_WRONLY);
+	if (IS_ERR(migf->filp)) {
+		ret = PTR_ERR(migf->filp);
+		kfree(migf);
+		return ERR_PTR(ret);
+	}
+
+	stream_open(migf->filp->f_inode, migf->filp);
+	mutex_init(&migf->lock);
+	migf->mdev_state = mdev_state;
+
+	mdev_state->resuming_migf = migf;
+
+	return migf;
+}
+
+static struct file *mtty_step_state(struct mdev_state *mdev_state,
+				     enum vfio_device_mig_state new)
+{
+	enum vfio_device_mig_state cur = mdev_state->state;
+
+	dev_dbg(mdev_state->vdev.dev, "%s: %d -> %d\n", __func__, cur, new);
+
+	/*
+	 * The following state transitions are no-op considering
+	 * mtty does not do DMA nor require any explicit start/stop.
+	 *
+	 *         RUNNING -> RUNNING_P2P
+	 *         RUNNING_P2P -> RUNNING
+	 *         RUNNING_P2P -> STOP
+	 *         PRE_COPY -> PRE_COPY_P2P
+	 *         PRE_COPY_P2P -> PRE_COPY
+	 *         STOP -> RUNNING_P2P
+	 */
+	if ((cur == VFIO_DEVICE_STATE_RUNNING &&
+	     new == VFIO_DEVICE_STATE_RUNNING_P2P) ||
+	    (cur == VFIO_DEVICE_STATE_RUNNING_P2P &&
+	     (new == VFIO_DEVICE_STATE_RUNNING ||
+	      new == VFIO_DEVICE_STATE_STOP)) ||
+	    (cur == VFIO_DEVICE_STATE_PRE_COPY &&
+	     new == VFIO_DEVICE_STATE_PRE_COPY_P2P) ||
+	    (cur == VFIO_DEVICE_STATE_PRE_COPY_P2P &&
+	     new == VFIO_DEVICE_STATE_PRE_COPY) ||
+	    (cur == VFIO_DEVICE_STATE_STOP &&
+	     new == VFIO_DEVICE_STATE_RUNNING_P2P))
+		return NULL;
+
+	/*
+	 * The following state transitions simply close migration files,
+	 * with the exception of RESUMING -> STOP, which needs to load
+	 * the state first.
+	 *
+	 *         RESUMING -> STOP
+	 *         PRE_COPY -> RUNNING
+	 *         PRE_COPY_P2P -> RUNNING_P2P
+	 *         STOP_COPY -> STOP
+	 */
+	if (cur == VFIO_DEVICE_STATE_RESUMING &&
+	    new == VFIO_DEVICE_STATE_STOP) {
+		int ret;
+
+		ret = mtty_load_state(mdev_state);
+		if (ret)
+			return ERR_PTR(ret);
+		mtty_disable_files(mdev_state);
+		return NULL;
+	}
+
+	if ((cur == VFIO_DEVICE_STATE_PRE_COPY &&
+	     new == VFIO_DEVICE_STATE_RUNNING) ||
+	    (cur == VFIO_DEVICE_STATE_PRE_COPY_P2P &&
+	     new == VFIO_DEVICE_STATE_RUNNING_P2P) ||
+	    (cur == VFIO_DEVICE_STATE_STOP_COPY &&
+	     new == VFIO_DEVICE_STATE_STOP)) {
+		mtty_disable_files(mdev_state);
+		return NULL;
+	}
+
+	/*
+	 * The following state transitions return migration files.
+	 *
+	 *         RUNNING -> PRE_COPY
+	 *         RUNNING_P2P -> PRE_COPY_P2P
+	 *         STOP -> STOP_COPY
+	 *         STOP -> RESUMING
+	 *         PRE_COPY_P2P -> STOP_COPY
+	 */
+	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) ||
+	    (cur == VFIO_DEVICE_STATE_STOP &&
+	     new == VFIO_DEVICE_STATE_STOP_COPY) ||
+	    (cur == VFIO_DEVICE_STATE_PRE_COPY_P2P &&
+	     new == VFIO_DEVICE_STATE_STOP_COPY)) {
+		struct mtty_migration_file *migf;
+
+		migf = mtty_save_device_data(mdev_state, new);
+		if (IS_ERR(migf))
+			return ERR_CAST(migf);
+
+		if (migf) {
+			get_file(migf->filp);
+
+			return migf->filp;
+		}
+		return NULL;
+	}
+
+	if (cur == VFIO_DEVICE_STATE_STOP &&
+	    new == VFIO_DEVICE_STATE_RESUMING) {
+		struct mtty_migration_file *migf;
+
+		migf = mtty_resume_device_data(mdev_state);
+		if (IS_ERR(migf))
+			return ERR_CAST(migf);
+
+		get_file(migf->filp);
+
+		return migf->filp;
+	}
+
+	/* vfio_mig_get_next_state() does not use arcs other than the above */
+	WARN_ON(true);
+	return ERR_PTR(-EINVAL);
+}
+
+static struct file *mtty_set_state(struct vfio_device *vdev,
+				   enum vfio_device_mig_state new_state)
+{
+	struct mdev_state *mdev_state =
+		container_of(vdev, struct mdev_state, vdev);
+	struct file *ret = NULL;
+
+	dev_dbg(vdev->dev, "%s -> %d\n", __func__, new_state);
+
+	mutex_lock(&mdev_state->state_mutex);
+	while (mdev_state->state != new_state) {
+		enum vfio_device_mig_state next_state;
+		int rc = vfio_mig_get_next_state(vdev, mdev_state->state,
+						 new_state, &next_state);
+		if (rc) {
+			ret = ERR_PTR(rc);
+			break;
+		}
+
+		ret = mtty_step_state(mdev_state, next_state);
+		if (IS_ERR(ret))
+			break;
+
+		mdev_state->state = next_state;
+
+		if (WARN_ON(ret && new_state != next_state)) {
+			fput(ret);
+			ret = ERR_PTR(-EINVAL);
+			break;
+		}
+	}
+	mtty_state_mutex_unlock(mdev_state);
+	return ret;
+}
+
+static int mtty_get_state(struct vfio_device *vdev,
+			  enum vfio_device_mig_state *current_state)
+{
+	struct mdev_state *mdev_state =
+		container_of(vdev, struct mdev_state, vdev);
+
+	mutex_lock(&mdev_state->state_mutex);
+	*current_state = mdev_state->state;
+	mtty_state_mutex_unlock(mdev_state);
+	return 0;
+}
+
+static int mtty_get_data_size(struct vfio_device *vdev,
+			      unsigned long *stop_copy_length)
+{
+	struct mdev_state *mdev_state =
+		container_of(vdev, struct mdev_state, vdev);
+
+	*stop_copy_length = mtty_data_size(mdev_state);
+	return 0;
+}
+
+static const struct vfio_migration_ops mtty_migration_ops = {
+	.migration_set_state = mtty_set_state,
+	.migration_get_state = mtty_get_state,
+	.migration_get_data_size = mtty_get_data_size,
+};
+
+static int mtty_log_start(struct vfio_device *vdev,
+			  struct rb_root_cached *ranges,
+			  u32 nnodes, u64 *page_size)
+{
+	return 0;
+}
+
+static int mtty_log_stop(struct vfio_device *vdev)
+{
+	return 0;
+}
+
+static int mtty_log_read_and_clear(struct vfio_device *vdev,
+				   unsigned long iova, unsigned long length,
+				   struct iova_bitmap *dirty)
+{
+	return 0;
+}
+
+static const struct vfio_log_ops mtty_log_ops = {
+	.log_start = mtty_log_start,
+	.log_stop = mtty_log_stop,
+	.log_read_and_clear = mtty_log_read_and_clear,
+};
+
 static int mtty_init_dev(struct vfio_device *vdev)
 {
 	struct mdev_state *mdev_state =
@@ -749,6 +1343,16 @@ static int mtty_init_dev(struct vfio_device *vdev)
 	mutex_init(&mdev_state->ops_lock);
 	mdev_state->mdev = mdev;
 	mtty_create_config_space(mdev_state);
+
+	mutex_init(&mdev_state->state_mutex);
+	mutex_init(&mdev_state->reset_mutex);
+	vdev->migration_flags = VFIO_MIGRATION_STOP_COPY |
+				VFIO_MIGRATION_P2P |
+				VFIO_MIGRATION_PRE_COPY;
+	vdev->mig_ops = &mtty_migration_ops;
+	vdev->log_ops = &mtty_log_ops;
+	mdev_state->state = VFIO_DEVICE_STATE_RUNNING;
+
 	return 0;
 
 err_nr_ports:
@@ -782,6 +1386,8 @@ static void mtty_release_dev(struct vfio_device *vdev)
 	struct mdev_state *mdev_state =
 		container_of(vdev, struct mdev_state, vdev);
 
+	mutex_destroy(&mdev_state->reset_mutex);
+	mutex_destroy(&mdev_state->state_mutex);
 	atomic_add(mdev_state->nr_ports, &mdev_avail_ports);
 	kfree(mdev_state->vconfig);
 }
@@ -798,6 +1404,15 @@ static int mtty_reset(struct mdev_state *mdev_state)
 {
 	pr_info("%s: called\n", __func__);
 
+	mutex_lock(&mdev_state->reset_mutex);
+	mdev_state->deferred_reset = true;
+	if (!mutex_trylock(&mdev_state->state_mutex)) {
+		mutex_unlock(&mdev_state->reset_mutex);
+		return 0;
+	}
+	mutex_unlock(&mdev_state->reset_mutex);
+	mtty_state_mutex_unlock(mdev_state);
+
 	return 0;
 }
 
@@ -921,6 +1536,25 @@ static ssize_t mtty_write(struct vfio_device *vdev, const char __user *buf,
 	return -EFAULT;
 }
 
+static void mtty_disable_intx(struct mdev_state *mdev_state)
+{
+	if (mdev_state->intx_evtfd) {
+		eventfd_ctx_put(mdev_state->intx_evtfd);
+		mdev_state->intx_evtfd = NULL;
+		mdev_state->intx_mask = false;
+		mdev_state->irq_index = -1;
+	}
+}
+
+static void mtty_disable_msi(struct mdev_state *mdev_state)
+{
+	if (mdev_state->msi_evtfd) {
+		eventfd_ctx_put(mdev_state->msi_evtfd);
+		mdev_state->msi_evtfd = NULL;
+		mdev_state->irq_index = -1;
+	}
+}
+
 static int mtty_set_irqs(struct mdev_state *mdev_state, uint32_t flags,
 			 unsigned int index, unsigned int start,
 			 unsigned int count, void *data)
@@ -932,59 +1566,113 @@ static int mtty_set_irqs(struct mdev_state *mdev_state, uint32_t flags,
 	case VFIO_PCI_INTX_IRQ_INDEX:
 		switch (flags & VFIO_IRQ_SET_ACTION_TYPE_MASK) {
 		case VFIO_IRQ_SET_ACTION_MASK:
+			if (!is_intx(mdev_state) || start != 0 || count != 1) {
+				ret = -EINVAL;
+				break;
+			}
+
+			if (flags & VFIO_IRQ_SET_DATA_NONE) {
+				mdev_state->intx_mask = true;
+			} else if (flags & VFIO_IRQ_SET_DATA_BOOL) {
+				uint8_t mask = *(uint8_t *)data;
+
+				if (mask)
+					mdev_state->intx_mask = true;
+			} else if (flags &  VFIO_IRQ_SET_DATA_EVENTFD) {
+				ret = -ENOTTY; /* No support for mask fd */
+			}
+			break;
 		case VFIO_IRQ_SET_ACTION_UNMASK:
+			if (!is_intx(mdev_state) || start != 0 || count != 1) {
+				ret = -EINVAL;
+				break;
+			}
+
+			if (flags & VFIO_IRQ_SET_DATA_NONE) {
+				mdev_state->intx_mask = false;
+			} else if (flags & VFIO_IRQ_SET_DATA_BOOL) {
+				uint8_t mask = *(uint8_t *)data;
+
+				if (mask)
+					mdev_state->intx_mask = false;
+			} else if (flags &  VFIO_IRQ_SET_DATA_EVENTFD) {
+				ret = -ENOTTY; /* No support for unmask fd */
+			}
 			break;
 		case VFIO_IRQ_SET_ACTION_TRIGGER:
-		{
-			if (flags & VFIO_IRQ_SET_DATA_NONE) {
-				pr_info("%s: disable INTx\n", __func__);
-				if (mdev_state->intx_evtfd)
-					eventfd_ctx_put(mdev_state->intx_evtfd);
+			if (is_intx(mdev_state) && !count &&
+			    (flags & VFIO_IRQ_SET_DATA_NONE)) {
+				mtty_disable_intx(mdev_state);
+				break;
+			}
+
+			if (!(is_intx(mdev_state) || is_noirq(mdev_state)) ||
+			    start != 0 || count != 1) {
+				ret = -EINVAL;
 				break;
 			}
 
 			if (flags & VFIO_IRQ_SET_DATA_EVENTFD) {
 				int fd = *(int *)data;
+				struct eventfd_ctx *evt;
 
-				if (fd > 0) {
-					struct eventfd_ctx *evt;
-
-					evt = eventfd_ctx_fdget(fd);
-					if (IS_ERR(evt)) {
-						ret = PTR_ERR(evt);
-						break;
-					}
-					mdev_state->intx_evtfd = evt;
-					mdev_state->irq_fd = fd;
-					mdev_state->irq_index = index;
+				mtty_disable_intx(mdev_state);
+
+				if (fd < 0)
+					break;
+
+				evt = eventfd_ctx_fdget(fd);
+				if (IS_ERR(evt)) {
+					ret = PTR_ERR(evt);
 					break;
 				}
+				mdev_state->intx_evtfd = evt;
+				mdev_state->irq_index = index;
+				break;
+			}
+
+			if (!is_intx(mdev_state)) {
+				ret = -EINVAL;
+				break;
+			}
+
+			if (flags & VFIO_IRQ_SET_DATA_NONE) {
+				mtty_trigger_interrupt(mdev_state);
+			} else if (flags & VFIO_IRQ_SET_DATA_BOOL) {
+				uint8_t trigger = *(uint8_t *)data;
+
+				if (trigger)
+					mtty_trigger_interrupt(mdev_state);
 			}
 			break;
 		}
-		}
 		break;
 	case VFIO_PCI_MSI_IRQ_INDEX:
 		switch (flags & VFIO_IRQ_SET_ACTION_TYPE_MASK) {
 		case VFIO_IRQ_SET_ACTION_MASK:
 		case VFIO_IRQ_SET_ACTION_UNMASK:
+			ret = -ENOTTY;
 			break;
 		case VFIO_IRQ_SET_ACTION_TRIGGER:
-			if (flags & VFIO_IRQ_SET_DATA_NONE) {
-				if (mdev_state->msi_evtfd)
-					eventfd_ctx_put(mdev_state->msi_evtfd);
-				pr_info("%s: disable MSI\n", __func__);
-				mdev_state->irq_index = VFIO_PCI_INTX_IRQ_INDEX;
+			if (is_msi(mdev_state) && !count &&
+			    (flags & VFIO_IRQ_SET_DATA_NONE)) {
+				mtty_disable_msi(mdev_state);
+				break;
+			}
+
+			if (!(is_msi(mdev_state) || is_noirq(mdev_state)) ||
+			    start != 0 || count != 1) {
+				ret = -EINVAL;
 				break;
 			}
+
 			if (flags & VFIO_IRQ_SET_DATA_EVENTFD) {
 				int fd = *(int *)data;
 				struct eventfd_ctx *evt;
 
-				if (fd <= 0)
-					break;
+				mtty_disable_msi(mdev_state);
 
-				if (mdev_state->msi_evtfd)
+				if (fd < 0)
 					break;
 
 				evt = eventfd_ctx_fdget(fd);
@@ -993,20 +1681,37 @@ static int mtty_set_irqs(struct mdev_state *mdev_state, uint32_t flags,
 					break;
 				}
 				mdev_state->msi_evtfd = evt;
-				mdev_state->irq_fd = fd;
 				mdev_state->irq_index = index;
+				break;
+			}
+
+			if (!is_msi(mdev_state)) {
+				ret = -EINVAL;
+				break;
+			}
+
+			if (flags & VFIO_IRQ_SET_DATA_NONE) {
+				mtty_trigger_interrupt(mdev_state);
+			} else if (flags & VFIO_IRQ_SET_DATA_BOOL) {
+				uint8_t trigger = *(uint8_t *)data;
+
+				if (trigger)
+					mtty_trigger_interrupt(mdev_state);
 			}
 			break;
-	}
-	break;
+		}
+		break;
 	case VFIO_PCI_MSIX_IRQ_INDEX:
-		pr_info("%s: MSIX_IRQ\n", __func__);
+		dev_dbg(mdev_state->vdev.dev, "%s: MSIX_IRQ\n", __func__);
+		ret = -ENOTTY;
 		break;
 	case VFIO_PCI_ERR_IRQ_INDEX:
-		pr_info("%s: ERR_IRQ\n", __func__);
+		dev_dbg(mdev_state->vdev.dev, "%s: ERR_IRQ\n", __func__);
+		ret = -ENOTTY;
 		break;
 	case VFIO_PCI_REQ_IRQ_INDEX:
-		pr_info("%s: REQ_IRQ\n", __func__);
+		dev_dbg(mdev_state->vdev.dev, "%s: REQ_IRQ\n", __func__);
+		ret = -ENOTTY;
 		break;
 	}
 
@@ -1014,33 +1719,6 @@ static int mtty_set_irqs(struct mdev_state *mdev_state, uint32_t flags,
 	return ret;
 }
 
-static int mtty_trigger_interrupt(struct mdev_state *mdev_state)
-{
-	int ret = -1;
-
-	if ((mdev_state->irq_index == VFIO_PCI_MSI_IRQ_INDEX) &&
-	    (!mdev_state->msi_evtfd))
-		return -EINVAL;
-	else if ((mdev_state->irq_index == VFIO_PCI_INTX_IRQ_INDEX) &&
-		 (!mdev_state->intx_evtfd)) {
-		pr_info("%s: Intr eventfd not found\n", __func__);
-		return -EINVAL;
-	}
-
-	if (mdev_state->irq_index == VFIO_PCI_MSI_IRQ_INDEX)
-		ret = eventfd_signal(mdev_state->msi_evtfd, 1);
-	else
-		ret = eventfd_signal(mdev_state->intx_evtfd, 1);
-
-#if defined(DEBUG_INTR)
-	pr_info("Intx triggered\n");
-#endif
-	if (ret != 1)
-		pr_err("%s: eventfd signal failed (%d)\n", __func__, ret);
-
-	return ret;
-}
-
 static int mtty_get_region_info(struct mdev_state *mdev_state,
 			 struct vfio_region_info *region_info,
 			 u16 *cap_type_id, void **cap_type)
@@ -1084,22 +1762,16 @@ static int mtty_get_region_info(struct mdev_state *mdev_state,
 
 static int mtty_get_irq_info(struct vfio_irq_info *irq_info)
 {
-	switch (irq_info->index) {
-	case VFIO_PCI_INTX_IRQ_INDEX:
-	case VFIO_PCI_MSI_IRQ_INDEX:
-	case VFIO_PCI_REQ_IRQ_INDEX:
-		break;
-
-	default:
+	if (irq_info->index != VFIO_PCI_INTX_IRQ_INDEX &&
+	    irq_info->index != VFIO_PCI_MSI_IRQ_INDEX)
 		return -EINVAL;
-	}
 
 	irq_info->flags = VFIO_IRQ_INFO_EVENTFD;
 	irq_info->count = 1;
 
 	if (irq_info->index == VFIO_PCI_INTX_IRQ_INDEX)
-		irq_info->flags |= (VFIO_IRQ_INFO_MASKABLE |
-				VFIO_IRQ_INFO_AUTOMASKED);
+		irq_info->flags |= VFIO_IRQ_INFO_MASKABLE |
+				   VFIO_IRQ_INFO_AUTOMASKED;
 	else
 		irq_info->flags |= VFIO_IRQ_INFO_NORESIZE;
 
@@ -1262,6 +1934,16 @@ static unsigned int mtty_get_available(struct mdev_type *mtype)
 	return atomic_read(&mdev_avail_ports) / type->nr_ports;
 }
 
+static void mtty_close(struct vfio_device *vdev)
+{
+	struct mdev_state *mdev_state =
+				container_of(vdev, struct mdev_state, vdev);
+
+	mtty_disable_files(mdev_state);
+	mtty_disable_intx(mdev_state);
+	mtty_disable_msi(mdev_state);
+}
+
 static const struct vfio_device_ops mtty_dev_ops = {
 	.name = "vfio-mtty",
 	.init = mtty_init_dev,
@@ -1273,6 +1955,7 @@ static const struct vfio_device_ops mtty_dev_ops = {
 	.unbind_iommufd	= vfio_iommufd_emulated_unbind,
 	.attach_ioas	= vfio_iommufd_emulated_attach_ioas,
 	.detach_ioas	= vfio_iommufd_emulated_detach_ioas,
+	.close_device	= mtty_close,
 };
 
 static struct mdev_driver mtty_driver = {