[Intel-gfx] [PATCH v4 09/19] vfio/pci: Accept device fd for hot reset

[Tian, Kevin]

> From: Liu, Yi L <yi.l.liu at intel.com>
> Sent: Tuesday, February 21, 2023 11:48 AM
> 
>  	/*
>  	 * We can't let userspace give us an arbitrarily large buffer to copy,
> -	 * so verify how many we think there could be.  Note groups can have
> -	 * multiple devices so one group per device is the max.
> +	 * so verify how many we think there could be.  Note user may
> provide
> +	 * a set of groups, group can have multiple devices so one group per
> +	 * device is the max.

well this change doesn't include cdev

> @@ -1320,7 +1321,7 @@ static int vfio_pci_ioctl_pci_hot_reset(struct
> vfio_pci_core_device *vdev,
>  		}
> 
>  		/* Ensure the FD is a vfio FD.*/
> -		if (!vfio_file_is_valid(file)) {
> +		if (!vfio_file_is_device_opened(file)) {
>  			fput(file);
>  			ret = -EINVAL;
>  			break;

that function is not just for checking device.

Probably rename it to vfio_file_is_reset_valid().

btw this patch is insufficient to handle device fd. The current logic
requires every device in the dev_set covered by provided fd's:

static bool vfio_dev_in_groups(struct vfio_pci_core_device *vdev,
                               struct vfio_pci_group_info *groups)
{
 	unsigned int i;

	for (i = 0; i < groups->count; i++)
		if (vfio_file_has_dev(groups->files[i], &vdev->vdev))
			return true;
	return false;
}

Presumably when cdev fd is provided above should compare iommu
group of the fd and that of the vdev. Otherwise it expects the user
to have full access to every device in the set which is impractical.