Re: [PATCH v1 08/12] iommufd/viommu: Replace ops->viommu_alloc with ops->viommu_init

[Baolu Lu <baolu.lu@linux.intel.com>]

On 6/10/25 01:13, Nicolin Chen wrote:
> To ease the for-driver iommufd APIs, get_viommu_size and viommu_init ops
> are introduced. Now, those existing vIOMMU supported drivers implemented
> these two ops too.
> 
> Relace the ops->viommu_alloc call with the two new ones.
> 
> Note that this will fail a !viommu->ops case from now on, since a vIOMMU
> is expected to support alloc_domain_nested at least.

Does this mean that the viommu implementation in the iommu driver is
required to implement alloc_domain_nested? I suppose viommu should soon
be extended to support TEE/IO.

> 
> Suggested-by: Jason Gunthorpe<jgg@nvidia.com>
> Signed-off-by: Nicolin Chen<nicolinc@nvidia.com>
> ---
>   drivers/iommu/iommufd/viommu.c | 30 +++++++++++++++++++++++++++---
>   1 file changed, 27 insertions(+), 3 deletions(-)
> 

<...>

> @@ -68,6 +82,16 @@ int iommufd_viommu_alloc_ioctl(struct iommufd_ucmd *ucmd)
>   	 */
>   	viommu->iommu_dev = __iommu_get_iommu_dev(idev->dev);
>   
> +	rc = ops->viommu_init(viommu, hwpt_paging->common.domain);
> +	if (rc)
> +		goto out_abort;
> +
> +	/* It is a driver bug that viommu->ops isn't filled */
> +	if (WARN_ON_ONCE(!viommu->ops)) {
> +		rc = -EINVAL;
> +		goto out_abort;
> +	}
> +
>   	cmd->out_viommu_id = viommu->obj.id;
>   	rc = iommufd_ucmd_respond(ucmd, sizeof(*cmd));
>   	if (rc)

Thanks,
baolu