Re: [PATCH v6 1/6] iommu: Add cache_invalidate_user op

[Jason Gunthorpe <jgg@nvidia.com>]

On Fri, Nov 17, 2023 at 05:07:12AM -0800, Yi Liu wrote:
> +/**
> + * struct iommu_user_data_array - iommu driver specific user space data array
> + * @type: The data type of all the entries in the user buffer array
> + * @uptr: Pointer to the user buffer array for copy_from_user()
> + * @entry_len: The fixed-width length of a entry in the array, in bytes
> + * @entry_num: The number of total entries in the array
> + *
> + * A array having a @entry_num number of @entry_len sized entries, each entry is
> + * user space data, an uAPI defined in include/uapi/linux/iommufd.h where @type
> + * is also defined as enum iommu_xyz_data_type.
> + */
> +struct iommu_user_data_array {
> +	unsigned int type;
> +	void __user *uptr;
> +	size_t entry_len;
> +	int entry_num;

These are u32 in the uapi, they should probably be u32 here
too. Otherwise we have to worry about truncation.

> @@ -465,6 +492,9 @@ struct iommu_domain_ops {
>  			      size_t size);
>  	void (*iotlb_sync)(struct iommu_domain *domain,
>  			   struct iommu_iotlb_gather *iotlb_gather);
> +	int (*cache_invalidate_user)(struct iommu_domain *domain,
> +				     struct iommu_user_data_array *array,
> +				     u32 *error_code);

Regarding the other conversation I worry a u32 error_code is too small.

Unfortunately there is no obvious place to put something better so if
we reach it we will have to add more error_code space via normal
extension.

Maybe expand this to u64? That is 64 bits of error register data and
the consumer index. It should do for SMMUv3 at least?

Jason