RE: [PATCH v3 01/10] iommufd: Add data structure for Intel VT-d stage-1 domain allocation

[Zhang, Tina]

Hi,

> -----Original Message-----
> From: Yi Liu <yi.l.liu@xxxxxxxxx>
> Sent: Thursday, May 11, 2023 10:51 PM
> To: joro@xxxxxxxxxx; alex.williamson@xxxxxxxxxx; jgg@xxxxxxxxxx; Tian,
> Kevin <kevin.tian@xxxxxxxxx>; robin.murphy@xxxxxxx;
> baolu.lu@xxxxxxxxxxxxxxx
> Cc: cohuck@xxxxxxxxxx; eric.auger@xxxxxxxxxx; nicolinc@xxxxxxxxxx;
> kvm@xxxxxxxxxxxxxxx; mjrosato@xxxxxxxxxxxxx;
> chao.p.peng@xxxxxxxxxxxxxxx; Liu, Yi L <yi.l.liu@xxxxxxxxx>;
> yi.y.sun@xxxxxxxxxxxxxxx; peterx@xxxxxxxxxx; jasowang@xxxxxxxxxx;
> shameerali.kolothum.thodi@xxxxxxxxxx; lulu@xxxxxxxxxx;
> suravee.suthikulpanit@xxxxxxx; iommu@xxxxxxxxxxxxxxx; linux-
> kernel@xxxxxxxxxxxxxxx; linux-kselftest@xxxxxxxxxxxxxxx; Duan, Zhenzhong
> <zhenzhong.duan@xxxxxxxxx>
> Subject: [PATCH v3 01/10] iommufd: Add data structure for Intel VT-d stage-1
> domain allocation
> 
> This adds IOMMU_HWPT_TYPE_VTD_S1 for stage-1 hw_pagetable of Intel VT-
> d and the corressponding data structure for userspace specified parameter
> for the domain allocation.
> 
> Signed-off-by: Yi Liu <yi.l.liu@xxxxxxxxx>
> ---
>  include/uapi/linux/iommufd.h | 57
> ++++++++++++++++++++++++++++++++++++
>  1 file changed, 57 insertions(+)
> 
> diff --git a/include/uapi/linux/iommufd.h b/include/uapi/linux/iommufd.h
> index 06dcad6ab082..c2658394827a 100644
> --- a/include/uapi/linux/iommufd.h
> +++ b/include/uapi/linux/iommufd.h
> @@ -353,9 +353,64 @@ struct iommu_vfio_ioas {
>  /**
>   * enum iommu_hwpt_type - IOMMU HWPT Type
>   * @IOMMU_HWPT_TYPE_DEFAULT: default
> + * @IOMMU_HWPT_TYPE_VTD_S1: Intel VT-d stage-1 page table
>   */
>  enum iommu_hwpt_type {
>  	IOMMU_HWPT_TYPE_DEFAULT,
> +	IOMMU_HWPT_TYPE_VTD_S1,
> +};
> +
> +/**
> + * enum iommu_hwpt_intel_vtd_flags - Intel VT-d stage-1 page
> + *				     table entry attributes
> + * @IOMMU_VTD_PGTBL_SRE: Supervisor request
> + * @IOMMU_VTD_PGTBL_EAFE: Extended access enable
> + * @IOMMU_VTD_PGTBL_PCD: Page-level cache disable
> + * @IOMMU_VTD_PGTBL_PWT: Page-level write through
> + * @IOMMU_VTD_PGTBL_EMTE: Extended mem type enable
> + * @IOMMU_VTD_PGTBL_CD: PASID-level cache disable
> + * @IOMMU_VTD_PGTBL_WPE: Write protect enable  */ enum
> +iommu_hwpt_intel_vtd_flags {
> +	IOMMU_VTD_PGTBL_SRE = 1 << 0,
> +	IOMMU_VTD_PGTBL_EAFE = 1 << 1,
> +	IOMMU_VTD_PGTBL_PCD = 1 << 2,
> +	IOMMU_VTD_PGTBL_PWT = 1 << 3,
> +	IOMMU_VTD_PGTBL_EMTE = 1 << 4,
> +	IOMMU_VTD_PGTBL_CD = 1 << 5,
> +	IOMMU_VTD_PGTBL_WPE = 1 << 6,
> +	IOMMU_VTD_PGTBL_LAST = 1 << 7,
> +};
> +
> +/**
> + * struct iommu_hwpt_intel_vtd - Intel VT-d specific user-managed
> + *                               stage-1 page table info
> + * @flags: Combination of enum iommu_hwpt_intel_vtd_flags
> + * @pgtbl_addr: The base address of the user-managed stage-1 page table.
> + * @pat: Page attribute table data to compute effective memory type
> + * @emt: Extended memory type
> + * @addr_width: The address width of the untranslated addresses that are
> + *              subjected to the user-managed stage-1 page table.
> + * @__reserved: Must be 0
> + *
> + * The Intel VT-d specific data for creating hw_pagetable to represent
> + * the user-managed stage-1 page table that is used in nested translation.
> + *
> + * In nested translation, the stage-1 page table locates in the address
> + * space that defined by the corresponding stage-2 page table. Hence
> +the
> + * stage-1 page table base address value should not be higher than the
> + * maximum untranslated address of stage-2 page table.
> + *
> + * The paging level of the stage-1 page table should be compatible with
> + * the hardware iommu. Otherwise, the allocation would be failed.
> + */
> +struct iommu_hwpt_intel_vtd {
> +	__u64 flags;
> +	__u64 pgtbl_addr;
> +	__u32 pat;
> +	__u32 emt;
Do we need the emt field as part of the stage-1 page table info? IIUC, according to vt-d spec, the emt field is in stage-2 page table entries. Since in nested mode we only expose stage-1 page table to guest vt-d driver, I'm curious how does guest vt-d driver populate this EMT?

Thanks
-Tina
> +	__u32 addr_width;
> +	__u32 __reserved;
>  };
> 
>  /**
> @@ -391,6 +446,8 @@ enum iommu_hwpt_type {
>   * +------------------------------+-------------------------------------+-----------+
>   * | IOMMU_HWPT_TYPE_DEFAULT      |               N/A                   |    IOAS   |
>   * +------------------------------+-------------------------------------+-----------+
> + * | IOMMU_HWPT_TYPE_VTD_S1       |      struct iommu_hwpt_intel_vtd    |
> HWPT   |
> + *
> + +------------------------------+-------------------------------------+
> + -----------+
>   */
>  struct iommu_hwpt_alloc {
>  	__u32 size;
> --
> 2.34.1
>