Re: [PATCH v7 2/9] iommufd: Add IOMMU_HWPT_INVALIDATE

[Yi Liu <yi.l.liu@intel.com>]

On 2023/12/22 11:19, Tian, Kevin wrote:
>> From: Liu, Yi L <yi.l.liu@intel.com>
>> Sent: Thursday, December 21, 2023 11:40 PM
>> +
>> +int iommufd_hwpt_invalidate(struct iommufd_ucmd *ucmd)
>> +{
>> +	struct iommu_hwpt_invalidate *cmd = ucmd->cmd;
>> +	struct iommu_user_data_array data_array = {
>> +		.type = cmd->req_type,
>> +		.uptr = u64_to_user_ptr(cmd->reqs_uptr),
>> +		.entry_len = cmd->req_len,
>> +		.entry_num = cmd->req_num,
>> +	};
>> +	struct iommufd_hw_pagetable *hwpt;
>> +	u32 done_num = 0;
>> +	int rc;
>> +
>> +	if (cmd->req_num && (!cmd->reqs_uptr || !cmd->req_len)) {
>> +		rc = -EINVAL;
>> +		goto out;
>> +	}
> 
> miss a check on the __reserved field.

done.

> 
>> @@ -323,6 +323,7 @@ union ucmd_buffer {
>>   	struct iommu_hwpt_alloc hwpt;
>>   	struct iommu_hwpt_get_dirty_bitmap get_dirty_bitmap;
>>   	struct iommu_hwpt_set_dirty_tracking set_dirty_tracking;
>> +	struct iommu_hwpt_invalidate cache;
> 
> this should be in alphabetic order. I gave this comment in v6 too:
> 
> https://lore.kernel.org/linux-iommu/BN9PR11MB5276D8406BF08B853329288C8CB4A@BN9PR11MB5276.namprd11.prod.outlook.com/

oops, I did miss it. sigh, I paid too much attention to the error code
discussion in the end. :( I've revisited the comments and should have
done now.

>> +/**
>> + * enum iommu_hwpt_invalidate_data_type - IOMMU HWPT Cache
>> Invalidation
>> + *                                        Data Type
>> + * @IOMMU_HWPT_INVALIDATE_DATA_VTD_S1: Invalidation data for
>> VTD_S1
>> + */
>> +enum iommu_hwpt_invalidate_data_type {
>> +	IOMMU_HWPT_INVALIDATE_DATA_VTD_S1,
>> +};
> 
> Defining DATA_VTD_S1 at this point is fine, if there is no usage on
> DATA_NONE. But following vtd specific definitions should be moved
> to the later vtd specific patches. they are not used by the common
> code anyway.

sure.

>> +
>> +/**
>> + * enum iommu_hwpt_vtd_s1_invalidate_flags - Flags for Intel VT-d
>> + *                                           stage-1 cache invalidation
>> + * @IOMMU_VTD_INV_FLAGS_LEAF: The LEAF flag indicates whether only
>> the
>> + *                            leaf PTE caching needs to be invalidated
>> + *                            and other paging structure caches can be
>> + *                            preserved.
> 
> "indicates whether the invalidation applies to all-levels page structure
> cache or just the leaf PTE cache"

done.

>> + */
>> +enum iommu_hwpt_vtd_s1_invalidate_flags {
>> +	IOMMU_VTD_INV_FLAGS_LEAF = 1 << 0,
>> +};
>> +
>> +/**
>> + * enum iommu_hwpt_vtd_s1_invalidate_error - Result of invalidation
> 
> "hardware error of invalidation"

done.

> 
>> + * @IOMMU_HWPT_INVALIDATE_VTD_S1_ICE: Invalidation Completion
>> Error, details
>> + *                                     refer to 11.4.7.1 Fault Status Register
>> + *                                     of VT-d specification.
>> + * @IOMMU_HWPT_INVALIDATE_VTD_S1_ITE: Invalidation Time-out Error,
>> details
>> + *                                     refer to 11.4.7.1 Fault Status Register
>> + *                                     of VT-d specification.
>> + */
>> +enum iommu_hwpt_vtd_s1_invalidate_error {
>> +	IOMMU_HWPT_INVALIDATE_VTD_S1_ICE = 1 << 0,
>> +	IOMMU_HWPT_INVALIDATE_VTD_S1_ITE = 1 << 1,
>> +};
>> +
>> +/**
>> + * struct iommu_hwpt_vtd_s1_invalidate - Intel VT-d cache invalidation
>> + *                                       (IOMMU_HWPT_INVALIDATE_DATA_VTD_S1)
>> + * @addr: The start address of the addresses to be invalidated. It needs
>> + *        to be 4KB aligned.
> 
> 'of the range'

done.

> 
>> + * @npages: Number of contiguous 4K pages to be invalidated.
>> + * @flags: Combination of enum iommu_hwpt_vtd_s1_invalidate_flags
>> + * @inv_error: One of enum iommu_hwpt_vtd_s1_invalidate_error
> 
> '@hw_error'

done.

>> + *
>> + * The Intel VT-d specific invalidation data for user-managed stage-1 cache
>> + * invalidation in nested translation. Userspace uses this structure to
>> + * tell the impacted cache scope after modifying the stage-1 page table.
>> + *
>> + * Invalidating all the caches related to the page table by setting @addr
>> + * to be 0 and @npages to be U64_MAX.
> 
> here should clarify that the invalidation applies to device TLB automatically
> for VT-d.

done.

>> + *
>> + * @inv_error is meaningful only if the request is handled by kernel. This
>> + * can be known by checking struct iommu_hwpt_invalidate::req_num
>> output.
>> + * @inv_error only covers the errors detected by hardware after submitting
>> the
>> + * invalidation. The software detected errors would go through the normal
>> + * ioctl errno.
>> + */
>> +struct iommu_hwpt_vtd_s1_invalidate {
>> +	__aligned_u64 addr;
>> +	__aligned_u64 npages;
>> +	__u32 flags;
>> +	__u32 inv_error;
>> +};
>> +
>> +/**
>> + * struct iommu_hwpt_invalidate - ioctl(IOMMU_HWPT_INVALIDATE)
>> + * @size: sizeof(struct iommu_hwpt_invalidate)
>> + * @hwpt_id: HWPT ID of a nested HWPT for cache invalidation
>> + * @reqs_uptr: User pointer to an array having @req_num of cache
>> invalidation
>> + *             requests. The request entries in the array are of fixed width
>> + *             @req_len, and contain a user data structure for invalidation
>> + *             request specific to the given hardware page table.
> 
> Just:
> 
> 'User pointer to an array of driver-specific cache invalidation requests'

done.

>> + * @req_type: One of enum iommu_hwpt_invalidate_data_type, defining
>> the data
>> + *            type of all the entries in the invalidation request array. It
>> + *            should be a type supported by the hwpt pointed by @hwpt_id.
>> + * @req_len: Length (in bytes) of a request entry in the request array
>> + * @req_num: Input the number of cache invalidation requests in the array.

s/req_len/entry_len, s/req_num/entry_num per the comments in last version.

>> + *           Output the number of requests successfully handled by kernel.
>> + * @__reserved: Must be 0.
>> + *
>> + * Invalidate the iommu cache for user-managed page table. Modifications
>> on a
>> + * user-managed page table should be followed by this operation to sync
>> cache.
>> + * Each ioctl can support one or more cache invalidation requests in the
>> array
>> + * that has a total size of @req_len * @req_num.
>> + *
>> + * An empty invalidation request array by setting @req_num==0 is allowed,
>> and
>> + * @req_len and @reqs_uptr would be ignored in this case. This can be
>> used to
>> + * check if the given @req_type is supported or not by kernel.
>> + */
>> +struct iommu_hwpt_invalidate {
>> +	__u32 size;
>> +	__u32 hwpt_id;
>> +	__aligned_u64 reqs_uptr;
>> +	__u32 req_type;
>> +	__u32 req_len;
>> +	__u32 req_num;
>> +	__u32 __reserved;
>> +};
>> +#define IOMMU_HWPT_INVALIDATE _IO(IOMMUFD_TYPE,
>> IOMMUFD_CMD_HWPT_INVALIDATE)
>>   #endif
>> --
>> 2.34.1
> 

-- 
Regards,
Yi Liu