Re: [PATCH v6 2/6] iommufd: Add IOMMU_HWPT_INVALIDATE

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

On 2023/11/20 16:09, Tian, Kevin wrote:
>> From: Liu, Yi L <yi.l.liu@intel.com>
>> Sent: Friday, November 17, 2023 9:07 PM
>> +
>> +	hwpt = iommufd_hw_pagetable_get_nested(ucmd, cmd->hwpt_id);
>> +	if (IS_ERR(hwpt))
>> +		return PTR_ERR(hwpt);
>> +
>> +	rc = hwpt->domain->ops->cache_invalidate_user(hwpt->domain,
>> &data_array,
>> +						      &cmd-
>>> out_driver_error_code);
>> +	cmd->req_num = data_array.entry_num;
>> +	if (iommufd_ucmd_respond(ucmd, sizeof(*cmd)))
>> +		return -EFAULT;
>> +	iommufd_put_object(&hwpt->obj);
> 
> the object is not put when ucmd response fails. It can be put right
> after calling @cache_invalidate_user()

yes.

>> @@ -309,6 +309,7 @@ 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;
> 
> In alphabetic order this should be put before "hwpt_set_dirty"

yes.

>>   	struct iommu_ioas_alloc alloc;
>>   	struct iommu_ioas_allow_iovas allow_iovas;
>>   	struct iommu_ioas_copy ioas_copy;
>> @@ -348,6 +349,8 @@ static const struct iommufd_ioctl_op
>> iommufd_ioctl_ops[] = {
>>   		 struct iommu_hwpt_get_dirty_bitmap, data),
>>   	IOCTL_OP(IOMMU_HWPT_SET_DIRTY_TRACKING,
>> iommufd_hwpt_set_dirty_tracking,
>>   		 struct iommu_hwpt_set_dirty_tracking, __reserved),
>> +	IOCTL_OP(IOMMU_HWPT_INVALIDATE, iommufd_hwpt_invalidate,
>> +		 struct iommu_hwpt_invalidate, out_driver_error_code),
> 
> ditto
> 
>> +/**
>> + * 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
> 
> remove the first 'HWPT'

>> + * @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.
>> + * @req_type: One of enum iommu_hwpt_data_type, defining the data
>> type of all
>> + *            the entries in the invalidation request array. It should suit
>> + *            with the data_type passed per the allocation of the hwpt pointed
>> + *            by @hwpt_id.
> 
> "It should match the data_type given to the specified hwpt"

above comments are well received. :)

>> + * @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.
>> + *           Output the number of requests successfully handled by kernel.
>> + * @out_driver_error_code: Report a driver speicifc error code upon failure.
>> + *                         It's optional, driver has a choice to fill it or
>> + *                         not.
> 
> Being optional how does the user tell whether the code is filled or not?

seems like we need a flag for it. otherwise, a reserved special value is
required. or is it enough to just document it that this field is available
or not per the iommu_hw_info_type?

>> + *
>> + * 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.
>> + */
>> +struct iommu_hwpt_invalidate {
>> +	__u32 size;
>> +	__u32 hwpt_id;
>> +	__aligned_u64 reqs_uptr;
>> +	__u32 req_type;
>> +	__u32 req_len;
>> +	__u32 req_num;
>> +	__u32 out_driver_error_code;
>> +};
> 
> Probably 'data_uptr', 'data_type', "entry_len' and 'entry_num" read slightly
> better.

sure.

-- 
Regards,
Yi Liu