Re: [PATCH v2] vfio iommu type1: Improve vfio_iommu_type1_pin_pages performance

["xuxiaoyang (C)" <xuxiaoyang2@huawei.com>]

On 2020/12/9 22:42, Eric Farman wrote:
> 
> 
> On 12/9/20 6:54 AM, Cornelia Huck wrote:
>> On Tue, 8 Dec 2020 21:55:53 +0800
>> "xuxiaoyang (C)" <xuxiaoyang2@huawei.com> wrote:
>>
>>> On 2020/11/21 15:58, xuxiaoyang (C) wrote:
>>>> vfio_pin_pages() accepts an array of unrelated iova pfns and processes
>>>> each to return the physical pfn.  When dealing with large arrays of
>>>> contiguous iovas, vfio_iommu_type1_pin_pages is very inefficient because
>>>> it is processed page by page.In this case, we can divide the iova pfn
>>>> array into multiple continuous ranges and optimize them.  For example,
>>>> when the iova pfn array is {1,5,6,7,9}, it will be divided into three
>>>> groups {1}, {5,6,7}, {9} for processing.  When processing {5,6,7}, the
>>>> number of calls to pin_user_pages_remote is reduced from 3 times to once.
>>>> For single page or large array of discontinuous iovas, we still use
>>>> vfio_pin_page_external to deal with it to reduce the performance loss
>>>> caused by refactoring.
>>>>
>>>> Signed-off-by: Xiaoyang Xu <xuxiaoyang2@huawei.com>
>>
>> (...)
>>
>>>
>>> hi Cornelia Huck, Eric Farman, Zhenyu Wang, Zhi Wang
>>>
>>> vfio_pin_pages() accepts an array of unrelated iova pfns and processes
>>> each to return the physical pfn.  When dealing with large arrays of
>>> contiguous iovas, vfio_iommu_type1_pin_pages is very inefficient because
>>> it is processed page by page.  In this case, we can divide the iova pfn
>>> array into multiple continuous ranges and optimize them.  I have a set
>>> of performance test data for reference.
>>>
>>> The patch was not applied
>>>                      1 page           512 pages
>>> no huge pages：     1638ns           223651ns
>>> THP：               1668ns           222330ns
>>> HugeTLB：           1526ns           208151ns
>>>
>>> The patch was applied
>>>                      1 page           512 pages
>>> no huge pages       1735ns           167286ns
>>> THP：               1934ns           126900ns
>>> HugeTLB：           1713ns           102188ns
>>>
>>> As Alex Williamson said, this patch lacks proof that it works in the
>>> real world. I think you will have some valuable opinions.
>>
>> Looking at this from the vfio-ccw angle, I'm not sure how much this
>> would buy us, as we deal with IDAWs, which are designed so that they
>> can be non-contiguous. I guess this depends a lot on what the guest
>> does.
> 
> This would be my concern too, but I don't have data off the top of my head to say one way or another...
> 
>>
>> Eric, any opinion? Do you maybe also happen to have a test setup that
>> mimics workloads actually seen in the real world?
>>
> 
> ...I do have some test setups, which I will try to get some data from in a couple days. At the moment I've broken most of those setups trying to implement some other stuff, and can't revert back at the moment. Will get back to this.
> 
> Eric
> .

Thank you for your reply. Looking forward to your test data.

Regards,
Xu