From: Zhou Wang <wangzhou1@hisilicon.com>
To: Greg KH <gregkh@linuxfoundation.org>
Cc: jean-philippe@linaro.org, kevin.tian@intel.com,
Sihang Chen <chensihang1@hisilicon.com>,
jgg@ziepe.ca, linux-api@vger.kernel.org,
linux-kernel@vger.kernel.org,
Andy Lutomirski <luto@amacapital.net>,
linux-mm@kvack.org, iommu@lists.linux-foundation.org,
Alexander Viro <viro@zeniv.linux.org.uk>,
zhangfei.gao@linaro.org,
Andrew Morton <akpm@linux-foundation.org>,
liguozhu@hisilicon.com, linux-arm-kernel@lists.infradead.org
Subject: Re: [RFC PATCH v3 1/2] mempinfd: Add new syscall to provide memory pin
Date: Tue, 9 Feb 2021 20:20:18 +0800 [thread overview]
Message-ID: <2237506a-0c98-7ba6-5d5f-b60b637174c5@hisilicon.com> (raw)
In-Reply-To: <YCJ5k/Bxxkg3BNNj@kroah.com>
On 2021/2/9 20:01, Greg KH wrote:
> On Tue, Feb 09, 2021 at 07:58:15PM +0800, Zhou Wang wrote:
>> On 2021/2/9 17:37, Greg KH wrote:
>>> On Tue, Feb 09, 2021 at 05:17:46PM +0800, Zhou Wang wrote:
>>>> On 2021/2/8 6:02, Andy Lutomirski wrote:
>>>>>
>>>>>
>>>>>> On Feb 7, 2021, at 12:31 AM, Zhou Wang <wangzhou1@hisilicon.com> wrote:
>>>>>>
>>>>>> SVA(share virtual address) offers a way for device to share process virtual
>>>>>> address space safely, which makes more convenient for user space device
>>>>>> driver coding. However, IO page faults may happen when doing DMA
>>>>>> operations. As the latency of IO page fault is relatively big, DMA
>>>>>> performance will be affected severely when there are IO page faults.
>>>>>> From a long term view, DMA performance will be not stable.
>>>>>>
>>>>>> In high-performance I/O cases, accelerators might want to perform
>>>>>> I/O on a memory without IO page faults which can result in dramatically
>>>>>> increased latency. Current memory related APIs could not achieve this
>>>>>> requirement, e.g. mlock can only avoid memory to swap to backup device,
>>>>>> page migration can still trigger IO page fault.
>>>>>>
>>>>>> Various drivers working under traditional non-SVA mode are using
>>>>>> their own specific ioctl to do pin. Such ioctl can be seen in v4l2,
>>>>>> gpu, infiniband, media, vfio, etc. Drivers are usually doing dma
>>>>>> mapping while doing pin.
>>>>>>
>>>>>> But, in SVA mode, pin could be a common need which isn't necessarily
>>>>>> bound with any drivers, and neither is dma mapping needed by drivers
>>>>>> since devices are using the virtual address of CPU. Thus, It is better
>>>>>> to introduce a new common syscall for it.
>>>>>>
>>>>>> This patch leverages the design of userfaultfd and adds mempinfd for pin
>>>>>> to avoid messing up mm_struct. A fd will be got by mempinfd, then user
>>>>>> space can do pin/unpin pages by ioctls of this fd, all pinned pages under
>>>>>> one file will be unpinned in file release process. Like pin page cases in
>>>>>> other places, can_do_mlock is used to check permission and input
>>>>>> parameters.
>>>>>
>>>>>
>>>>> Can you document what the syscall does?
>>>>
>>>> Will add related document in Documentation/vm.
>>>
>>> A manpage is always good, and will be required eventually :)
>>
>> manpage is maintained in another repo. Do you mean add a manpage
>> patch in this series?
>
> It's good to show how it will be used, don't you think?
Agree, will add it in next version.
Thanks,
Zhou
>
> thanks,
>
> greg k-h
>
> .
>
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu
next prev parent reply other threads:[~2021-02-09 12:20 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-02-07 8:18 [RFC PATCH v3 0/2] mempinfd: Add new syscall to provide memory pin Zhou Wang
2021-02-07 8:18 ` [RFC PATCH v3 1/2] " Zhou Wang
2021-02-07 21:34 ` Matthew Wilcox
2021-02-07 22:24 ` Song Bao Hua (Barry Song)
2021-02-08 1:30 ` Matthew Wilcox
2021-02-08 2:27 ` Song Bao Hua (Barry Song)
2021-02-08 8:21 ` David Hildenbrand
2021-02-08 10:13 ` Song Bao Hua (Barry Song)
2021-02-08 10:37 ` David Hildenbrand
2021-02-08 20:52 ` Song Bao Hua (Barry Song)
2021-02-08 2:18 ` David Rientjes via iommu
2021-02-08 5:34 ` Song Bao Hua (Barry Song)
2021-02-07 21:51 ` Arnd Bergmann
2021-02-09 9:27 ` Zhou Wang
2021-02-07 22:02 ` Andy Lutomirski
2021-02-09 9:17 ` Zhou Wang
2021-02-09 9:37 ` Greg KH
2021-02-09 11:58 ` Zhou Wang
2021-02-09 12:01 ` Greg KH
2021-02-09 12:20 ` Zhou Wang [this message]
2021-02-10 18:50 ` Matthew Wilcox
2021-02-08 8:14 ` David Hildenbrand
2021-02-08 18:33 ` Jason Gunthorpe
2021-02-08 20:35 ` Song Bao Hua (Barry Song)
2021-02-08 21:30 ` Jason Gunthorpe
2021-02-09 3:01 ` Song Bao Hua (Barry Song)
2021-02-09 13:53 ` Jason Gunthorpe
2021-02-09 22:22 ` Song Bao Hua (Barry Song)
2021-02-10 18:04 ` Jason Gunthorpe
2021-02-10 21:39 ` Song Bao Hua (Barry Song)
2021-02-11 10:28 ` David Hildenbrand
2021-02-07 8:18 ` [RFC PATCH v3 2/2] selftests/vm: add mempinfd test Zhou Wang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=2237506a-0c98-7ba6-5d5f-b60b637174c5@hisilicon.com \
--to=wangzhou1@hisilicon.com \
--cc=akpm@linux-foundation.org \
--cc=chensihang1@hisilicon.com \
--cc=gregkh@linuxfoundation.org \
--cc=iommu@lists.linux-foundation.org \
--cc=jean-philippe@linaro.org \
--cc=jgg@ziepe.ca \
--cc=kevin.tian@intel.com \
--cc=liguozhu@hisilicon.com \
--cc=linux-api@vger.kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=luto@amacapital.net \
--cc=viro@zeniv.linux.org.uk \
--cc=zhangfei.gao@linaro.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).