Re: [PATCH v1] fs/fuse: Fix missing FOLL_PIN for direct-io

[Miklos Szeredi]

On Wed, 6 Mar 2024 at 12:16, Bernd Schubert <bernd.schubert@xxxxxxxxxxx> wrote:
>
>
>
> On 3/6/24 11:01, Miklos Szeredi wrote:
> > On Tue, 29 Aug 2023 at 20:37, Lei Huang <lei.huang@xxxxxxxxxxxxxxx> wrote:
> >>
> >> Our user space filesystem relies on fuse to provide POSIX interface.
> >> In our test, a known string is written into a file and the content
> >> is read back later to verify correct data returned. We observed wrong
> >> data returned in read buffer in rare cases although correct data are
> >> stored in our filesystem.
> >>
> >> Fuse kernel module calls iov_iter_get_pages2() to get the physical
> >> pages of the user-space read buffer passed in read(). The pages are
> >> not pinned to avoid page migration. When page migration occurs, the
> >> consequence are two-folds.
> >>
> >> 1) Applications do not receive correct data in read buffer.
> >> 2) fuse kernel writes data into a wrong place.
> >>
> >> Using iov_iter_extract_pages() to pin pages fixes the issue in our
> >> test.
> >>
> >> An auxiliary variable "struct page **pt_pages" is used in the patch
> >> to prepare the 2nd parameter for iov_iter_extract_pages() since
> >> iov_iter_get_pages2() uses a different type for the 2nd parameter.
> >>
> >> Signed-off-by: Lei Huang <lei.huang@xxxxxxxxxxxxxxx>
> >
> > Applied, with a modification to only unpin if
> > iov_iter_extract_will_pin() returns true.
>
> Hi Miklos,
>
> do you have an idea if this needs to be back ported and to which kernel
> version?
> I had tried to reproduce data corruption with 4.18 - Lei wrote that he
> could see issues with older kernels as well, but I never managed to
> trigger anything on 4.18-RHEL. Typically I use ql-fstest
> (https://github.com/bsbernd/ql-fstest) and even added random DIO as an
> option - nothing report with weeks of run time. I could try again with
> more recent kernels that have folios.

I don't think that corruption will happen in real life.  So I'm not
sure we need to bother with backporting, and definitely not before
when the infrastructure was introduced.

Thanks,
Miklos