[dm-devel] [PATCH 4/5] dax: remove the copy_from_iter and copy_to_iter methods

[Dan Williams]

On Fri, Dec 10, 2021 at 6:17 AM Vivek Goyal <vgoyal at redhat.com> wrote:
>
> On Thu, Dec 09, 2021 at 07:38:27AM +0100, Christoph Hellwig wrote:
> > These methods indirect the actual DAX read/write path.  In the end pmem
> > uses magic flush and mc safe variants and fuse and dcssblk use plain ones
> > while device mapper picks redirects to the underlying device.
> >
> > Add set_dax_virtual() and set_dax_nomcsafe() APIs for fuse to skip these
> > special variants, then use them everywhere as they fall back to the plain
> > ones on s390 anyway and remove an indirect call from the read/write path
> > as well as a lot of boilerplate code.
> >
> > Signed-off-by: Christoph Hellwig <hch at lst.de>
> > ---
> >  drivers/dax/super.c           | 36 ++++++++++++++--
> >  drivers/md/dm-linear.c        | 20 ---------
> >  drivers/md/dm-log-writes.c    | 80 -----------------------------------
> >  drivers/md/dm-stripe.c        | 20 ---------
> >  drivers/md/dm.c               | 50 ----------------------
> >  drivers/nvdimm/pmem.c         | 20 ---------
> >  drivers/s390/block/dcssblk.c  | 14 ------
> >  fs/dax.c                      |  5 ---
> >  fs/fuse/virtio_fs.c           | 19 +--------
> >  include/linux/dax.h           |  9 ++--
> >  include/linux/device-mapper.h |  4 --
> >  11 files changed, 37 insertions(+), 240 deletions(-)
> >
>
> [..]
> > diff --git a/fs/fuse/virtio_fs.c b/fs/fuse/virtio_fs.c
> > index 5c03a0364a9bb..754319ce2a29b 100644
> > --- a/fs/fuse/virtio_fs.c
> > +++ b/fs/fuse/virtio_fs.c
> > @@ -753,20 +753,6 @@ static long virtio_fs_direct_access(struct dax_device *dax_dev, pgoff_t pgoff,
> >       return nr_pages > max_nr_pages ? max_nr_pages : nr_pages;
> >  }
> >
> > -static size_t virtio_fs_copy_from_iter(struct dax_device *dax_dev,
> > -                                    pgoff_t pgoff, void *addr,
> > -                                    size_t bytes, struct iov_iter *i)
> > -{
> > -     return copy_from_iter(addr, bytes, i);
> > -}
> > -
> > -static size_t virtio_fs_copy_to_iter(struct dax_device *dax_dev,
> > -                                    pgoff_t pgoff, void *addr,
> > -                                    size_t bytes, struct iov_iter *i)
> > -{
> > -     return copy_to_iter(addr, bytes, i);
> > -}
> > -
> >  static int virtio_fs_zero_page_range(struct dax_device *dax_dev,
> >                                    pgoff_t pgoff, size_t nr_pages)
> >  {
> > @@ -783,8 +769,6 @@ static int virtio_fs_zero_page_range(struct dax_device *dax_dev,
> >
> >  static const struct dax_operations virtio_fs_dax_ops = {
> >       .direct_access = virtio_fs_direct_access,
> > -     .copy_from_iter = virtio_fs_copy_from_iter,
> > -     .copy_to_iter = virtio_fs_copy_to_iter,
> >       .zero_page_range = virtio_fs_zero_page_range,
> >  };
> >
> > @@ -853,7 +837,8 @@ static int virtio_fs_setup_dax(struct virtio_device *vdev, struct virtio_fs *fs)
> >       fs->dax_dev = alloc_dax(fs, &virtio_fs_dax_ops);
> >       if (IS_ERR(fs->dax_dev))
> >               return PTR_ERR(fs->dax_dev);
> > -
> > +     set_dax_cached(fs->dax_dev);
>
> Looks good to me from virtiofs point of view.
>
> Reviewed-by: Vivek Goyal <vgoyal at redhat.com>
>
> Going forward, I am wondering should virtiofs use flushcache version as
> well. What if host filesystem is using DAX and mapping persistent memory
> pfn directly into qemu address space. I have never tested that.
>
> Right now we are relying on applications to do fsync/msync on virtiofs
> for data persistence.

This sounds like it would need coordination with a paravirtualized
driver that can indicate whether the host side is pmem or not, like
the virtio_pmem driver. However, if the guest sends any fsync/msync
you would still need to go explicitly cache flush any dirty page
because you can't necessarily trust that the guest did that already.

>
> > +     set_dax_nomcsafe(fs->dax_dev);
> >       return devm_add_action_or_reset(&vdev->dev, virtio_fs_cleanup_dax,
> >                                       fs->dax_dev);
> >  }
>
> Thanks
> Vivek
>
>