Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.36 as permitted sender) client-ip=23.128.96.36;
MIME-Version: 1.0
References: <20231205083444.3029239-1-lulu@redhat.com> <20231205083444.3029239-6-lulu@redhat.com>
In-Reply-To: <20231205083444.3029239-6-lulu@redhat.com>
From:   Jason Wang <jasowang@redhat.com>
Date:   Thu, 7 Dec 2023 16:46:12 +0800
Message-ID: <CACGkMEv+myMJ1MXNz5RAfggkYy-0Q-VVyZKkpg_DD+_wfMxFQQ@mail.gmail.com>
Subject: Re: [PATCH v3 5/7] vduse: Add file operation for mmap
To:     Cindy Lu <lulu@redhat.com>
Cc:     mst@redhat.com, xieyongji@bytedance.com,
        linux-kernel@vger.kernel.org, maxime.coquelin@redhat.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Tue, Dec 5, 2023 at 4:35=E2=80=AFPM Cindy Lu <lulu@redhat.com> wrote:
>
> Add the operation for mmap, This function  will be used by the user space
> application to map the pages to the user space.
>
> Signed-off-by: Cindy Lu <lulu@redhat.com>
> ---
>  drivers/vdpa/vdpa_user/vduse_dev.c | 75 ++++++++++++++++++++++++++++++
>  1 file changed, 75 insertions(+)
>
> diff --git a/drivers/vdpa/vdpa_user/vduse_dev.c b/drivers/vdpa/vdpa_user/=
vduse_dev.c
> index 52ccde636406..f55f415629de 100644
> --- a/drivers/vdpa/vdpa_user/vduse_dev.c
> +++ b/drivers/vdpa/vdpa_user/vduse_dev.c
> @@ -1381,6 +1381,79 @@ static struct vduse_dev *vduse_dev_get_from_minor(=
int minor)
>         return dev;
>  }
>
> +static vm_fault_t vduse_vm_fault(struct vm_fault *vmf)
> +{
> +       struct vduse_dev *dev =3D vmf->vma->vm_file->private_data;
> +       struct vm_area_struct *vma =3D vmf->vma;
> +       u16 index =3D vma->vm_pgoff;
> +       struct vduse_virtqueue *vq;
> +       unsigned long vaddr;
> +
> +       /* index 0  page reserved for vduse status*/

I'd suggest to tweak it as "reserved for device state"

> +       if (index =3D=3D 0) {
> +               vaddr =3D dev->vdpa_reconnect_vaddr;
> +       } else {
> +               /* index 1+vq_number page reserved for vduse vqs*/

And "reserved for virtqueue state"

> +               vq =3D &dev->vqs[index - 1];
> +               vaddr =3D vq->vdpa_reconnect_vaddr;
> +       }
> +       if (remap_pfn_range(vma, vmf->address & PAGE_MASK,
> +                           PFN_DOWN(virt_to_phys((void *)vaddr)),
> +                           VDUSE_RECONNCT_MMAP_SIZE, vma->vm_page_prot))

If there's no chance for VDUSE_RECONNCT_MMAP_SIZE to differ from
PAGE_SIZE. Let's just use PAGE_SIZE here.

> +               return VM_FAULT_SIGBUS;
> +       return VM_FAULT_NOPAGE;
> +}
> +
> +static const struct vm_operations_struct vduse_vm_ops =3D {
> +       .fault =3D vduse_vm_fault,
> +};
> +
> +static int vduse_dev_mmap(struct file *file, struct vm_area_struct *vma)
> +{
> +       struct vduse_dev *dev =3D file->private_data;
> +       unsigned long vaddr =3D 0;
> +       unsigned long index =3D vma->vm_pgoff;
> +       struct vduse_virtqueue *vq;
> +
> +       if (vma->vm_end - vma->vm_start !=3D PAGE_SIZE)
> +               return -EINVAL;
> +       if ((vma->vm_flags & VM_SHARED) =3D=3D 0)
> +               return -EINVAL;
> +
> +       /*check if Userspace App map the page number larger than kernel a=
llocated*/

Code explains itself, let's just drop the comment please.

> +       if (index > dev->vq_num + 1)
> +               return -EINVAL;
> +
> +       /* index 0  page reserved for vduse status*/

This comment duplicates the one in fault. I'd suggest to move it to uAPI.

> +       if (index =3D=3D 0) {
> +               vaddr =3D dev->vdpa_reconnect_vaddr;
> +       } else {
> +               /* index 1+vq_number page reserved for vduse vqs*/
> +               vq =3D &dev->vqs[index - 1];
> +               vaddr =3D vq->vdpa_reconnect_vaddr;
> +       }
> +       /* Check whether the memory for the mmap was allocated by the ker=
nel.
> +        * If not, this device may not have been created/destroyed correc=
tly.
> +        */

Code explains, so let's drop the comment.

> +       if (vaddr =3D=3D 0)
> +               return -EOPNOTSUPP;
> +
> +       /* check if the address is page aligned, if not,
> +        * this address maybe damaged
> +        */

And this comment.

Thanks