vdpa/mlx5: Fix mlx5_vdpa_get_config() endianness on big-endian machines

[PATCH v2] vdpa/mlx5: Fix mlx5_vdpa_get_config() endianness on big-endian machines

Posted by Konstantin Shkolnyy 10 months, 1 week ago

mlx5_vdpa_dev_add() doesn’t initialize mvdev->actual_features. It’s
initialized later by mlx5_vdpa_set_driver_features(). However,
mlx5_vdpa_get_config() depends on the VIRTIO_F_VERSION_1 flag in
actual_features, to return data with correct endianness. When it’s called
before mlx5_vdpa_set_driver_features(), the data are incorrectly returned
as big-endian on big-endian machines, while QEMU then interprets them as
little-endian.

The fix is to initialize this VIRTIO_F_VERSION_1 as early as possible,
especially considering that mlx5_vdpa_dev_add() insists on this flag to
always be set anyway.

Signed-off-by: Konstantin Shkolnyy <kshk@linux.ibm.com>
---
Changes in V2: Initialize only VIRTIO_F_VERSION_1 in actual_features,
rather than all flags.

 drivers/vdpa/mlx5/net/mlx5_vnet.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
index 36099047560d..cccc49a08a1a 100644
--- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
+++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
@@ -3884,6 +3884,9 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name,
 	ndev->mvdev.max_vqs = max_vqs;
 	mvdev = &ndev->mvdev;
 	mvdev->mdev = mdev;
+	/* cpu_to_mlx5vdpa16() below depends on this flag */
+	mvdev->actual_features =
+			(device_features & BIT_ULL(VIRTIO_F_VERSION_1));
 
 	ndev->vqs = kcalloc(max_vqs, sizeof(*ndev->vqs), GFP_KERNEL);
 	ndev->event_cbs = kcalloc(max_vqs + 1, sizeof(*ndev->event_cbs), GFP_KERNEL);
-- 
2.34.1

Re: [PATCH v2] vdpa/mlx5: Fix mlx5_vdpa_get_config() endianness on big-endian machines

Posted by Dragos Tatulea 10 months, 1 week ago

On 02/04, Konstantin Shkolnyy wrote:
> mlx5_vdpa_dev_add() doesn’t initialize mvdev->actual_features. It’s
> initialized later by mlx5_vdpa_set_driver_features(). However,
> mlx5_vdpa_get_config() depends on the VIRTIO_F_VERSION_1 flag in
> actual_features, to return data with correct endianness. When it’s called
> before mlx5_vdpa_set_driver_features(), the data are incorrectly returned
> as big-endian on big-endian machines, while QEMU then interprets them as
> little-endian.
> 
> The fix is to initialize this VIRTIO_F_VERSION_1 as early as possible,
> especially considering that mlx5_vdpa_dev_add() insists on this flag to
> always be set anyway.
> 
> Signed-off-by: Konstantin Shkolnyy <kshk@linux.ibm.com>
> ---
> Changes in V2: Initialize only VIRTIO_F_VERSION_1 in actual_features,
> rather than all flags.
> 
>  drivers/vdpa/mlx5/net/mlx5_vnet.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> index 36099047560d..cccc49a08a1a 100644
> --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
> +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> @@ -3884,6 +3884,9 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name,
>  	ndev->mvdev.max_vqs = max_vqs;
>  	mvdev = &ndev->mvdev;
>  	mvdev->mdev = mdev;
> +	/* cpu_to_mlx5vdpa16() below depends on this flag */
> +	mvdev->actual_features =
> +			(device_features & BIT_ULL(VIRTIO_F_VERSION_1));
>  
>  	ndev->vqs = kcalloc(max_vqs, sizeof(*ndev->vqs), GFP_KERNEL);
>  	ndev->event_cbs = kcalloc(max_vqs + 1, sizeof(*ndev->event_cbs), GFP_KERNEL);
> -- 
> 2.34.1
>
The patch looks good now. Thanks!

Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>

Re: [PATCH v2] vdpa/mlx5: Fix mlx5_vdpa_get_config() endianness on big-endian machines

Posted by Jason Wang 10 months, 1 week ago

On Wed, Feb 5, 2025 at 2:16 AM Dragos Tatulea <dtatulea@nvidia.com> wrote:
>
> On 02/04, Konstantin Shkolnyy wrote:
> > mlx5_vdpa_dev_add() doesn’t initialize mvdev->actual_features. It’s
> > initialized later by mlx5_vdpa_set_driver_features(). However,
> > mlx5_vdpa_get_config() depends on the VIRTIO_F_VERSION_1 flag in
> > actual_features, to return data with correct endianness. When it’s called
> > before mlx5_vdpa_set_driver_features(), the data are incorrectly returned
> > as big-endian on big-endian machines, while QEMU then interprets them as
> > little-endian.
> >
> > The fix is to initialize this VIRTIO_F_VERSION_1 as early as possible,
> > especially considering that mlx5_vdpa_dev_add() insists on this flag to
> > always be set anyway.
> >
> > Signed-off-by: Konstantin Shkolnyy <kshk@linux.ibm.com>
> > ---
> > Changes in V2: Initialize only VIRTIO_F_VERSION_1 in actual_features,
> > rather than all flags.
> >
> >  drivers/vdpa/mlx5/net/mlx5_vnet.c | 3 +++
> >  1 file changed, 3 insertions(+)
> >
> > diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > index 36099047560d..cccc49a08a1a 100644
> > --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > @@ -3884,6 +3884,9 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name,
> >       ndev->mvdev.max_vqs = max_vqs;
> >       mvdev = &ndev->mvdev;
> >       mvdev->mdev = mdev;
> > +     /* cpu_to_mlx5vdpa16() below depends on this flag */
> > +     mvdev->actual_features =
> > +                     (device_features & BIT_ULL(VIRTIO_F_VERSION_1));
> >
> >       ndev->vqs = kcalloc(max_vqs, sizeof(*ndev->vqs), GFP_KERNEL);
> >       ndev->event_cbs = kcalloc(max_vqs + 1, sizeof(*ndev->event_cbs), GFP_KERNEL);
> > --
> > 2.34.1
> >
> The patch looks good now. Thanks!
>
> Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
>

Acked-by: Jason Wang <jasowang@redhat.com>

Thanks