Re: [PATCH] vdpa/mlx5: Enlarge queue size to 512 entries

Re: [PATCH] vdpa/mlx5: Enlarge queue size to 512 entries

[Jason Wang]

On Tue, Feb 22, 2022 at 3:00 AM Eugenio Perez Martin
<eperezma@redhat.com> wrote:
>
> On Mon, Feb 21, 2022 at 2:20 PM Eli Cohen <elic@nvidia.com> wrote:
> >
> > Currently we can work with 512 entries with any MTU size. Change
> > MLX5_VDPA_MAX_VQ_ENTRIES accordingly.
> >
> > Signed-off-by: Eli Cohen <elic@nvidia.com>
> > ---
> >  drivers/vdpa/mlx5/net/mlx5_vnet.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > index be095dc1134e..de115e3fe761 100644
> > --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > @@ -2062,7 +2062,7 @@ static void mlx5_vdpa_set_config_cb(struct vdpa_device *vdev, struct vdpa_callba
> >         ndev->config_cb = *cb;
> >  }
> >
> > -#define MLX5_VDPA_MAX_VQ_ENTRIES 256
> > +#define MLX5_VDPA_MAX_VQ_ENTRIES 512

I'd expect this should be something that can be read from the fw or
hw? Otherwise we can't increase it in the future.

Thanks

> >  static u16 mlx5_vdpa_get_vq_num_max(struct vdpa_device *vdev)
> >  {
> >         return MLX5_VDPA_MAX_VQ_ENTRIES;
> > --
> > 2.35.1
> >
>
> Acked-by: Eugenio Pérez <eperezma@redhat.com>
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH] vdpa/mlx5: Enlarge queue size to 512 entries

[Jason Wang]

On Tue, Feb 22, 2022 at 2:33 PM Eli Cohen <elic@nvidia.com> wrote:
>
> On Tue, Feb 22, 2022 at 11:18:14AM +0800, Jason Wang wrote:
> > On Tue, Feb 22, 2022 at 3:00 AM Eugenio Perez Martin
> > <eperezma@redhat.com> wrote:
> > >
> > > On Mon, Feb 21, 2022 at 2:20 PM Eli Cohen <elic@nvidia.com> wrote:
> > > >
> > > > Currently we can work with 512 entries with any MTU size. Change
> > > > MLX5_VDPA_MAX_VQ_ENTRIES accordingly.
> > > >
> > > > Signed-off-by: Eli Cohen <elic@nvidia.com>
> > > > ---
> > > >  drivers/vdpa/mlx5/net/mlx5_vnet.c | 2 +-
> > > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > >
> > > > diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > > > index be095dc1134e..de115e3fe761 100644
> > > > --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > > > +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > > > @@ -2062,7 +2062,7 @@ static void mlx5_vdpa_set_config_cb(struct vdpa_device *vdev, struct vdpa_callba
> > > >         ndev->config_cb = *cb;
> > > >  }
> > > >
> > > > -#define MLX5_VDPA_MAX_VQ_ENTRIES 256
> > > > +#define MLX5_VDPA_MAX_VQ_ENTRIES 512
> >
> > I'd expect this should be something that can be read from the fw or
> > hw? Otherwise we can't increase it in the future.
> >
>
> We do read all the required parameters from firmware in order to fulfill
> any valid value up to the spec defined limit of 32K entries. But, there
> are FW related issues that with 9K MTU we will start seeing firmware
> errors if we go beyond 512. So I thought it would be better to increase
> number of entries to a safe value till we have other issues resolved.

I see.

So

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

>
> > Thanks
> >
> > > >  static u16 mlx5_vdpa_get_vq_num_max(struct vdpa_device *vdev)
> > > >  {
> > > >         return MLX5_VDPA_MAX_VQ_ENTRIES;
> > > > --
> > > > 2.35.1
> > > >
> > >
> > > Acked-by: Eugenio Pérez <eperezma@redhat.com>
> > >
> >
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization