Michael S. Tsirkin
2023-Mar-20 20:02 UTC
[PATCH v3 2/2] vdpa/mlx5: Make VIRTIO_NET_F_MRG_RXBUF off by default
On Mon, Mar 20, 2023 at 01:49:30PM +0200, Eli Cohen wrote:> One can still enable it when creating the vdpa device using vdpa tool by > providing features that include it. > > For example: > $ vdpa dev add name vdpa0 mgmtdev pci/0000:86:00.2 device_features 0x300cb982b > > Please be aware that this feature was not supported before the previous > patch in this list was introduced so we don't change user experience.so I would say the patches have to be reordered to avoid a regression?> Current firmware versions show degradation in packet rate when using > MRG_RXBUF. Users who favor memory saving over packet rate could enable > this feature but we want to keep it off by default. > > Signed-off-by: Eli Cohen <elic at nvidia.com>OK and when future firmware will (maybe) fix this up how will you know it's ok to enable by default? Some version check I guess? It would be better if firmware specified flags to enable by default ...> --- > drivers/vdpa/mlx5/net/mlx5_vnet.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c > index 5285dd76c793..24397a71d6f3 100644 > --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c > +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c > @@ -3146,6 +3146,8 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name, > return -EINVAL; > } > device_features &= add_config->device_features; > + } else { > + device_features &= ~BIT_ULL(VIRTIO_NET_F_MRG_RXBUF); > } > if (!(device_features & BIT_ULL(VIRTIO_F_VERSION_1) && > device_features & BIT_ULL(VIRTIO_F_ACCESS_PLATFORM))) { > -- > 2.38.1