Re: [PATCH V2 2/3] vdpa_sim_net: support feature provisioning

From: Jason Wang
Date: Mon Sep 26 2022 - 03:12:06 EST


On Sat, Sep 24, 2022 at 4:01 AM Si-Wei Liu <si-wei.liu@xxxxxxxxxx> wrote:
>
>
>
> On 9/21/2022 7:43 PM, Jason Wang wrote:
> > This patch implements features provisioning for vdpa_sim_net.
> >
> > 1) validating the provisioned features to be a subset of the parent
> > features.
> > 2) clearing the features that is not wanted by the userspace
> >
> > For example:
> >
> > # vdpa mgmtdev show
> > vdpasim_net:
> > supported_classes net
> > max_supported_vqs 3
> > dev_features MTU MAC CTRL_VQ CTRL_MAC_ADDR ANY_LAYOUT VERSION_1 ACCESS_PLATFORM
> Sighs, not to blame any one and it's perhaps too late, but this
> "dev_features" attr in "mgmtdev show" command output should have been
> called "supported_features" in the first place.

Not sure I get this, but I guess this is the negotiated features actually.

I think Ling Shan is working on reporting both negotiated features
with the device features.

>
> >
> > 1) provision vDPA device with all features that are supported by the
> > net simulator
> >
> > # vdpa dev add name dev1 mgmtdev vdpasim_net
> > # vdpa dev config show
> > dev1: mac 00:00:00:00:00:00 link up link_announce false mtu 1500
> > negotiated_features MTU MAC CTRL_VQ CTRL_MAC_ADDR VERSION_1 ACCESS_PLATFORM
> Maybe not in this patch, but for completeness for the whole series,
> could we also add device_features to the output?

Lingshan, could you please share your thoughts or patch on this?

>
> When simply look at the "vdpa dev config show" output, I cannot really
> tell the actual device_features that was used in vdpa creation. For e.g.
> there is a missing feature ANY_LAYOUT from negotiated_features compared
> with supported_features in mgmtdev, but the orchestration software
> couldn't tell if the vdpa device on destination host should be created
> with or without the ANY_LAYOUT feature.

I think VERSION_1 implies ANY_LAYOUT. But it should be sufficient to
use features_src & feature_dst in this case. Actually, it should work
similar as to the cpu flags, the management software should introduce
the concept of cluster which means the maximal set of common features
is calculated and provisioned during device creation to allow
migration among the nodes inside the cluster.

Thanks

>
> Thanks,
> -Siwei
>
>
> >
> > 2) provision vDPA device with a subset of the features
> >
> > # vdpa dev add name dev1 mgmtdev vdpasim_net device_features 0x300020000
> > # vdpa dev config show
> > dev1: mac 00:00:00:00:00:00 link up link_announce false mtu 1500
> > negotiated_features CTRL_VQ VERSION_1 ACCESS_PLATFORM
> >
> > Reviewed-by: Eli Cohen <elic@xxxxxxxxxx>
> > Signed-off-by: Jason Wang <jasowang@xxxxxxxxxx>
> > ---
> > drivers/vdpa/vdpa_sim/vdpa_sim_net.c | 11 ++++++++++-
> > 1 file changed, 10 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/vdpa/vdpa_sim/vdpa_sim_net.c b/drivers/vdpa/vdpa_sim/vdpa_sim_net.c
> > index 886449e88502..a9ba02be378b 100644
> > --- a/drivers/vdpa/vdpa_sim/vdpa_sim_net.c
> > +++ b/drivers/vdpa/vdpa_sim/vdpa_sim_net.c
> > @@ -254,6 +254,14 @@ static int vdpasim_net_dev_add(struct vdpa_mgmt_dev *mdev, const char *name,
> > dev_attr.work_fn = vdpasim_net_work;
> > dev_attr.buffer_size = PAGE_SIZE;
> >
> > + if (config->mask & BIT_ULL(VDPA_ATTR_DEV_FEATURES)) {
> > + if (config->device_features &
> > + ~dev_attr.supported_features)
> > + return -EINVAL;
> > + dev_attr.supported_features &=
> > + config->device_features;
> > + }
> > +
> > simdev = vdpasim_create(&dev_attr);
> > if (IS_ERR(simdev))
> > return PTR_ERR(simdev);
> > @@ -294,7 +302,8 @@ static struct vdpa_mgmt_dev mgmt_dev = {
> > .id_table = id_table,
> > .ops = &vdpasim_net_mgmtdev_ops,
> > .config_attr_mask = (1 << VDPA_ATTR_DEV_NET_CFG_MACADDR |
> > - 1 << VDPA_ATTR_DEV_NET_CFG_MTU),
> > + 1 << VDPA_ATTR_DEV_NET_CFG_MTU |
> > + 1 << VDPA_ATTR_DEV_FEATURES),
> > .max_supported_vqs = VDPASIM_NET_VQ_NUM,
> > .supported_features = VDPASIM_NET_FEATURES,
> > };
>