On Mon, Oct 20, 2025 at 11:10:14AM +0200, Sabrina Dubroca wrote:
> > +/**
> > + * netdev_compute_master_upper_features - compute feature from lowers
>
> nit: I'm slightly annoyed (that's not quite the right word, sorry)
> that we're adding a new function to "compute features" that doesn't
> touch netdev->features, but I can't come up with a better name
> (the best I got was "compute extra features" and it doesn't help).
Ah, yes, the term "compute features" can be confusing since we don’t actually
update netdev->features. We can rename it if there’s a better alternative.
>
> > + * @dev: the upper device
> > + * @update_header: whether to update upper device's
> > header_len/headroom/tailroom
> > + *
> > + * Recompute the upper device's feature based on all lower devices.
> > + */
> > +void netdev_compute_master_upper_features(struct net_device *dev, bool
> > update_header)
> > +{
> [...]
> > + netif_set_tso_max_segs(dev, tso_max_segs);
> > + netif_set_tso_max_size(dev, tso_max_size);
> > +
> > + netdev_change_features(dev);
>
> Maybe a dumb idea: I'm wondering if we're doing this from the wrong
> side.
>
> Right now we have:
>
> [some device op] -> [this new function] -> netdev_change_features ->
> __netdev_update_features -> ndo_fix_features
>
> Would it make more sense to go instead:
>
> [some device op] -> netdev_change_features -> __netdev_update_features ->
> ndo_fix_features -> [this new function]
Since we actually doesn't touch netdev->feature. I think [this new function]
and netdev_change_features() should be in parallel relationship.
>
> Possible benefit: not forgetting to fix up the "extra" features in
> some cases? (ie calling netdev_change_features when we should have
> called netdev_compute_master_upper_features)
That’s a good reason to call them together. However, ndo_fix_features is used
for computing new features for later use. Since we both compute and set them,
maybe we should put this in ndo_set_features instead?
Thanks
Hangbin
