Re: [RFC bpf-next 3/7] bpf: implement devtx hook points

[Simon Horman <simon.horman@corigine.com>]

On Tue, Jun 13, 2023 at 12:00:25PM -0700, Stanislav Fomichev wrote:
> On Tue, Jun 13, 2023 at 8:08 AM Simon Horman <simon.horman@corigine.com> wrote:
> > On Mon, Jun 12, 2023 at 10:23:03AM -0700, Stanislav Fomichev wrote:

...

> > > +void devtx_complete(struct net_device *netdev, struct devtx_frame *ctx)
> > > +{
> > > +     rcu_read_lock();
> > > +     devtx_run(netdev, ctx, &netdev->devtx_cp);
> > > +     rcu_read_unlock();
> > > +}
> > > +EXPORT_SYMBOL_GPL(devtx_complete);
> > > +
> > > +/**
> > > + * devtx_sb - Called for every egress netdev packet
> >
> > As this is a kernel doc, it would be good to document the ctx parameter here.
> 
> I didn't really find a convincing way to add a comment, I've had the
> following which I've removed prio to submission:
> @ctx devtx_frame context
> 
> But it doesn't seem like it brings anything useful? Or ok to keep it that way?

Thanks Stan,

I see what you are saying wrt it not bringing much value.
But I'm more thinking that something is better than nothing.
Anyway, I'll drop this topic if you prefer.