On Tue, 07 Apr 2026 20:03:44 +0200,
Janne Johansson <icepic.dz@gmail.com> wrote:
> 
> > >       ifp->if_xflags = IFXF_MPSAFE;
> > > +#ifndef SMALL_KERNEL
> > > +     ifp->if_xflags |= IFXF_LRO;
> > > +#endif
> > >       ifp->if_capabilities = IFCAP_VLAN_MTU | IFCAP_CSUM_TCPv4 |
> > >           IFCAP_CSUM_UDPv4 | IFCAP_CSUM_TCPv6 | IFCAP_CSUM_UDPv6;
> > > +#ifndef SMALL_KERNEL
> > > +     ifp->if_capabilities |= IFCAP_LRO;
> > > +#endif
> > >
> > >
> > > -     ml_enqueue(ml, m);
> > > +#ifndef SMALL_KERNEL
> > > +     if (__predict_true(ISSET(ifp->if_xflags, IFXF_LRO)) &&
> > > +         __predict_true(!ISSET(word2, PIP_WQE_WORD2_IP_NI)) &&
> >
> > I would leave these branch hints out as they mainly add visual noise
> > here. Also, the predict true hint does not seem correct with IFXF_LRO
> > because it is a valid admin choice to not use LRO.
> 
> It is a valid choice to turn it off, but it defaults to on for cnmac(4) devices.
> Then if compiler prediction hints have any measurable effect is
> another matter, but it would be the "correct" hint for a lot of
> installations if this diff goes in.
> 

I just mimic that condition:

	if (__predict_true(!ISSET(word2, PIP_WQE_WORD2_IP_NI))) {

but, frankly, I almost sure that you can't measure impact of any
__predict_true, it is noise and I'd like follow Visa suggestion.

-- 
wbr, Kirill