Re: [PATCH bpf 6/6] net: enetc: use truesize as XDP RxQ info frag_size

[Larysa Zaremba <larysa.zaremba@intel.com>]

On Fri, Feb 06, 2026 at 06:57:35PM -0800, Jakub Kicinski wrote:
> On Fri, 6 Feb 2026 09:36:21 +0100 Larysa Zaremba wrote:
> > > FWIW my feeling is that instead of nickel and diming leftover space 
> > > in the frags if someone actually cared about growing mbufs we should
> > > have the helper allocate a new page from the PP and append it to the
> > > shinfo. Much simpler, "infinite space", and works regardless of the
> > > driver. I don't mean that to suggest you implement it, purely to point
> > > out that I think nobody really uses positive offsets.. So we can as
> > > well switch more complicated drivers back to xdp_rxq_info_reg().
> > 
> > As Vladimir has mentioned, if the driver does not use header split, frags will 
> > have a tailroom of a size of skb_shared_info, so tail growing does work in 
> > practice.
> > 
> > Allocating a page_pool buffer (given XDP queue has one attached) is certainly an 
> > option, although I am not sure if anyone needs it. Furthermore, growing tail 
> > would still fail for a single-buf case.
> 
> sbuf is a different code path, sbuf has precise frame_sz per frame,
> not a single value in rxq, no? Don't mean to argue, just making sure
> my mental model is correct ;)
> 

You are right, xdp_buff in sbuf case is self-sufficient inside of bpf ops. sbuf 
mostly requires memory info from XDP RxQ for proper xdp_frame handling.

What I meant is that intended conditions of failure to grow tail are the same 
for sbuf and mbuf case, the last (which can be the same as the first) buffer 
needs to have enough leftover space.