Re: [PATCH net-next v1 00/11] xsk: Zero copy through page-pool memory providers
From: Björn Töpel
Date: Sun Nov 22 2026 - 13:13:22 EST
Björn Töpel <bjorn@xxxxxxxxxx> writes:
> Hi!
>
> AF_XDP zero copy needs a second RX allocator in every driver. This
> series makes UMEM a page-pool memory provider instead: a page_pool
> driver opts in, and keeps its refill, recycling and DMA sync code.
> fbnic is the first user. Patches 1-7 are core, 8-11 fbnic.
>
> Since the RFC [RFC], net_iov is a first class netmem, the core owns
> AF_XDP setup, and a refill contract replaces the NAPI kicks. The fixes
> are out.
Hey,
Thanks for all the input!
TL;DR for new readers: Too much churn, especially on the driver side.
net_iov was not intended to be readable from the core side of things.
The net_iov part needs some deeper thinking. UMEM memory is readable
(the kernel can read the memory), but needs a non-page refcounting, and
benefits from the rx_buf_len (larger chunks). iouring TCP ZC frags are
also readable, and the kernel does read from it from time to time, but
it still unreadable. I have some ideas how to fit UMEM to the puzzle.
:-)
I'm working on a v2, but it'll have a different take than this version,
to try to reduce the amount of driver churn. It'll probably be an RFC or
a brand new v1. I'll try to address all input from this design.
Cheers, and thanks all!
Björn