Re: [PATCH net-next v1 06/11] page_pool: Coordinate refill with memory providers
From: Björn Töpel
Date: Sun Nov 15 2026 - 14:17:31 EST
netdev-bot+sashiko@xxxxxxxxxx writes:
[...]
> Should netif_mp_napi_del() also clear rxq->napi when it matches, or
> should netif_mp_update_napi() only publish a NAPI that is still
> listed? If drivers are expected to clear the association themselves,
> could netmem.rst item 8 say so?
The first. v2 drops mp_napi, so wakeups use rxq->napi itself, under RCU,
and the core clears it when the NAPI is deleted.
[...]
> Should the documented contract require that, or should the core handle
> the disabled-NAPI case itself?
The contract. fbnic polls the vector right after napi_enable()
(FBNIC_INTR_SET), so the queue refills once, from the reuse stack or
FILL, and refill_done marks it starved again if both are empty. A wakeup
lost on a stopped queue doesn't matter then. v2 documents that a driver
polls the queue once after enabling its NAPI.
> Could uses_need_wakeup be cached in struct xsk_pp when it is created, or
> read only after pp->fq is confirmed under pp->lock?
Same root cause as the failed removal in 05, and the same fix. The pool
then stays until the queue DMA has stopped, and no refill runs after
that.
> Should the error path arm from the configuration that is running
> again, or should __netif_mp_open_rxq() and __netif_mp_close_rxq() call
> netif_mp_update_napi() once they have settled rxq->mp_params?
Neither is needed -- with mp_napi and mp_armed gone in v2, there is no
wakeup target to arm from the wrong configuration.
Björn