On 1/14/26 03:36, Jakub Kicinski wrote:
On Fri, 9 Jan 2026 11:28:46 +0000 Pavel Begunkov wrote:
@@ -4342,7 +4343,8 @@ static void bnxt_init_ring_struct(struct bnxt *bp)
if (!rxr)
goto skip_rx;
- rxr->rx_page_size = BNXT_RX_PAGE_SIZE;
+ rxq = __netif_get_rx_queue(bp->dev, i);
+ rxr->rx_page_size = rxq->qcfg.rx_page_size;
Pretty sure I asked for the netdev_queue_config() helper to make
a return, instead of drivers poking directly into core state.
Having the config live in rxq directly is also ugh.
Having a helper would be a good idea, but I went for stashing
configs in the queue as it's simpler, while dynamic allocations
were of no benefit for this series. Maybe there are some further
plans for it, but as you mentioned, it'd be better to do on top.
But at this stage we're probably better off if you just respin
to fix the nits from Paolo and I try to de-lobotimize the driver
facing API. This is close enough.
Ok
--
Pavel Begunkov
