On Mon, Jul 15, 2024 at 06:45:51AM -0700, Jakub Kicinski wrote:
> On Sun, 14 Jul 2024 20:40:20 -0700 Shradha Gupta wrote:
> > + if (ring->rx_jumbo_pending) {
> > + netdev_err(ndev, "%s: rx_jumbo_pending not supported\n",
> > __func__);
> > + return -EINVAL;
> > + }
> > + if (ring->rx_mini_pending) {
> > + netdev_err(ndev, "%s: rx_mini_pending not supported\n",
> > __func__);
> > + return -EINVAL;
> > + }
>
> I think that core already checks this
>
> > + if (!apc)
> > + return -EINVAL;
>
> Provably impossible, apc is netdev + sizeof(netdev) so it'd have to
> wrap a 64b integer to be NULL :|
>
> > + old_tx = apc->tx_queue_size;
> > + old_rx = apc->rx_queue_size;
> > + new_tx = clamp_t(u32, ring->tx_pending, MIN_TX_BUFFERS_PER_QUEUE,
> > MAX_TX_BUFFERS_PER_QUEUE);
> > + new_rx = clamp_t(u32, ring->rx_pending, MIN_RX_BUFFERS_PER_QUEUE,
> > MAX_RX_BUFFERS_PER_QUEUE);
> > +
> > + if (new_tx == old_tx && new_rx == old_rx)
> > + return 0;
>
> Pretty sure core will also not call you if there's no change.
> If it does please update core instead of catching this in the driver.
Thanks for the comments Jakub. I'll verify these and make the relevant changes
in the next version
>
> Please keep in mind that net-next will be closed for the duration
> of the merge window.
Noted, Thanks.
> --
> pw-bot: cr
