Re: [PATCH net-next v7 1/5] net: move ARFS rmap management to core

[Jakub Kicinski <kuba@kernel.org>]

On Tue,  4 Feb 2025 15:06:18 -0700 Ahmed Zaki wrote:
> +void netif_napi_set_irq_locked(struct napi_struct *napi, int irq)
> +{
> +	int rc;
> +
> +	/* Remove existing rmap entries */
> +	if (napi->dev->rx_cpu_rmap_auto &&
> +	    napi->irq != irq && napi->irq > 0)

this condition gets a bit hairy by the end of the series.
could you add a napi state bit that indicates that a notifier is
installed? Then here:

	if (napi->irq == irq)
		return;

	if (test_and_clear_bit(NAPI_STATE_HAS_NOTIFIER, &napi->state))
		irq_set_affinity_notifier(napi->irq, NULL);
	if (irq < 0)
		return;

And you can similarly simplify napi_disable_locked().

Speaking of which, why do the auto-removal in napi_disable()
rather than netif_napi_del() ? We don't reinstall on napi_enable()
and doing a disable() + enable() is fairly common during driver
reconfig.

> +		irq_set_affinity_notifier(napi->irq, NULL);
> +
> +	napi->irq = irq;
> +	if (irq > 0) {
> +		rc = napi_irq_cpu_rmap_add(napi, irq);
> +		if (rc)
> +			netdev_warn(napi->dev, "Unable to update ARFS map (%d)\n",

nit: not sure I'd grasp this message as a user, maybe:

	"Unable to install aRFS CPU to Rx queue mapping"

? Not great either, I guess.