Re: [PATCH net-next v3 2/5] net: phy: microchip_ptp : Add ptp library for Microchip phys

[Jakub Kicinski <kuba@kernel.org>]

On Tue, 12 Nov 2024 19:07:21 +0530 Divya Koppera wrote:
> +	/* Iterate over all RX timestamps and match it with the received skbs */
> +	spin_lock_irqsave(&ptp_clock->rx_ts_lock, flags);
> +	list_for_each_entry_safe(rx_ts, tmp, &ptp_clock->rx_ts_list, list) {
> +		/* Check if we found the signature we were looking for. */
> +		if (skb_sig != rx_ts->seq_id)
> +			continue;
> +
> +		match = true;
> +		break;
> +	}
> +	spin_unlock_irqrestore(&ptp_clock->rx_ts_lock, flags);
> +
> +	if (match) {
> +		shhwtstamps = skb_hwtstamps(skb);
> +		shhwtstamps->hwtstamp = ktime_set(rx_ts->seconds, rx_ts->nsec);
> +		netif_rx(skb);
> +
> +		list_del(&rx_ts->list);
> +		kfree(rx_ts);
> +	} else {
> +		skb_queue_tail(&ptp_clock->rx_queue, skb);
> +	}

coccicheck complains that you are using rx_ts after the loop, 
even though it's a loop iterator. Instead of using bool match
make that variable a pointer, set it to NULL and act on it only 
if set. That will make the code easier for static checkers.

Coincidentally, I haven't looked closely, but you seem to have
a spin lock protecting the list, and yet you list_del() without
holding that spin lock? Sus.
-- 
pw-bot: cr