Re: [PATCH v4 net-next 3/8] net: Set skb->mono_delivery_time and clear it after sch_handle_ingress()

[Daniel Borkmann <daniel@xxxxxxxxxxxxx>]

On 2/11/22 8:12 AM, Martin KaFai Lau wrote:
[...]
> +
> +DECLARE_STATIC_KEY_FALSE(netstamp_needed_key);
> +
> +/* It is used in the ingress path to clear the delivery_time.
> + * If needed, set the skb->tstamp to the (rcv) timestamp.
> + */
> +static inline void skb_clear_delivery_time(struct sk_buff *skb)
> +{
> +	if (unlikely(skb->mono_delivery_time)) {
> +		skb->mono_delivery_time = 0;
> +		if (static_branch_unlikely(&netstamp_needed_key))
> +			skb->tstamp = ktime_get_real();
> +		else
> +			skb->tstamp = 0;
> +	}
>   }
>   
>   static inline void skb_clear_tstamp(struct sk_buff *skb)
> @@ -3946,6 +3961,14 @@ static inline void skb_clear_tstamp(struct sk_buff *skb)
>   	skb->tstamp = 0;
>   }
>   
> +static inline ktime_t skb_tstamp(const struct sk_buff *skb)
> +{
> +	if (unlikely(skb->mono_delivery_time))
> +		return 0;
> +
> +	return skb->tstamp;
> +}
> +
>   static inline u8 skb_metadata_len(const struct sk_buff *skb)
>   {

Just small nit, but I don't think here and in other patches as well the conditional
for skb->mono_delivery_time should be marked unlikely(). For container workloads
this is very likely.