[PATCH] netvsc: optimize stats counters performance

Long Li longli at microsoft.com
Fri Aug 2 18:48:47 CEST 2024


> Subject: [PATCH] netvsc: optimize stats counters performance
> 
> Optimized the performance of updating the statistics counters by reducing the
> number of branches.
> 
> Ordered the packet size comparisons according to the probability with typical
> internet traffic mix.
> 
> Signed-off-by: Morten Brørup <mb at smartsharesystems.com>
> ---
>  drivers/net/netvsc/hn_rxtx.c | 32 ++++++++++----------------------
>  1 file changed, 10 insertions(+), 22 deletions(-)
> 
> diff --git a/drivers/net/netvsc/hn_rxtx.c b/drivers/net/netvsc/hn_rxtx.c index
> 9bf1ec5509..b704b2c971 100644
> --- a/drivers/net/netvsc/hn_rxtx.c
> +++ b/drivers/net/netvsc/hn_rxtx.c
> @@ -110,30 +110,18 @@ hn_update_packet_stats(struct hn_stats *stats,
> const struct rte_mbuf *m)
>  	uint32_t s = m->pkt_len;
>  	const struct rte_ether_addr *ea;
> 
> -	if (s == 64) {
> -		stats->size_bins[1]++;
> -	} else if (s > 64 && s < 1024) {
> -		uint32_t bin;
> -
> -		/* count zeros, and offset into correct bin */
> -		bin = (sizeof(s) * 8) - rte_clz32(s) - 5;
> -		stats->size_bins[bin]++;
> -	} else {
> -		if (s < 64)
> -			stats->size_bins[0]++;
> -		else if (s < 1519)
> -			stats->size_bins[6]++;
> -		else
> -			stats->size_bins[7]++;
> -	}
> +	if (s >= 1024)
> +		stats->size_bins[6 + (s > 1518)]++;
> +	else if (s <= 64)
> +		stats->size_bins[s >> 6]++;
> +	else
> +		stats->size_bins[32UL - rte_clz32(s) - 5]++;

This part looks good.

> 
>  	ea = rte_pktmbuf_mtod(m, const struct rte_ether_addr *);
> -	if (rte_is_multicast_ether_addr(ea)) {
> -		if (rte_is_broadcast_ether_addr(ea))
> -			stats->broadcast++;
> -		else
> -			stats->multicast++;
> -	}
> +	RTE_BUILD_BUG_ON(offsetof(struct hn_stats, broadcast) !=
> +			offsetof(struct hn_stats, multicast) + sizeof(uint64_t));
> +	if (unlikely(rte_is_multicast_ether_addr(ea)))
> +		(&stats->multicast)[rte_is_broadcast_ether_addr(ea)]++;
>  }

This makes the code a little harder to read. How about just add "unlikely" to rte_is_multicast_ether_addr(ea) and keep the rest unchanged?

Thank you,

Long



More information about the dev mailing list