[PATCH][net-next] virtio_net: skip RCU read lock by checking xdp_enabled of vi

[PATCH][net-next] virtio_net: skip RCU read lock by checking xdp_enabled of vi

[Li RongQing]

networking benchmark shows that __rcu_read_lock and
__rcu_read_unlock takes some cpu cycles, and we can avoid
calling them partially in virtio rx path by check xdp_enabled
of vi, and xdp is disabled most of time

Signed-off-by: Li RongQing <lirongqing@baidu.com>
---
 drivers/net/virtio_net.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 2ed49884565f..74d2be438180 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -734,6 +734,12 @@ static struct sk_buff *receive_small(struct net_device *dev,
 		dev->stats.rx_length_errors++;
 		goto err_len;
 	}
+
+	if (likely(!vi->xdp_enabled)) {
+		xdp_prog = NULL;
+		goto skip_xdp;
+	}
+
 	rcu_read_lock();
 	xdp_prog = rcu_dereference(rq->xdp_prog);
 	if (xdp_prog) {
@@ -816,6 +822,7 @@ static struct sk_buff *receive_small(struct net_device *dev,
 	}
 	rcu_read_unlock();
 
+skip_xdp:
 	skb = build_skb(buf, buflen);
 	if (!skb) {
 		put_page(page);
@@ -897,6 +904,12 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 		dev->stats.rx_length_errors++;
 		goto err_skb;
 	}
+
+	if (likely(!vi->xdp_enabled)) {
+		xdp_prog = NULL;
+		goto skip_xdp;
+	}
+
 	rcu_read_lock();
 	xdp_prog = rcu_dereference(rq->xdp_prog);
 	if (xdp_prog) {
@@ -1024,6 +1037,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 	}
 	rcu_read_unlock();
 
+skip_xdp:
 	head_skb = page_to_skb(vi, rq, page, offset, len, truesize, !xdp_prog,
 			       metasize, headroom);
 	curr_skb = head_skb;
-- 
2.33.0.69.gc420321.dirty

Re: [PATCH][net-next] virtio_net: skip RCU read lock by checking xdp_enabled of vi

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net-next.git (master)
by David S. Miller <davem@davemloft.net>:

On Sat,  9 Oct 2021 17:32:43 +0800 you wrote:
> networking benchmark shows that __rcu_read_lock and
> __rcu_read_unlock takes some cpu cycles, and we can avoid
> calling them partially in virtio rx path by check xdp_enabled
> of vi, and xdp is disabled most of time
> 
> Signed-off-by: Li RongQing <lirongqing@baidu.com>
> 
> [...]

Here is the summary with links:
  - [net-next] virtio_net: skip RCU read lock by checking xdp_enabled of vi
    https://git.kernel.org/netdev/net-next/c/6213f07cb542

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html