netdev - Re: [PATCH net-next v3 6/8] virtio-net: xsk zero copy xmit kick by threshold

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <b04d7110-95ec-0aec-a802-631baf260a62@redhat.com>
Date:   Tue, 6 Apr 2021 14:59:00 +0800
From:   Jason Wang <jasowang@...hat.com>
To:     Xuan Zhuo <xuanzhuo@...ux.alibaba.com>, netdev@...r.kernel.org
Cc:     "Michael S. Tsirkin" <mst@...hat.com>,
        "David S. Miller" <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>,
        Björn Töpel <bjorn@...nel.org>,
        Magnus Karlsson <magnus.karlsson@...el.com>,
        Jonathan Lemon <jonathan.lemon@...il.com>,
        Alexei Starovoitov <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        Jesper Dangaard Brouer <hawk@...nel.org>,
        John Fastabend <john.fastabend@...il.com>,
        Andrii Nakryiko <andrii@...nel.org>,
        Martin KaFai Lau <kafai@...com>,
        Song Liu <songliubraving@...com>, Yonghong Song <yhs@...com>,
        KP Singh <kpsingh@...nel.org>,
        virtualization@...ts.linux-foundation.org, bpf@...r.kernel.org,
        Dust Li <dust.li@...ux.alibaba.com>
Subject: Re: [PATCH net-next v3 6/8] virtio-net: xsk zero copy xmit kick by
 threshold


在 2021/3/31 下午3:11, Xuan Zhuo 写道:
> After testing, the performance of calling kick every time is not stable.
> And if all the packets are sent and kicked again, the performance is not
> good. So add a module parameter to specify how many packets are sent to
> call a kick.
>
> 8 is a relatively stable value with the best performance.


Please add some perf numbers here.

Thanks


>
> Signed-off-by: Xuan Zhuo <xuanzhuo@...ux.alibaba.com>
> Reviewed-by: Dust Li <dust.li@...ux.alibaba.com>
> ---
>   drivers/net/virtio_net.c | 23 ++++++++++++++++++++---
>   1 file changed, 20 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 259fafcf6028..d7e95f55478d 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -29,10 +29,12 @@ module_param(napi_weight, int, 0444);
>   
>   static bool csum = true, gso = true, napi_tx = true;
>   static int xsk_budget = 32;
> +static int xsk_kick_thr = 8;
>   module_param(csum, bool, 0444);
>   module_param(gso, bool, 0444);
>   module_param(napi_tx, bool, 0644);
>   module_param(xsk_budget, int, 0644);
> +module_param(xsk_kick_thr, int, 0644);
>   
>   /* FIXME: MTU in config. */
>   #define GOOD_PACKET_LEN (ETH_HLEN + VLAN_HLEN + ETH_DATA_LEN)
> @@ -2612,6 +2614,8 @@ static int virtnet_xsk_xmit_batch(struct send_queue *sq,
>   	struct xdp_desc desc;
>   	int err, packet = 0;
>   	int ret = -EAGAIN;
> +	int need_kick = 0;
> +	int kicks = 0;
>   
>   	if (sq->xsk.last_desc.addr) {
>   		err = virtnet_xsk_xmit(sq, pool, &sq->xsk.last_desc);
> @@ -2619,6 +2623,7 @@ static int virtnet_xsk_xmit_batch(struct send_queue *sq,
>   			return -EBUSY;
>   
>   		++packet;
> +		++need_kick;
>   		--budget;
>   		sq->xsk.last_desc.addr = 0;
>   	}
> @@ -2642,13 +2647,25 @@ static int virtnet_xsk_xmit_batch(struct send_queue *sq,
>   		}
>   
>   		++packet;
> +		++need_kick;
> +		if (need_kick > xsk_kick_thr) {
> +			if (virtqueue_kick_prepare(sq->vq) &&
> +			    virtqueue_notify(sq->vq))
> +				++kicks;
> +
> +			need_kick = 0;
> +		}
>   	}
>   
>   	if (packet) {
> -		if (virtqueue_kick_prepare(sq->vq) &&
> -		    virtqueue_notify(sq->vq)) {
> +		if (need_kick) {
> +			if (virtqueue_kick_prepare(sq->vq) &&
> +			    virtqueue_notify(sq->vq))
> +				++kicks;
> +		}
> +		if (kicks) {
>   			u64_stats_update_begin(&sq->stats.syncp);
> -			sq->stats.kicks += 1;
> +			sq->stats.kicks += kicks;
>   			u64_stats_update_end(&sq->stats.syncp);
>   		}
>