[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20200802003818-mutt-send-email-mst@kernel.org>
Date: Sun, 2 Aug 2020 00:40:25 -0400
From: "Michael S. Tsirkin" <mst@...hat.com>
To: Mao Wenan <wenan.mao@...ux.alibaba.com>
Cc: jasowang@...hat.com, davem@...emloft.net, kuba@...nel.org,
netdev@...r.kernel.org
Subject: Re: [PATCH -next] virtio_net: Avoid loop in virtnet_poll
On Sun, Aug 02, 2020 at 11:41:23AM +0800, Mao Wenan wrote:
> The loop may exist if vq->broken is true,
> virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split
> will return NULL, so virtnet_poll will reschedule napi to
> receive packet, it will lead cpu usage(si) to 100%.
>
> call trace as below:
> virtnet_poll
> virtnet_receive
> virtqueue_get_buf_ctx
> virtqueue_get_buf_ctx_packed
> virtqueue_get_buf_ctx_split
> virtqueue_napi_complete
> virtqueue_napi_schedule //it will reschedule napi
>
> Signed-off-by: Mao Wenan <wenan.mao@...ux.alibaba.com>
I think it's more a bug in virtqueue_poll : virtqueue_get_buf reports
NULL on broken, so virtqueue_poll should report false for consistency.
> ---
> drivers/net/virtio_net.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index ba38765..a058da1 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -327,7 +327,8 @@ static void virtqueue_napi_complete(struct napi_struct *napi,
>
> opaque = virtqueue_enable_cb_prepare(vq);
> if (napi_complete_done(napi, processed)) {
> - if (unlikely(virtqueue_poll(vq, opaque)))
> + if (unlikely(virtqueue_poll(vq, opaque)) &&
> + unlikely(!virtqueue_is_broken(vq)))
> virtqueue_napi_schedule(napi, vq);
> } else {
> virtqueue_disable_cb(vq);
> --
> 1.8.3.1
Powered by blists - more mailing lists