I do a ipvs connection sync test in a 3.10 version kernel which has the 7c13f97ffde6 commit which succeed. I will modify the fixes information of the patch and replace the skb_queue_empty with skb_queue_empty_lockless. Thanks.
-----邮件原件----- 发件人: Julian Anastasov [mailto:j...@ssi.bg] 发送时间: Thursday, July 16, 2020 1:36 收件人: Guodeqing (A) <geffrey....@huawei.com> 抄送: wens...@linux-vs.org; ho...@verge.net.au; pa...@netfilter.org; kad...@netfilter.org; f...@strlen.de; da...@davemloft.net; k...@kernel.org; netdev@vger.kernel.org; lvs-de...@vger.kernel.org; netfilter-de...@vger.kernel.org 主题: Re: [PATCH] ipvs: fix the connection sync failed in some cases Hello, On Wed, 15 Jul 2020, guodeqing wrote: > The sync_thread_backup only checks sk_receive_queue is empty or not, > there is a situation which cannot sync the connection entries when > sk_receive_queue is empty and sk_rmem_alloc is larger than sk_rcvbuf, > the sync packets are dropped in __udp_enqueue_schedule_skb, this is > because the packets in reader_queue is not read, so the rmem is not > reclaimed. Good catch. We missed this change in UDP... > Here I add the check of whether the reader_queue of the udp sock is > empty or not to solve this problem. > > Fixes: 7c13f97ffde6 ("udp: do fwd memory scheduling on dequeue") Why this commit and not 2276f58ac589 which adds reader_queue to udp_poll() ? May be both? > Reported-by: zhouxudong <zhouxudo...@huawei.com> > Signed-off-by: guodeqing <geffrey....@huawei.com> > --- > net/netfilter/ipvs/ip_vs_sync.c | 12 ++++++++---- > 1 file changed, 8 insertions(+), 4 deletions(-) > > diff --git a/net/netfilter/ipvs/ip_vs_sync.c > b/net/netfilter/ipvs/ip_vs_sync.c index 605e0f6..abe8d63 100644 > --- a/net/netfilter/ipvs/ip_vs_sync.c > +++ b/net/netfilter/ipvs/ip_vs_sync.c > @@ -1717,6 +1717,8 @@ static int sync_thread_backup(void *data) { > struct ip_vs_sync_thread_data *tinfo = data; > struct netns_ipvs *ipvs = tinfo->ipvs; > + struct sock *sk = tinfo->sock->sk; > + struct udp_sock *up = udp_sk(sk); > int len; > > pr_info("sync thread started: state = BACKUP, mcast_ifn = %s, " > @@ -1724,12 +1726,14 @@ static int sync_thread_backup(void *data) > ipvs->bcfg.mcast_ifn, ipvs->bcfg.syncid, tinfo->id); > > while (!kthread_should_stop()) { > - wait_event_interruptible(*sk_sleep(tinfo->sock->sk), > - !skb_queue_empty(&tinfo->sock->sk->sk_receive_queue) > - || kthread_should_stop()); > + wait_event_interruptible(*sk_sleep(sk), > + > !skb_queue_empty(&sk->sk_receive_queue) || > + !skb_queue_empty(&up->reader_queue) || May be we should use skb_queue_empty_lockless for 5.4+ and skb_queue_empty() for backports to 4.14 and 4.19... > + kthread_should_stop()); > > /* do we have data now? */ > - while (!skb_queue_empty(&(tinfo->sock->sk->sk_receive_queue))) { > + while (!skb_queue_empty(&sk->sk_receive_queue) || > + !skb_queue_empty(&up->reader_queue)) { Here too > len = ip_vs_receive(tinfo->sock, tinfo->buf, > ipvs->bcfg.sync_maxlen); > if (len <= 0) { > -- > 2.7.4 Regards -- Julian Anastasov <j...@ssi.bg>
