From mboxrd@z Thu Jan 1 00:00:00 1970 From: Simon Horman Subject: Re: [PATCH net 4/4] ipvs: stats should not depend on CPU 0 Date: Fri, 30 Aug 2013 17:53:32 +0900 Message-ID: <20130830085332.GA10484@verge.net.au> References: <1377848768.12846.122.camel@hawk.mlab.se> Mime-Version: 1.0 Return-path: Content-Disposition: inline In-Reply-To: <1377848768.12846.122.camel@hawk.mlab.se> Sender: lvs-devel-owner@vger.kernel.org List-ID: Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Julian Anastasov , Hans Schillstrom Cc: lvs-devel@vger.kernel.org On Fri, Aug 30, 2013 at 10:31:57AM +0300, Julian Anastasov wrote: > > Hello, > > On Fri, 30 Aug 2013, Simon Horman wrote: > > > [ Cc: Hans Schillstrom ] > > > > On Wed, Aug 28, 2013 at 07:29:36PM +0300, Julian Anastasov wrote: > > > When reading percpu stats we need to properly reset > > > the sum when CPU 0 is not present in the possible mask. > > > > It seems reasonable to me that this condition could occur, > > though I am not aware of a specific case where that is so. > > > > Can I confirm that this problem was introduced by > > b17fc9963f837ef1 ("IPVS: netns, ip_vs_stats and its procfs")? > > That is correct, I forgot to track the initial > commit, you can freely add such information before > applying. > > Regards > > -- > Julian Anastasov > On Fri, Aug 30, 2013 at 09:46:08AM +0200, Hans Schillstrom wrote: > > On Wed, 2013-08-28 at 19:29 +0300, Julian Anastasov wrote: > > When reading percpu stats we need to properly reset > > the sum when CPU 0 is not present in the possible mask. > > Not a very common case, but still possible. > > > Signed-off-by: Julian Anastasov > > Acked-by: Hans Schillstrom Thanks guys, I have queued up the following in the ipvs tree along with the rest of this series. >From 056a057b581b7a98669e65745c9866d233f82a18 Mon Sep 17 00:00:00 2001 From: Julian Anastasov Date: Wed, 28 Aug 2013 19:29:36 +0300 Subject: [PATCH] ipvs: stats should not depend on CPU 0 When reading percpu stats we need to properly reset the sum when CPU 0 is not present in the possible mask. This problem was introduced by 17fc9963f837ef1 ("IPVS: netns, ip_vs_stats and its procfs"). Signed-off-by: Julian Anastasov Acked-by: Hans Schillstrom Signed-off-by: Simon Horman --- net/netfilter/ipvs/ip_vs_est.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/net/netfilter/ipvs/ip_vs_est.c b/net/netfilter/ipvs/ip_vs_est.c index 6bee6d0..1425e9a 100644 --- a/net/netfilter/ipvs/ip_vs_est.c +++ b/net/netfilter/ipvs/ip_vs_est.c @@ -59,12 +59,13 @@ static void ip_vs_read_cpu_stats(struct ip_vs_stats_user *sum, struct ip_vs_cpu_stats __percpu *stats) { int i; + bool add = false; for_each_possible_cpu(i) { struct ip_vs_cpu_stats *s = per_cpu_ptr(stats, i); unsigned int start; __u64 inbytes, outbytes; - if (i) { + if (add) { sum->conns += s->ustats.conns; sum->inpkts += s->ustats.inpkts; sum->outpkts += s->ustats.outpkts; @@ -76,6 +77,7 @@ static void ip_vs_read_cpu_stats(struct ip_vs_stats_user *sum, sum->inbytes += inbytes; sum->outbytes += outbytes; } else { + add = true; sum->conns = s->ustats.conns; sum->inpkts = s->ustats.inpkts; sum->outpkts = s->ustats.outpkts; -- 1.8.3.2