[PATCH] inet: ping: Fix icmp out counting

[PATCH] inet: ping: Fix icmp out counting

[yuan.gao]

From: "yuan.gao" <yuan.gao@ucloud.cn>

When the ping program uses an IPPROTO_ICMP socket to send ICMP_ECHO
messages, ICMP_MIB_OUTMSGS is counted twice.

    ping_v4_sendmsg
      ping_v4_push_pending_frames
        ip_push_pending_frames
          ip_finish_skb
            __ip_make_skb
              icmp_out_count(net, icmp_type); // first count
      icmp_out_count(sock_net(sk), user_icmph.type); // second count

However, when the ping program uses an IPPROTO_RAW socket,
ICMP_MIB_OUTMSGS is counted correctly only once.

Therefore, the first count should be removed.

Fixes: c319b4d76b9e ("net: ipv4: add IPPROTO_ICMP socket kind")
Signed-off-by: yuan.gao <yuan.gao@ucloud.cn>
---
 net/ipv4/ping.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/net/ipv4/ping.c b/net/ipv4/ping.c
index 4cb0c896c..c662d6821 100644
--- a/net/ipv4/ping.c
+++ b/net/ipv4/ping.c
@@ -833,10 +833,8 @@ static int ping_v4_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 out_free:
 	if (free)
 		kfree(ipc.opt);
-	if (!err) {
-		icmp_out_count(sock_net(sk), user_icmph.type);
+	if (!err)
 		return len;
-	}
 	return err;
 
 do_confirm:
-- 
2.32.0

Re: [PATCH] inet: ping: Fix icmp out counting

[Ido Schimmel]

Next time, please tag the patch as [PATCH net]. See:

https://docs.kernel.org/process/maintainer-netdev.html

On Wed, Dec 24, 2025 at 02:31:45PM +0800, yuan.gao@ucloud.cn wrote:
> From: "yuan.gao" <yuan.gao@ucloud.cn>
> 
> When the ping program uses an IPPROTO_ICMP socket to send ICMP_ECHO
> messages, ICMP_MIB_OUTMSGS is counted twice.
> 
>     ping_v4_sendmsg
>       ping_v4_push_pending_frames
>         ip_push_pending_frames
>           ip_finish_skb
>             __ip_make_skb
>               icmp_out_count(net, icmp_type); // first count
>       icmp_out_count(sock_net(sk), user_icmph.type); // second count
> 
> However, when the ping program uses an IPPROTO_RAW socket,
> ICMP_MIB_OUTMSGS is counted correctly only once.
> 
> Therefore, the first count should be removed.

Looks correct.

Before:

# sysctl -wq net.ipv4.ping_group_range="0 4294967294"
# nstat -z -j | jq '.[]["IcmpOutEchos"]'
0
# ping -c1 127.0.0.1 &> /dev/null
# nstat -z -j | jq '.[]["IcmpOutEchos"]'
2

After:

# sysctl -wq net.ipv4.ping_group_range="0 4294967294"
# nstat -z -j | jq '.[]["IcmpOutEchos"]'
0
# ping -c1 127.0.0.1 &> /dev/null
# nstat -z -j | jq '.[]["IcmpOutEchos"]'
1

And it's consistent with IPv6:

# sysctl -wq net.ipv4.ping_group_range="0 4294967294"
# nstat -z -j | jq '.[]["Icmp6OutEchos"]'
0
# ping -c1 ::1 &> /dev/null
# nstat -z -j | jq '.[]["Icmp6OutEchos"]'
1

> 
> Fixes: c319b4d76b9e ("net: ipv4: add IPPROTO_ICMP socket kind")
> Signed-off-by: yuan.gao <yuan.gao@ucloud.cn>

Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Tested-by: Ido Schimmel <idosch@nvidia.com>

Re: [PATCH] inet: ping: Fix icmp out counting

[yuan.gao]

On Wed, Dec 24, 2025 at 06:08:47PM +0200, Ido Schimmel wrote:
> Next time, please tag the patch as [PATCH net]. See:
> 
> https://docs.kernel.org/process/maintainer-netdev.html
> 
> On Wed, Dec 24, 2025 at 02:31:45PM +0800, yuan.gao@ucloud.cn wrote:
> > From: "yuan.gao" <yuan.gao@ucloud.cn>
> > 
> > When the ping program uses an IPPROTO_ICMP socket to send ICMP_ECHO
> > messages, ICMP_MIB_OUTMSGS is counted twice.
> > 
> >     ping_v4_sendmsg
> >       ping_v4_push_pending_frames
> >         ip_push_pending_frames
> >           ip_finish_skb
> >             __ip_make_skb
> >               icmp_out_count(net, icmp_type); // first count
> >       icmp_out_count(sock_net(sk), user_icmph.type); // second count
> > 
> > However, when the ping program uses an IPPROTO_RAW socket,
> > ICMP_MIB_OUTMSGS is counted correctly only once.
> > 
> > Therefore, the first count should be removed.
> 
> Looks correct.
> 
> Before:
> 
> # sysctl -wq net.ipv4.ping_group_range="0 4294967294"
> # nstat -z -j | jq '.[]["IcmpOutEchos"]'
> 0
> # ping -c1 127.0.0.1 &> /dev/null
> # nstat -z -j | jq '.[]["IcmpOutEchos"]'
> 2
> 
> After:
> 
> # sysctl -wq net.ipv4.ping_group_range="0 4294967294"
> # nstat -z -j | jq '.[]["IcmpOutEchos"]'
> 0
> # ping -c1 127.0.0.1 &> /dev/null
> # nstat -z -j | jq '.[]["IcmpOutEchos"]'
> 1
> 
> And it's consistent with IPv6:
> 
> # sysctl -wq net.ipv4.ping_group_range="0 4294967294"
> # nstat -z -j | jq '.[]["Icmp6OutEchos"]'
> 0
> # ping -c1 ::1 &> /dev/null
> # nstat -z -j | jq '.[]["Icmp6OutEchos"]'
> 1
> 
> > 
> > Fixes: c319b4d76b9e ("net: ipv4: add IPPROTO_ICMP socket kind")
> > Signed-off-by: yuan.gao <yuan.gao@ucloud.cn>
> 
> Reviewed-by: Ido Schimmel <idosch@nvidia.com>
> Tested-by: Ido Schimmel <idosch@nvidia.com>

Thanks for the reminder!

Cheers,
Yuan Gao