[PATCH net-next 1/1] ping: fix ipv6 ping socket flow labels

[PATCH net-next 1/1] ping: fix ipv6 ping socket flow labels

[Tony Nguyen]

From: Alan Brady <alan.brady@intel.com>

Ping sockets don't appear to make any attempt to preserve flow labels
created and set by userspace. Instead they are always clobbered by
autolabels (if enabled) or zero.

This grabs the flowlabel out of the msghdr similar to how rawv6_sendmsg
does it and moves the memset up so we don't zero it.

Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Gurucharan <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
---
 net/ipv6/ping.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/net/ipv6/ping.c b/net/ipv6/ping.c
index ecf3a553a0dc..b1179f62bd23 100644
--- a/net/ipv6/ping.c
+++ b/net/ipv6/ping.c
@@ -64,6 +64,8 @@ static int ping_v6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 	if (err)
 		return err;
 
+	memset(&fl6, 0, sizeof(fl6));
+
 	if (msg->msg_name) {
 		DECLARE_SOCKADDR(struct sockaddr_in6 *, u, msg->msg_name);
 		if (msg->msg_namelen < sizeof(*u))
@@ -72,12 +74,15 @@ static int ping_v6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 			return -EAFNOSUPPORT;
 		}
 		daddr = &(u->sin6_addr);
+		if (np->sndflow)
+			fl6.flowlabel = u->sin6_flowinfo & IPV6_FLOWINFO_MASK;
 		if (__ipv6_addr_needs_scope_id(ipv6_addr_type(daddr)))
 			oif = u->sin6_scope_id;
 	} else {
 		if (sk->sk_state != TCP_ESTABLISHED)
 			return -EDESTADDRREQ;
 		daddr = &sk->sk_v6_daddr;
+		fl6.flowlabel = np->flow_label;
 	}
 
 	if (!oif)
@@ -101,7 +106,6 @@ static int ping_v6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 	ipc6.sockc.tsflags = sk->sk_tsflags;
 	ipc6.sockc.mark = sk->sk_mark;
 
-	memset(&fl6, 0, sizeof(fl6));
 	fl6.flowi6_oif = oif;
 
 	if (msg->msg_controllen) {
-- 
2.35.1

Re: [PATCH net-next 1/1] ping: fix ipv6 ping socket flow labels

[Jakub Kicinski]

On Tue, 12 Jul 2022 09:56:08 -0700 Tony Nguyen wrote:
> From: Alan Brady <alan.brady@intel.com>
> 
> Ping sockets don't appear to make any attempt to preserve flow labels
> created and set by userspace. Instead they are always clobbered by
> autolabels (if enabled) or zero.
> 
> This grabs the flowlabel out of the msghdr similar to how rawv6_sendmsg
> does it and moves the memset up so we don't zero it.
> 
> Signed-off-by: Alan Brady <alan.brady@intel.com>
> Tested-by: Gurucharan <gurucharanx.g@intel.com> (A Contingent worker at Intel)
> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

Thanks! Please add a selftest and s/fix/support/ in the subject
otherwise  the stable ML bot will think this is a fix, and its more 
of a missing feature.

Re: [PATCH net-next 1/1] ping: fix ipv6 ping socket flow labels

[Alan Brady]

On 7/13/2022 12:49 PM, Jakub Kicinski wrote:
> On Tue, 12 Jul 2022 09:56:08 -0700 Tony Nguyen wrote:
>> From: Alan Brady <alan.brady@intel.com>
>>
>> Ping sockets don't appear to make any attempt to preserve flow labels
>> created and set by userspace. Instead they are always clobbered by
>> autolabels (if enabled) or zero.
>>
>> This grabs the flowlabel out of the msghdr similar to how rawv6_sendmsg
>> does it and moves the memset up so we don't zero it.
>>
>> Signed-off-by: Alan Brady <alan.brady@intel.com>
>> Tested-by: Gurucharan <gurucharanx.g@intel.com> (A Contingent worker at Intel)
>> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
> 
> Thanks! Please add a selftest and s/fix/support/ in the subject
> otherwise  the stable ML bot will think this is a fix, and its more
> of a missing feature.

Gotcha, will do both, thanks!