get receiving interface index while no message is received

get receiving interface index while no message is received

[Yang Hongyang]

when I use getsockopt(sk, IPPROTO_IPV6, IPV6_2292PKTOPTIONS,(char *)incmsg, &cnt)
to get receiving interface index while no message is received,

What interface index should be returned?

[PATCH] IPv6:fix the return interface index when get it while no message is received

[Yang Hongyang]

Yang Hongyang wrote:
> when I use getsockopt(sk, IPPROTO_IPV6, IPV6_2292PKTOPTIONS,(char *)incmsg, &cnt)
> to get receiving interface index while no message is received,
> 
> What interface index should be returned?

When get receiving interface index while no message is received,
the bounded device's index of the socket should be returned?

Signed-off-by: Yang Hongyang<yanghy@cn.fujitsu.com>

--- a/ipv6_sockglue.c
+++ b/ipv6_sockglue.c
@@ -911,7 +911,7 @@ static int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
 		} else {
 			if (np->rxopt.bits.rxinfo) {
 				struct in6_pktinfo src_info;
-				src_info.ipi6_ifindex = np->mcast_oif;
+				src_info.ipi6_ifindex = np->mcast_oif ? np->mcast_oif : sk->sk_bound_dev_if;
 				ipv6_addr_copy(&src_info.ipi6_addr, &np->daddr);
 				put_cmsg(&msg, SOL_IPV6, IPV6_PKTINFO, sizeof(src_info), &src_info);
 			}
@@ -921,7 +921,7 @@ static int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
 			}
 			if (np->rxopt.bits.rxoinfo) {
 				struct in6_pktinfo src_info;
-				src_info.ipi6_ifindex = np->mcast_oif;
+				src_info.ipi6_ifindex = np->mcast_oif ? np->mcast_oif : sk->sk_bound_dev_if;
 				ipv6_addr_copy(&src_info.ipi6_addr, &np->daddr);
 				put_cmsg(&msg, SOL_IPV6, IPV6_2292PKTINFO, sizeof(src_info), &src_info);
 			}
-- 
1.5.3.8

Re: [PATCH] IPv6:fix the return interface index when get it while no message is received

[David Miller]

From: Yang Hongyang <yanghy@cn.fujitsu.com>
Date: Fri, 08 Aug 2008 13:59:54 +0800

> Yang Hongyang wrote:
> > when I use getsockopt(sk, IPPROTO_IPV6, IPV6_2292PKTOPTIONS,(char *)incmsg, &cnt)
> > to get receiving interface index while no message is received,
> > 
> > What interface index should be returned?
> 
> When get receiving interface index while no message is received,
> the bounded device's index of the socket should be returned?
> 
> Signed-off-by: Yang Hongyang<yanghy@cn.fujitsu.com>

To me it seems to be an undefined operation.

One cannot expect something valid from this socket option until a
packet really is received on the socket.

Re: [PATCH] IPv6:fix the return interface index when get it while no message is received

[Vlad Yasevich]

David Miller wrote:
> From: Yang Hongyang <yanghy@cn.fujitsu.com>
> Date: Fri, 08 Aug 2008 13:59:54 +0800
> 
>> Yang Hongyang wrote:
>>> when I use getsockopt(sk, IPPROTO_IPV6, IPV6_2292PKTOPTIONS,(char *)incmsg, &cnt)
>>> to get receiving interface index while no message is received,
>>>
>>> What interface index should be returned?
>> When get receiving interface index while no message is received,
>> the bounded device's index of the socket should be returned?
>>
>> Signed-off-by: Yang Hongyang<yanghy@cn.fujitsu.com>
> 
> To me it seems to be an undefined operation.
> 
> One cannot expect something valid from this socket option until a
> packet really is received on the socket.

Here is what 2292 has to say:

>   The corresponding receive option
> 
>        getsockopt(fd, IPPROTO_IPV6, IPV6_PKTOPTIONS, &buf, &len);
> 
>    returns a buffer with one or more ancillary data objects for all the
>    optional receive information that the application has previously
>    specified that it wants to receive.  The fourth argument points to
>    the buffer that is filled in by the call.  The fifth argument is a
>    pointer to a value-result integer: when the function is called the
>    integer specifies the size of the buffer pointed to by the fourth
>    argument, and on return this integer contains the actual number of
>    bytes that were returned.  The application processes this buffer
>    exactly as if the buffer were returned by recvmsg() as control
>    information.


So, there return value should be any sticky options set.

In the case of PKTINFO, the interface should be either 0 or the one
set by the sticky option.

The address should be should be from the options or IN6ADDR_ANY.

This is all for RFC2292 style.

For RFC 3542 style,  there what should be done:

   Issuing getsockopt() for the above options will return the sticky
   option value i.e., the value set with setsockopt().  If no sticky
   option value has been set getsockopt() will return the following
   values:

   -  For the IPV6_PKTINFO option, it will return an in6_pktinfo
      structure with ipi6_addr being in6addr_any and ipi6_ifindex being
      zero.

-vlad

Re: [PATCH] IPv6:fix the return interface index when get it while no message is received

[David Miller]

From: Vlad Yasevich <vladislav.yasevich@hp.com>
Date: Fri, 08 Aug 2008 10:58:04 -0400

> For RFC 3542 style,  there what should be done:
> 
>    Issuing getsockopt() for the above options will return the sticky
>    option value i.e., the value set with setsockopt().  If no sticky
>    option value has been set getsockopt() will return the following
>    values:
> 
>    -  For the IPV6_PKTINFO option, it will return an in6_pktinfo
>       structure with ipi6_addr being in6addr_any and ipi6_ifindex being
>       zero.

Ok so it looks like Yang's patch is ok.

Yank please resubmit your patch, thank you.

[PATCH] IPv6:fix the return interface index when get it while no message is received

[Yang Hongyang]

When get receiving interface index while no message is received,
the bounded device's index of the socket should be returned.

RFC 3542:
   Issuing getsockopt() for the above options will return the sticky
   option value i.e., the value set with setsockopt().  If no sticky
   option value has been set getsockopt() will return the following
   values:

   -  For the IPV6_PKTINFO option, it will return an in6_pktinfo
      structure with ipi6_addr being in6addr_any and ipi6_ifindex being
      zero.


Signed-off-by: Yang Hongyang<yanghy@cn.fujitsu.com>

---
 ipv6_sockglue.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/ipv6/ipv6_sockglue.c b/net/ipv6/ipv6_sockglue.c
index 2cabac9..e4f8d29 100644
--- a/net/ipv6/ipv6_sockglue.c
+++ b/net/ipv6/ipv6_sockglue.c
@@ -911,7 +911,7 @@ static int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
 		} else {
 			if (np->rxopt.bits.rxinfo) {
 				struct in6_pktinfo src_info;
-				src_info.ipi6_ifindex = np->mcast_oif;
+				src_info.ipi6_ifindex = np->mcast_oif ? np->mcast_oif : sk->sk_bound_dev_if;
 				ipv6_addr_copy(&src_info.ipi6_addr, &np->daddr);
 				put_cmsg(&msg, SOL_IPV6, IPV6_PKTINFO, sizeof(src_info), &src_info);
 			}
@@ -921,7 +921,7 @@ static int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
 			}
 			if (np->rxopt.bits.rxoinfo) {
 				struct in6_pktinfo src_info;
-				src_info.ipi6_ifindex = np->mcast_oif;
+				src_info.ipi6_ifindex = np->mcast_oif ? np->mcast_oif : sk->sk_bound_dev_if;
 				ipv6_addr_copy(&src_info.ipi6_addr, &np->daddr);
 				put_cmsg(&msg, SOL_IPV6, IPV6_2292PKTINFO, sizeof(src_info), &src_info);
 			}
-- 
1.5.3.8

Re: [PATCH] IPv6:fix the return interface index when get it while no message is received

[David Miller]

From: Yang Hongyang <yanghy@cn.fujitsu.com>
Date: Mon, 18 Aug 2008 14:07:41 +0800

> When get receiving interface index while no message is received,
> the bounded device's index of the socket should be returned.
> 
> RFC 3542:
>    Issuing getsockopt() for the above options will return the sticky
>    option value i.e., the value set with setsockopt().  If no sticky
>    option value has been set getsockopt() will return the following
>    values:
> 
>    -  For the IPV6_PKTINFO option, it will return an in6_pktinfo
>       structure with ipi6_addr being in6addr_any and ipi6_ifindex being
>       zero.
> 
> 
> Signed-off-by: Yang Hongyang<yanghy@cn.fujitsu.com>

Applied, thanks Yang.

Re: [PATCH] IPv6:fix the return interface index when get it while no message is received

[Vlad Yasevich]

Yang Hongyang wrote:
> When get receiving interface index while no message is received,
> the bounded device's index of the socket should be returned.
> 
> RFC 3542:
>    Issuing getsockopt() for the above options will return the sticky
>    option value i.e., the value set with setsockopt().  If no sticky
>    option value has been set getsockopt() will return the following
>    values:
> 
>    -  For the IPV6_PKTINFO option, it will return an in6_pktinfo
>       structure with ipi6_addr being in6addr_any and ipi6_ifindex being
>       zero.
> 
> 
> Signed-off-by: Yang Hongyang<yanghy@cn.fujitsu.com>
> 
> ---
>  ipv6_sockglue.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/net/ipv6/ipv6_sockglue.c b/net/ipv6/ipv6_sockglue.c
> index 2cabac9..e4f8d29 100644
> --- a/net/ipv6/ipv6_sockglue.c
> +++ b/net/ipv6/ipv6_sockglue.c
> @@ -911,7 +911,7 @@ static int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
>  		} else {
>  			if (np->rxopt.bits.rxinfo) {
>  				struct in6_pktinfo src_info;
> -				src_info.ipi6_ifindex = np->mcast_oif;
> +				src_info.ipi6_ifindex = np->mcast_oif ? np->mcast_oif : sk->sk_bound_dev_if;
>  				ipv6_addr_copy(&src_info.ipi6_addr, &np->daddr);
>  				put_cmsg(&msg, SOL_IPV6, IPV6_PKTINFO, sizeof(src_info), &src_info);
>  			}
> @@ -921,7 +921,7 @@ static int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
>  			}
>  			if (np->rxopt.bits.rxoinfo) {
>  				struct in6_pktinfo src_info;
> -				src_info.ipi6_ifindex = np->mcast_oif;
> +				src_info.ipi6_ifindex = np->mcast_oif ? np->mcast_oif : sk->sk_bound_dev_if;
>  				ipv6_addr_copy(&src_info.ipi6_addr, &np->daddr);
>  				put_cmsg(&msg, SOL_IPV6, IPV6_2292PKTINFO, sizeof(src_info), &src_info);
>  			}


I don't think that's correct at all.  The code path shows here is when there are no received
options and no sticky options set.  In such case, we shouldn't be returning multicast or
bound interfaces.  We should be returning 0.

Additionally the address returned is completely bogus as well.  We are returning the
address our peer instead of the one of our own addresses.

-vlad

Re: [PATCH] IPv6:fix the return interface index when get it while no message is received

[David Miller]

From: Vlad Yasevich <vladislav.yasevich@hp.com>
Date: Mon, 18 Aug 2008 09:31:07 -0400

> I don't think that's correct at all.  The code path shows here is
> when there are no received options and no sticky options set.  In
> such case, we shouldn't be returning multicast or bound interfaces.
> We should be returning 0.
>
> Additionally the address returned is completely bogus as well.  We
> are returning the address our peer instead of the one of our own
> addresses.

Yang, please fix this up, thank you.

Re: [PATCH] IPv6:fix the return interface index when get it while no message is received

[Yang Hongyang]

David Miller wrote:
> From: Vlad Yasevich <vladislav.yasevich@hp.com>
> Date: Mon, 18 Aug 2008 09:31:07 -0400
> 
>> I don't think that's correct at all.  The code path shows here is
>> when there are no received options and no sticky options set.  In
>> such case, we shouldn't be returning multicast or bound interfaces.
>> We should be returning 0.

I use setsockopt() to set the bounded interface of the socket, and 
then I get receiving interface index while no message is received through
the above socket,shoudn't the bounded interface be returned?

>>
>> Additionally the address returned is completely bogus as well. We
>> are returning the address our peer instead of the one of our own
>> addresses.

If message is received,the address returned is what the received
message refered to, otherwise the address returned is what I used setsockopt()
to set before.

> 
> Yang, please fix this up, thank you.
> 
> 

Seems that I misunderstood the RFC?

Re: [PATCH] IPv6:fix the return interface index when get it while no message is received

[Vlad Yasevich]

Yang Hongyang wrote:
> David Miller wrote:
>> From: Vlad Yasevich <vladislav.yasevich@hp.com>
>> Date: Mon, 18 Aug 2008 09:31:07 -0400
>>
>>> I don't think that's correct at all.  The code path shows here is
>>> when there are no received options and no sticky options set.  In
>>> such case, we shouldn't be returning multicast or bound interfaces.
>>> We should be returning 0.
> 
> I use setsockopt() to set the bounded interface of the socket, and 
> then I get receiving interface index while no message is received through
> the above socket,shoudn't the bounded interface be returned?

One of the problems is that we don't seem to support the sticky IPV6_PKTINFO
option as specified in rfc3542.

The setting of SO_BINDTODEVICE should not impact the return of the sticky option
values when they are not set.  A return of 0 in the sticky options means that
that value has not been specified.  If you return anything else, you give
the illusion that a sticky option was set.

> 
>>> Additionally the address returned is completely bogus as well. We
>>> are returning the address our peer instead of the one of our own
>>> addresses.
> 
> If message is received,the address returned is what the received
> message refered to, otherwise the address returned is what I used setsockopt()
> to set before.

We are returning np->daddr, which is the peer address.

The expected value is "destination of the received packet".  In the worst case,
it should be the locally bound address.  In the correct case, it should be the
address set in the sticky options if such is set.  It really should be 0::0, if there
are no sticky options and no packets have been received.

-vlad

> 
>> Yang, please fix this up, thank you.
>>
>>
> 
> Seems that I misunderstood the RFC?
> 

Re: [PATCH] IPv6:fix the return interface index when get it while no message is received

[Yang Hongyang]

Vlad Yasevich wrote:
> Yang Hongyang wrote:
>> David Miller wrote:
>>> From: Vlad Yasevich <vladislav.yasevich@hp.com>
>>> Date: Mon, 18 Aug 2008 09:31:07 -0400
>>>
>>>> I don't think that's correct at all.  The code path shows here is
>>>> when there are no received options and no sticky options set.  In
>>>> such case, we shouldn't be returning multicast or bound interfaces.
>>>> We should be returning 0.
>> I use setsockopt() to set the bounded interface of the socket, and 
>> then I get receiving interface index while no message is received through
>> the above socket,shoudn't the bounded interface be returned?
> 
> One of the problems is that we don't seem to support the sticky IPV6_PKTINFO
> option as specified in rfc3542.
> 
> The setting of SO_BINDTODEVICE should not impact the return of the sticky option
> values when they are not set.  A return of 0 in the sticky options means that
> that value has not been specified.  If you return anything else, you give
> the illusion that a sticky option was set.
> 
>>>> Additionally the address returned is completely bogus as well. We
>>>> are returning the address our peer instead of the one of our own
>>>> addresses.
>> If message is received,the address returned is what the received
>> message refered to, otherwise the address returned is what I used setsockopt()
>> to set before.
> 
> We are returning np->daddr, which is the peer address.
> 
> The expected value is "destination of the received packet".  In the worst case,
> it should be the locally bound address.  In the correct case, it should be the
> address set in the sticky options if such is set.  It really should be 0::0, if there
> are no sticky options and no packets have been received.

Since here we do not recive any packet,we should return 
src_info.ipi6_ifindex=0 and src_info.ipi6_addr=0::0,right?

> 
> -vlad
> 
>>> Yang, please fix this up, thank you.
>>>
>>>
>> Seems that I misunderstood the RFC?

Re: [PATCH] IPv6:fix the return interface index when get it while no message is received

[Vlad Yasevich]

Yang Hongyang wrote:
> Vlad Yasevich wrote:
>> Yang Hongyang wrote:
>>> David Miller wrote:
>>>> From: Vlad Yasevich <vladislav.yasevich@hp.com>
>>>> Date: Mon, 18 Aug 2008 09:31:07 -0400
>>>>
>>>>> I don't think that's correct at all.  The code path shows here is
>>>>> when there are no received options and no sticky options set.  In
>>>>> such case, we shouldn't be returning multicast or bound interfaces.
>>>>> We should be returning 0.
>>> I use setsockopt() to set the bounded interface of the socket, and 
>>> then I get receiving interface index while no message is received through
>>> the above socket,shoudn't the bounded interface be returned?
>> One of the problems is that we don't seem to support the sticky IPV6_PKTINFO
>> option as specified in rfc3542.
>>
>> The setting of SO_BINDTODEVICE should not impact the return of the sticky option
>> values when they are not set.  A return of 0 in the sticky options means that
>> that value has not been specified.  If you return anything else, you give
>> the illusion that a sticky option was set.
>>
>>>>> Additionally the address returned is completely bogus as well. We
>>>>> are returning the address our peer instead of the one of our own
>>>>> addresses.
>>> If message is received,the address returned is what the received
>>> message refered to, otherwise the address returned is what I used setsockopt()
>>> to set before.
>> We are returning np->daddr, which is the peer address.
>>
>> The expected value is "destination of the received packet".  In the worst case,
>> it should be the locally bound address.  In the correct case, it should be the
>> address set in the sticky options if such is set.  It really should be 0::0, if there
>> are no sticky options and no packets have been received.
> 
> Since here we do not recive any packet,we should return 
> src_info.ipi6_ifindex=0 and src_info.ipi6_addr=0::0,right?

That's what I've done in other implementations.

-vlad

> 
>> -vlad
>>
>>>> Yang, please fix this up, thank you.
>>>>
>>>>
>>> Seems that I misunderstood the RFC?
> 
> 
> 

Re: [PATCH] IPv6:fix the return interface index when get it while no message is received

[David Miller]

From: Vlad Yasevich <vladislav.yasevich@hp.com>
Date: Wed, 20 Aug 2008 09:10:34 -0400

> Yang Hongyang wrote:
> > Vlad Yasevich wrote:
> >> We are returning np->daddr, which is the peer address.
> >>
> >> The expected value is "destination of the received packet".  In the worst case,
> >> it should be the locally bound address.  In the correct case, it should be the
> >> address set in the sticky options if such is set.  It really should be 0::0, if there
> >> are no sticky options and no packets have been received.
> > 
> > Since here we do not recive any packet,we should return 
> > src_info.ipi6_ifindex=0 and src_info.ipi6_addr=0::0,right?
> 
> That's what I've done in other implementations.

I'm patiently waiting for a patch to fix this.

Thanks :)