On 12.10.2010 08:51, Kurt Van Dijck wrote:
> Hi,
>
> On Mon, Oct 11, 2010 at 05:14:20PM +0200, [email protected] wrote:
>> Hello Kurt,
>>
>>> IMO, a CAN frame from another socket should be treated as coming from
>>> another bus node.
>>> I know (J1939 is still in the queue) that some protocols may benefit from
>>> several sockets cooperating, but should that be exported to userspace?
>>
>>
>> I fully agree on that. Two different sockets shall behave as two
>> different CAN nodes on the same CAN bus even if they are opened
>> on the same hardware. Taking that in to account vcan shall be actually
>> called "virtual CAN bus" and not "virtual CAN interface"...
>>
>> My understanding of "self-reception" per-socket option is that the
>> messages send over the specific socket are not only received by
>> other sockets on the same HW, but also by the original socket (so
>> "echo" may be even better name for that). And the reason why I
>> want to receive (some of) my own Tx messages is because I want
>> to know *exactly* when whose messages where physically send over
>> CAN bus. And that I am getting by reading the timestamp of my
>> self-received message. And the per-message self-receive flag helps
>> me to distinguish between Rx messages (coming from other CAN nodes
>> or other sockets on the same CAN node) and my own Tx messages.
>
> Thanks for the update. Seems like my understanding was right.
Ah, ok. So it's not to know, whether the CAN frame came from outside BUT to
know, when my originated CAN frame from this socket touched the bus.
>
>>
>> Regards,
>> Vladislav
>
> Oliver, I assembled a patch showing my idea of what need to be done.
> never tested, never even compiled, it just shows my thinking.
I'll be on a business trip until Thursday.
So far i did not have any luck to push the MSG_DONTROUTE bit into userspace.
Maybe you can try.
Regards,
Oliver
>
> ---
> net/can/raw.c | 9 ++++++---
> 1 file changed, 6 insertions(+), 3 deletions(-)
>
>
> Index: linux/net/can/raw.c
> ===================================================================
> --- linux.orig/net/can/raw.c 2010-10-12 08:46:33.711192001 +0200
> +++ linux/net/can/raw.c 2010-10-12 08:49:17.191192001 +0200
> @@ -94,20 +94,23 @@
> return (struct raw_sock *)sk;
> }
>
> -static void raw_rcv(struct sk_buff *skb, void *data)
> +static void raw_rcv(struct sk_buff *oskb, void *data)
> {
> struct sock *sk = (struct sock *)data;
> struct raw_sock *ro = raw_sk(sk);
> struct sockaddr_can *addr;
> + struct sk_buff *skb; /* new */
>
> /* check the received tx sock reference */
> - if (!ro->recv_own_msgs && skb->sk == sk)
> + if (!ro->recv_own_msgs && oskb->sk == sk)
> return;
>
> /* clone the given skb to be able to enqueue it into the rcv queue */
> - skb = skb_clone(skb, GFP_ATOMIC);
> + skb = skb_clone(oskb, GFP_ATOMIC);
> if (!skb)
> return;
> + if (oskb->sk == sk)
> + skb->msg_flags |= MSG_DONTROUTE;
>
> /*
> * Put the datagram to the queue so that raw_recvmsg() can
_______________________________________________
Socketcan-core mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/socketcan-core
