[RFC PATCH] can: dev: update rx state in the can_get_echo_skb()

[RFC PATCH] can: dev: update rx state in the can_get_echo_skb()

[Hui Wang]

After a loopback packet is routed to rx interface, we forget to update
rx state both in the can_get_echo_skb() and in the controller specific
drivers, as a result, after we sent a loopback packet and executed
ifconfig, following state is outputed:
can0      Link encap:UNSPEC  HWaddr 00-00-00-00-<snip>
          UP RUNNING NOARP  MTU:16  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:1 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:10
          RX bytes:0 (0.0 B)  TX bytes:4 (4.0 B)

We add rx state update in the can_get_echo_skb() to solve this
problem.

Cc: linux-can@vger.kernel.org
Cc: Marc Kleine-Budde <mkl@pengutronix.de>
Cc: Wolfgang Grandegger <wg@grandegger.com>
Cc: Shawn Guo <shawn.guo@linaro.org>
Signed-off-by: Hui Wang <jason77.wang@gmail.com>
---
I don't know if it is designed like that on purpose, please instruct
me if i am wrong.

I use following commands to let flexcan to work in the loopback mode:
%>ip link set can0 type can bitrate 125000
%>ip link set can0 type can loopback on
%>ip link set can0 up
And send a packet:
%>cantest can0 123#AABBCCDD
When i check net device state, i found tx doesn't equal rx:
%>ifconfig
can0      Link encap:UNSPEC  HWaddr 00-00-00-00-<snip>
          UP RUNNING NOARP  MTU:16  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:1 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:10
          RX bytes:0 (0.0 B)  TX bytes:4 (4.0 B)

From my understanding, i suppose it should be:
can0      Link encap:UNSPEC  HWaddr 00-00-00-00-<snip>
          UP RUNNING NOARP  MTU:16  Metric:1
          RX packets:1 errors:0 dropped:0 overruns:0 frame:0
          TX packets:1 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:10 
          RX bytes:4 (4.0 B)  TX bytes:4 (4.0 B)

 drivers/net/can/dev.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)

diff --git a/drivers/net/can/dev.c b/drivers/net/can/dev.c
index f03d7a4..6b62f6a 100644
--- a/drivers/net/can/dev.c
+++ b/drivers/net/can/dev.c
@@ -340,6 +340,9 @@ unsigned int can_get_echo_skb(struct net_device *dev, unsigned int idx)
 		netif_rx(priv->echo_skb[idx]);
 		priv->echo_skb[idx] = NULL;
 
+		dev->stats.rx_bytes += dlc;
+		dev->stats.rx_packets++;
+
 		return dlc;
 	}
 
-- 
1.7.6

Re: [RFC PATCH] can: dev: update rx state in the can_get_echo_skb()

[Wolfgang Grandegger]

On 06/28/2012 11:42 AM, Hui Wang wrote:
> After a loopback packet is routed to rx interface, we forget to update
> rx state both in the can_get_echo_skb() and in the controller specific
> drivers, as a result, after we sent a loopback packet and executed
> ifconfig, following state is outputed:
> can0      Link encap:UNSPEC  HWaddr 00-00-00-00-<snip>
>           UP RUNNING NOARP  MTU:16  Metric:1
>           RX packets:0 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:1 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:10
>           RX bytes:0 (0.0 B)  TX bytes:4 (4.0 B)
> 
> We add rx state update in the can_get_echo_skb() to solve this
> problem.

Well...

> Cc: linux-can@vger.kernel.org
> Cc: Marc Kleine-Budde <mkl@pengutronix.de>
> Cc: Wolfgang Grandegger <wg@grandegger.com>
> Cc: Shawn Guo <shawn.guo@linaro.org>
> Signed-off-by: Hui Wang <jason77.wang@gmail.com>
> ---
> I don't know if it is designed like that on purpose, please instruct
> me if i am wrong.
> 
> I use following commands to let flexcan to work in the loopback mode:
> %>ip link set can0 type can bitrate 125000
> %>ip link set can0 type can loopback on
> %>ip link set can0 up
> And send a packet:
> %>cantest can0 123#AABBCCDD
> When i check net device state, i found tx doesn't equal rx:
> %>ifconfig
> can0      Link encap:UNSPEC  HWaddr 00-00-00-00-<snip>
>           UP RUNNING NOARP  MTU:16  Metric:1
>           RX packets:0 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:1 errors:0 dropped:0 overruns:0 carrier:0

This means that one message has successfully been sent to the *real* CAN
bus.

>           collisions:0 txqueuelen:10
>           RX bytes:0 (0.0 B)  TX bytes:4 (4.0 B)
> 
>>From my understanding, i suppose it should be:
> can0      Link encap:UNSPEC  HWaddr 00-00-00-00-<snip>
>           UP RUNNING NOARP  MTU:16  Metric:1
>           RX packets:1 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:1 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:10 
>           RX bytes:4 (4.0 B)  TX bytes:4 (4.0 B)

Why, you have not received any message from the CAN bus. If you want to
count looped back messages, it should be:

          RX packets:1 errors:0 dropped:0 overruns:0 frame:0
          TX packets:2 errors:0 dropped:0 overruns:0 carrier:0

Wolfgang.

Re: [RFC PATCH] can: dev: update rx state in the can_get_echo_skb()

[Hui Wang]

Wolfgang Grandegger wrote:
> On 06/28/2012 11:42 AM, Hui Wang wrote:
>   
>> After a loopback packet is routed to rx interface, we forget to update
>> rx state both in the can_get_echo_skb() and in the controller specific
>> drivers, as a result, after we sent a loopback packet and executed
>> ifconfig, following state is outputed:
>> can0      Link encap:UNSPEC  HWaddr 00-00-00-00-<snip>
>>           UP RUNNING NOARP  MTU:16  Metric:1
>>           RX packets:0 errors:0 dropped:0 overruns:0 frame:0
>>           TX packets:1 errors:0 dropped:0 overruns:0 carrier:0
>>           collisions:0 txqueuelen:10
>>           RX bytes:0 (0.0 B)  TX bytes:4 (4.0 B)
>>
>> We add rx state update in the can_get_echo_skb() to solve this
>> problem.
>>     
>
> Well...
>
>   
>> Cc: linux-can@vger.kernel.org
>> Cc: Marc Kleine-Budde <mkl@pengutronix.de>
>> Cc: Wolfgang Grandegger <wg@grandegger.com>
>> Cc: Shawn Guo <shawn.guo@linaro.org>
>> Signed-off-by: Hui Wang <jason77.wang@gmail.com>
>> ---
>> I don't know if it is designed like that on purpose, please instruct
>> me if i am wrong.
>>
>> I use following commands to let flexcan to work in the loopback mode:
>> %>ip link set can0 type can bitrate 125000
>> %>ip link set can0 type can loopback on
>> %>ip link set can0 up
>> And send a packet:
>> %>cantest can0 123#AABBCCDD
>> When i check net device state, i found tx doesn't equal rx:
>> %>ifconfig
>> can0      Link encap:UNSPEC  HWaddr 00-00-00-00-<snip>
>>           UP RUNNING NOARP  MTU:16  Metric:1
>>           RX packets:0 errors:0 dropped:0 overruns:0 frame:0
>>           TX packets:1 errors:0 dropped:0 overruns:0 carrier:0
>>     
>
> This means that one message has successfully been sent to the *real* CAN
> bus.
>
>   
>>           collisions:0 txqueuelen:10
>>           RX bytes:0 (0.0 B)  TX bytes:4 (4.0 B)
>>
>> >From my understanding, i suppose it should be:
>> can0      Link encap:UNSPEC  HWaddr 00-00-00-00-<snip>
>>           UP RUNNING NOARP  MTU:16  Metric:1
>>           RX packets:1 errors:0 dropped:0 overruns:0 frame:0
>>           TX packets:1 errors:0 dropped:0 overruns:0 carrier:0
>>           collisions:0 txqueuelen:10 
>>           RX bytes:4 (4.0 B)  TX bytes:4 (4.0 B)
>>     
>
> Why, you have not received any message from the CAN bus. If you want to
> count looped back messages, it should be:
>
>   
You mean CAN bus didn't receive a packet,  the RX state of CAN0 is 0. It 
is a reasonable explanation.

Thanks,
Hui.
>           RX packets:1 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:2 errors:0 dropped:0 overruns:0 carrier:0
>
> Wolfgang.
>
>   

Re: [RFC PATCH] can: dev: update rx state in the can_get_echo_skb()

[Kurt Van Dijck]

Hi,

A loopback packet enters the rx interface of the CAN subsystem,
but it had never entered the chip's receive path.
IMHO it's bad to alter the chip's statistics the way you propose.

Are you interested in the number of packets/bytes
that you were able to receive on your socket?
How about adding echo statistics in the CAN device stats,
as shown in
$ ip -d -s link show can0

Kind regards,
Kurt

On Thu, Jun 28, 2012 at 05:42:20PM +0800, Hui Wang wrote:
> After a loopback packet is routed to rx interface, we forget to update
> rx state both in the can_get_echo_skb() and in the controller specific
> drivers, as a result, after we sent a loopback packet and executed
> ifconfig, following state is outputed:
> can0      Link encap:UNSPEC  HWaddr 00-00-00-00-<snip>
>           UP RUNNING NOARP  MTU:16  Metric:1
>           RX packets:0 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:1 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:10
>           RX bytes:0 (0.0 B)  TX bytes:4 (4.0 B)
> 
> We add rx state update in the can_get_echo_skb() to solve this
> problem.
> 
> Cc: linux-can@vger.kernel.org
> Cc: Marc Kleine-Budde <mkl@pengutronix.de>
> Cc: Wolfgang Grandegger <wg@grandegger.com>
> Cc: Shawn Guo <shawn.guo@linaro.org>
> Signed-off-by: Hui Wang <jason77.wang@gmail.com>
> ---
> I don't know if it is designed like that on purpose, please instruct
> me if i am wrong.
> 
> I use following commands to let flexcan to work in the loopback mode:
> %>ip link set can0 type can bitrate 125000
> %>ip link set can0 type can loopback on
> %>ip link set can0 up
> And send a packet:
> %>cantest can0 123#AABBCCDD
> When i check net device state, i found tx doesn't equal rx:
> %>ifconfig
> can0      Link encap:UNSPEC  HWaddr 00-00-00-00-<snip>
>           UP RUNNING NOARP  MTU:16  Metric:1
>           RX packets:0 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:1 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:10
>           RX bytes:0 (0.0 B)  TX bytes:4 (4.0 B)
> 
> From my understanding, i suppose it should be:
> can0      Link encap:UNSPEC  HWaddr 00-00-00-00-<snip>
>           UP RUNNING NOARP  MTU:16  Metric:1
>           RX packets:1 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:1 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:10 
>           RX bytes:4 (4.0 B)  TX bytes:4 (4.0 B)
> 
>  drivers/net/can/dev.c |    3 +++
>  1 files changed, 3 insertions(+), 0 deletions(-)
> 
> diff --git a/drivers/net/can/dev.c b/drivers/net/can/dev.c
> index f03d7a4..6b62f6a 100644
> --- a/drivers/net/can/dev.c
> +++ b/drivers/net/can/dev.c
> @@ -340,6 +340,9 @@ unsigned int can_get_echo_skb(struct net_device *dev, unsigned int idx)
>  		netif_rx(priv->echo_skb[idx]);
>  		priv->echo_skb[idx] = NULL;
>  
> +		dev->stats.rx_bytes += dlc;
> +		dev->stats.rx_packets++;
> +
>  		return dlc;
>  	}
>  
> -- 
> 1.7.6
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-can" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

-- 
Kurt Van Dijck
GRAMMER EiA ELECTRONICS
http://www.eia.be
kurt.van.dijck@eia.be
+32-38708534

Re: [RFC PATCH] can: dev: update rx state in the can_get_echo_skb()

[Hui Wang]

Kurt Van Dijck wrote:
> Hi,
>
> A loopback packet enters the rx interface of the CAN subsystem,
> but it had never entered the chip's receive path.
> IMHO it's bad to alter the chip's statistics the way you propose.
>   
Your explanation makes sense to me.
> Are you interested in the number of packets/bytes
> that you were able to receive on your socket?
> How about adding echo statistics in the CAN device stats,
> as shown in
> $ ip -d -s link show can0
>   
It can work.

Thanks,
Hui.
> Kind regards,
> Kurt
>
> On Thu, Jun 28, 2012 at 05:42:20PM +0800, Hui Wang wrote:
>