On 10/18/2010 09:17 AM, Alexander Stein wrote:
> Hello,
> 
> On Saturday 16 October 2010, 11:50:27 Oliver Hartkopp wrote:
>>> On 10/14/2010 12:07 PM, Alexander Stein wrote:
>>>> Hello,
>>>>
>>>> On Thursday 14 October 2010, 12:00:06 Wolfgang Grandegger wrote:
>>>>>>> Yes, it's explained in section 3.2, 4.1.3 and 4.1.4 of:
>>>>>>>
>>>>>>> http://lxr.linux.no/#linux+v2.6.35.7/Documentation/networking/can.txt
>>>>>>
>>>>>> I read those sections several times, but I doubt I get it all
>>>>>> correctly. To my understanding CAN_RAW_LOOPBACK is the option to
>>>>>> enable reception of sent frames on _different_ sockets of the same
>>>>>> device. And CAN_RAW_RECV_OWN_MSGS is the option to enable reception
>>>>>> of sent frames on the _same_ socket of the same device.
>>>>>
>>>>> Sounds good.
>>>>>
>>>>>> If I'm right until here, why do I get my own sent messages on one
>>>>>> socket, if CAN_RAW_LOOPBACK = 1 and CAN_RAW_RECV_OWN_MSGS = 0 is set?
>>>>>
>>>>> You mean you get the message on the *same* socket you use for sending?
>>>>
>>>> Yes, I write a frame on this socket and after that I read which gets my
>>>> previously sent frame. I do bind this socket to a specific device, but I
>>>> doubt this does matter in this case.
>>
>> Looks like a deja-vu :-)
>>
>> Can you send your exact Kernel version?
> 
> I'm currently running v2.6.35-26-gfee7f2c (that's some custom patches).
> 
>> The correct behaviour hase been killed here
>>
>> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h
>> =fc6055a5ba31e2c14e36e8939f9bf2b6d586a7f5
>>
>> and fixed here
>>
>> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdi
>> ff;h=cff0d6e6edac7672b3f915bb4fb59f279243b7f9
>>
>> The discussion about this was here:
>>
>> can: expected receive behavior broken
>>
>> http://marc.info/?t=128030411900003&r=1&w=2
>>
>> Probably your kernel version is affected.
> 
> Yes, it seems my version is affected. I merged v2.6.35.7 into my tree and now 
> it works as expected on at91_can. Thanks a lot.
> 
> Wolfgang, I guess my can omit my test program now.

Of course. I obviously missed that issue/fix. Anyway, there is also a
little test program in the SVN repo.

Wolfgang.


_______________________________________________
Socketcan-users mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/socketcan-users

Reply via email to